**Greenhouse Effect**

Andrew A. Lacis *NASA Goddard Institute for Space Studies New York, NY* 

#### **1. Introduction**

The greenhouse effect is the physical warming of the ground surface that arises in semitransparent planetary atmospheres that are heated by solar radiation. In order for the greenhouse effect to operate, the planet's atmosphere must be sufficiently transparent at visible wavelengths to allow significant amounts of solar radiation to be absorbed by the ground surface. The atmosphere must also be sufficiently opaque at thermal wavelengths to prevent thermal radiation emitted by the ground surface from escaping directly out to space. In such planetary atmospheres, the long-wave opacity for the absorption of radiation at thermal wavelengths is provided by the so-called greenhouse gases (e.g., water vapor, carbon dioxide, methane, nitrous oxide), although absorption by particulate matter such as that due to clouds and aerosols can also make significant contributions to the greenhouse effect. The net effect of the planetary greenhouse effect is to keep the ground surface of the planet at a warmer temperature than it would otherwise be if there were no greenhouse effect. The operation of a greenhouse effect also imparts a temperature gradient in the planet's atmosphere, with the highest temperatures being at the ground, decreasing with height in the atmosphere. This is primarily a radiative effect that is expected to occur in all planetary atmospheres where the ground surface is heated by solar radiation, and there is substantial atmospheric opacity at thermal wavelengths of the spectrum.

### **2. Planetary comparisons**

In our solar system, the greenhouse effect operates on those terrestrial planets that have substantial atmospheres (i.e., Earth, Mars, and Venus), including also on Saturn's moon Titan. Typically, carbon dioxide is the key greenhouse gas that keeps the ground surface temperatures of the terrestrial planets warmer than they would be without the greenhouse effect. The importance of carbon dioxide as a greenhouse gas stems from its very strong absorption bands that cover much of the thermal spectrum, and also because carbon dioxide does not condense and precipitate from the atmosphere at the prevailing temperatures on the terrestrial planets. Being so much farther from the sun, the atmosphere of Titan is much colder than the atmospheres of the terrestrial planets, so much so that carbon dioxide and water vapor cannot be maintained in gaseous form on Titan. Instead, the greenhouse effect on Titan is sustained primarily by pressure-induced

Greenhouse Effect 277

primarily of carbon dioxide. Interestingly, the column amount of carbon dioxide in the Martian atmosphere is actually about fifty times greater than in the atmosphere of Earth. Nevertheless, Mars can only muster a small amount of greenhouse warming (about 5°C, or 9°F). The reason for this is that the atmospheric pressure is a very important factor in determining the efficiency of greenhouse gas performance. Because of the small atmospheric pressure on Mars (less than one hundredth that on Earth), the spectral absorption lines of carbon dioxide on Mars are very narrow, and therefore act like a picket fence that lets most of the thermal radiation emitted by the Martian ground surface to escape directly out to space. This does not happen on Earth because of the atmospheric pressure that is exerted by the radiatively inactive nitrogen and oxygen, causing the spectral absorption lines of carbon dioxide and water vapor to be greatly broadened, making them more effective absorbers of thermal radiation. And, for comparison, it is the extremely high pressure on Venus that makes the carbon dioxide absorption of thermal radiation particularly efficient in the Venus

On Mercury, there is no tangible atmosphere, and hence there is no opportunity for the greenhouse effect to operate. Accordingly, the surface temperature on Mercury is determined solely as the result of local thermal equilibrium with the absorbed solar radiation, producing scorching hot temperatures on the sunlit side, and being abysmally

The strength of the planetary greenhouse effect is something that can, at least in principle, be determined by direct measurement. Perhaps the most basic measurement of a planet's energy balance is the reflected solar energy. Given the planet's distance from the Sun, it is therefore taken for granted that the incident solar radiation is a known quantity. Upon subtracting the reflected solar radiation from the incident solar energy flux, the energy input to the planet is thus defined. Other potential energy sources such as tidal, geothermal, nuclear, etc. are several orders of magnitude smaller than the absorbed solar radiation, and therefore need not be considered. A very natural and logical assumption that is commonly made is that the planet is in global energy balance, meaning that the thermal radiation emitted by the planet is going to be equal to the absorbed solar energy. Accordingly, this determines the outgoing long-wave energy that the planet radiates out to space. The longwave energy emitted to space then serves to define the effective radiating temperature of the planet by equating the emitted energy to its equivalent Planck, or blackbody radiation, as

where R2 is the projected geometrical area of the planet, So is the solar constant, A is the planet's bond albedo on the solar side of the equation. On the thermal side 4R2 is the total surface area of the planet, is the Stefan-Boltzmann constant (5.67 x 10–8 W/m2 K–4), and TE is the effective radiating temperature of the planet. The outgoing long-wave flux is also a subject for direct measurement, while the effective temperature is defined numerically from

FLW = TE4 (2)

R2 So (1 – A) = 4R2 TE4 (1)

greenhouse effect.

cold on the dark side.

summarized in equation form by

and

the long-wave flux through the Stefan-Boltzmann law.

methane, hydrogen, and nitrogen, to keep the surface temperature on Titan some 12°C (22°F) warmer than it would otherwise be if Titan were warmed only by its absorbed solar radiation (McKay et al., 1991). With Saturn being 9.5 times further from the Sun than the Earth, the solar radiation incident on Titan is about 90 times fainter than the solar radiation that is incident on Earth, endowing Titan with a frigid atmosphere and a surface temperature of 82 K (−191°C, −312°F).

Of the terrestrial planets, Venus has by far the strongest greenhouse effect. This is because the atmosphere of Venus is nearly a 100 times more massive and denser than that of Earth, and also because the atmosphere of Venus is composed almost entirely of the prime greenhouse gas, carbon dioxide. The large atmospheric pressure on Venus greatly broadens the carbon dioxide absorption lines to provide strong absorption across the entire thermal spectrum. Only about 1% of the incident solar radiation penetrates to the ground, but the thermal opacity in the atmosphere of Venus is so large as to produce a greenhouse effect that is about 15 times greater than that on Earth. Because of this, the surface temperature on Venus is nearly 460°C (860°F), which is hot enough to melt lead and vaporize mercury. Moreover, this is about 500°C (900°F) hotter than the surface of Venus would be (−40°C, −40°F) if Venus were simply in thermal equilibrium with the global mean solar energy that is absorbed by the Venus atmosphere, without benefit of the greenhouse effect. Interestingly, the non-greenhouse equilibrium temperature of Venus is actually some 12°C (22°F) colder than the corresponding equilibrium temperature of Earth, even though Venus is closer to the Sun and receives twice the terrestrial insolation. This situation arises because the cloud-shrouded atmosphere of Venus is far more reflecting than that of Earth, reflecting more than 70% of the incident sunlight, whereas Earth reflects only about 30%.

On Earth, the strength of the greenhouse warming effect is a more moderate 33°C (60°F). This amount of greenhouse warming makes the global mean surface temperature of the Earth a relatively comfortable 15°C (59°F) instead of an otherwise frigid −18°C (−1°F), which would be the equilibrium temperature for Earth with a planetary albedo of 30% and no greenhouse effect to trap the absorbed solar radiation. Carbon dioxide is also the key greenhouse gas of the terrestrial atmosphere, but with a very important distinction. In the terrestrial atmosphere, most of the greenhouse effect is actually provided by water vapor and by clouds, with only about 20% due to carbon dioxide. This is because Earth is the 'water' planet where water in all of its forms has a central role in defining the properties of the terrestrial climate system. The greenhouse warming provided by carbon dioxide and the other non-condensing greenhouse gases like methane, nitrous oxide, and ozone, allows water vapor to evaporate and remain in the atmosphere. This produces a strong 'feedback effect' which magnifies that part of the greenhouse warming that is uniquely attributable to the non-condensing greenhouse gases. Thus, water vapor with its strong absorption bands in the thermal part of the spectrum, accounts for approximately 50% of the terrestrial greenhouse effect. Long-wave absorption by cloud particles, which absorb radiation at all wavelengths of the thermal spectrum, accounts for about 25% of the total greenhouse effect. The highly interactive atmosphere of Earth serves to produce the uniquely variable climate that exists nowhere else in the solar system.

In contrast to Venus and Earth, the Martian atmosphere is very thin and tenuous, being scarcely 1% of the Earth's atmosphere. Like Venus, the Martian atmosphere is composed

methane, hydrogen, and nitrogen, to keep the surface temperature on Titan some 12°C (22°F) warmer than it would otherwise be if Titan were warmed only by its absorbed solar radiation (McKay et al., 1991). With Saturn being 9.5 times further from the Sun than the Earth, the solar radiation incident on Titan is about 90 times fainter than the solar radiation that is incident on Earth, endowing Titan with a frigid atmosphere and a surface

Of the terrestrial planets, Venus has by far the strongest greenhouse effect. This is because the atmosphere of Venus is nearly a 100 times more massive and denser than that of Earth, and also because the atmosphere of Venus is composed almost entirely of the prime greenhouse gas, carbon dioxide. The large atmospheric pressure on Venus greatly broadens the carbon dioxide absorption lines to provide strong absorption across the entire thermal spectrum. Only about 1% of the incident solar radiation penetrates to the ground, but the thermal opacity in the atmosphere of Venus is so large as to produce a greenhouse effect that is about 15 times greater than that on Earth. Because of this, the surface temperature on Venus is nearly 460°C (860°F), which is hot enough to melt lead and vaporize mercury. Moreover, this is about 500°C (900°F) hotter than the surface of Venus would be (−40°C, −40°F) if Venus were simply in thermal equilibrium with the global mean solar energy that is absorbed by the Venus atmosphere, without benefit of the greenhouse effect. Interestingly, the non-greenhouse equilibrium temperature of Venus is actually some 12°C (22°F) colder than the corresponding equilibrium temperature of Earth, even though Venus is closer to the Sun and receives twice the terrestrial insolation. This situation arises because the cloud-shrouded atmosphere of Venus is far more reflecting than that of Earth, reflecting more than 70% of the incident

On Earth, the strength of the greenhouse warming effect is a more moderate 33°C (60°F). This amount of greenhouse warming makes the global mean surface temperature of the Earth a relatively comfortable 15°C (59°F) instead of an otherwise frigid −18°C (−1°F), which would be the equilibrium temperature for Earth with a planetary albedo of 30% and no greenhouse effect to trap the absorbed solar radiation. Carbon dioxide is also the key greenhouse gas of the terrestrial atmosphere, but with a very important distinction. In the terrestrial atmosphere, most of the greenhouse effect is actually provided by water vapor and by clouds, with only about 20% due to carbon dioxide. This is because Earth is the 'water' planet where water in all of its forms has a central role in defining the properties of the terrestrial climate system. The greenhouse warming provided by carbon dioxide and the other non-condensing greenhouse gases like methane, nitrous oxide, and ozone, allows water vapor to evaporate and remain in the atmosphere. This produces a strong 'feedback effect' which magnifies that part of the greenhouse warming that is uniquely attributable to the non-condensing greenhouse gases. Thus, water vapor with its strong absorption bands in the thermal part of the spectrum, accounts for approximately 50% of the terrestrial greenhouse effect. Long-wave absorption by cloud particles, which absorb radiation at all wavelengths of the thermal spectrum, accounts for about 25% of the total greenhouse effect. The highly interactive atmosphere of Earth serves to produce the uniquely variable climate

In contrast to Venus and Earth, the Martian atmosphere is very thin and tenuous, being scarcely 1% of the Earth's atmosphere. Like Venus, the Martian atmosphere is composed

temperature of 82 K (−191°C, −312°F).

sunlight, whereas Earth reflects only about 30%.

that exists nowhere else in the solar system.

primarily of carbon dioxide. Interestingly, the column amount of carbon dioxide in the Martian atmosphere is actually about fifty times greater than in the atmosphere of Earth. Nevertheless, Mars can only muster a small amount of greenhouse warming (about 5°C, or 9°F). The reason for this is that the atmospheric pressure is a very important factor in determining the efficiency of greenhouse gas performance. Because of the small atmospheric pressure on Mars (less than one hundredth that on Earth), the spectral absorption lines of carbon dioxide on Mars are very narrow, and therefore act like a picket fence that lets most of the thermal radiation emitted by the Martian ground surface to escape directly out to space. This does not happen on Earth because of the atmospheric pressure that is exerted by the radiatively inactive nitrogen and oxygen, causing the spectral absorption lines of carbon dioxide and water vapor to be greatly broadened, making them more effective absorbers of thermal radiation. And, for comparison, it is the extremely high pressure on Venus that makes the carbon dioxide absorption of thermal radiation particularly efficient in the Venus greenhouse effect.

On Mercury, there is no tangible atmosphere, and hence there is no opportunity for the greenhouse effect to operate. Accordingly, the surface temperature on Mercury is determined solely as the result of local thermal equilibrium with the absorbed solar radiation, producing scorching hot temperatures on the sunlit side, and being abysmally cold on the dark side.

The strength of the planetary greenhouse effect is something that can, at least in principle, be determined by direct measurement. Perhaps the most basic measurement of a planet's energy balance is the reflected solar energy. Given the planet's distance from the Sun, it is therefore taken for granted that the incident solar radiation is a known quantity. Upon subtracting the reflected solar radiation from the incident solar energy flux, the energy input to the planet is thus defined. Other potential energy sources such as tidal, geothermal, nuclear, etc. are several orders of magnitude smaller than the absorbed solar radiation, and therefore need not be considered. A very natural and logical assumption that is commonly made is that the planet is in global energy balance, meaning that the thermal radiation emitted by the planet is going to be equal to the absorbed solar energy. Accordingly, this determines the outgoing long-wave energy that the planet radiates out to space. The longwave energy emitted to space then serves to define the effective radiating temperature of the planet by equating the emitted energy to its equivalent Planck, or blackbody radiation, as summarized in equation form by

$$
\pi \mathbf{R}^2 \,\mathrm{S}\_{\mathrm{o}} \,\mathrm{(1 - A)} = 4\pi \mathbf{R}^2 \,\mathrm{\sigma} \,\mathrm{T}\_{\mathrm{E}} \,\mathrm{4} \tag{1}
$$

where R2 is the projected geometrical area of the planet, So is the solar constant, A is the planet's bond albedo on the solar side of the equation. On the thermal side 4R2 is the total surface area of the planet, is the Stefan-Boltzmann constant (5.67 x 10–8 W/m2 K–4), and TE is the effective radiating temperature of the planet. The outgoing long-wave flux is also a subject for direct measurement, while the effective temperature is defined numerically from the long-wave flux through the Stefan-Boltzmann law.

$$\mathbf{F}\_{\rm LW} \equiv \sigma \mathbf{'} \mathbf{T}\_{\rm E}^{4} \tag{2}$$

and

Greenhouse Effect 279

observational data. GT expresses the strength of the greenhouse effect directly in terms of the temperature difference by which the global mean surface temperature would be increased in comparison to the temperature that would exist if there were no greenhouse effect. The numbers given in Table 1 represent annually-averaged global-mean values. Obviously, the strength of the greenhouse effect varies with geographic location and with season. While the well-mixed non-condensing greenhouse gas contribution to the greenhouse effect may tend to be globally uniform and steady in time, the water vapor and the cloud feedback contribution to the greenhouse effect is strongly variable,

It is interesting that of the numbers in Table 1, the outgoing long-wave fluxes of the three planets, as represented by the effective temperature, are so similar. More noteworthy are the factors of a 100 difference in the massiveness of the respective atmospheres, as represented by the surface pressure PS. In all three atmospheres, carbon dioxide is the key greenhouse gas. But the Earth is unique in that its greenhouse effect includes feedback amplification.

It is the combination of solar radiative heating and the strength of the greenhouse effect that determines the surface temperature of a planet. In the early days of the solar system some 4.5 billion years ago, the Sun had only about 75% of its present luminosity. But, as hydrogen in the interior of the Sun gets transformed into heavier elements, this causes the temperature and luminosity of the Sun increase, which in turn produces a slow and steady rise in the energy input to the terrestrial climate system. This is sometimes called the astrophysical climate forcing, and this type of solar radiative forcing will continue unabated. In several

While changes in solar insolation act to regulate the inflow of energy to the atmosphere and ground, it is the strength of the greenhouse effect that exerts its influence on the terrestrial climate system by regulating the energy outflow, and by generating additional heating of the ground surface by down-welling long-wave radiation. Here, carbon dioxide plays the leading role. In the early days of the Earth when the Sun was much less luminous, high concentrations of carbon dioxide where able to keep the temperature habitable for life forms

Over million-year time scales, volcanoes are the principal source of atmospheric carbon dioxide, and weathering of rocks is the principal sink (Berner, 2004), with the biosphere of the Earth also participating both as a source and as a sink. Because the principal sources operate independently of each other, the atmospheric level of carbon dioxide can fluctuate unchecked. There is geological evidence that shows glaciation was taking place at tropical latitudes 650 to 750 million years ago (Kirschvink, 1992), suggesting that snowball Earth conditions might have existed at that time. If, over millions of years, rock weathering were to exceed volcanic replenishment, atmospheric carbon dioxide could fall below its critical value, thus sending the Earth to snowball conditions. Eventually, with the ocean covered by ice, and thus prevented from taking up carbon dioxide, sufficient buildup of atmospheric carbon dioxide from volcanic eruptions would gradually bring the Earth out of its snowball

geographically, seasonally, and even inter-annually.

billion years time it will make the Earth too hot to be habitable.

**3. Geological perspective** 

to get established and to take root.

state climate.

$$\mathbf{T\_E = [S\_o \ (1 - A)/4\sigma]^{1/4}}\tag{3}$$

Also needed is the information about the planet's surface temperature and the up-welling thermal flux that is emitted by the ground surface, both of which are measurable quantities. Direct measurements have been made by space probes that landed on the surfaces of Mars and Venus. Space based measurements are also possible by using selected spectral windows in the long-wave spectrum where radiation from the ground surface can escape directly out to space. Such measurements are a topic in remote sensing. They rely on associating the intensity of the measured spectral radiance with the equivalent spectral brightness of Planck radiation. By such means, microwave measurements of Venus were the first indication that the surface temperature of Venus was exceedingly hot, before verification by direct space probe measurements. A summary of planetary greenhouse parameters is listed in Table 1.


Table 1. Planetary greenhouse parameters. (After Lacis et al., 2011).

The greenhouse parameters listed in Table 1 are all measurable quantities so that, at least in principle, the strength of the greenhouse effect of a planet is empirically determinable. As explained above, the effective temperature, TE(K), is actually a measure of either the emitted long-wave radiation at the top of the atmosphere, or the absorbed solar radiation by the planet, as given by Equation (3). Direct measurements of the globally averaged radiative fluxes at the top of the atmosphere and at the ground surface are actually quite difficult to obtain, particularly in view of the large variability of the solar and thermal radiative fluxes both geographically and with time, and also as a function of wavelength and viewing geometry. In practice, it is a combination of observational data analyses and theoretical modeling that is required to refine our knowledge of the planetary greenhouse parameters.

Basically, it is the long-wave radiative fluxes at the top of the atmosphere and at the ground surface that define the greenhouse effect. The strength of the planetary greenhouse effect is expressed as the difference between the up-welling long-wave flux emitted by the ground surface, and the outgoing long-wave flux at the top of the atmosphere. Thus,

$$\mathbf{C\_{F}} = \sigma \mathbf{T\_{S}} 4 - \sigma \mathbf{T\_{E}} 4 \tag{4}$$

and

$$\mathbf{G}\_{\rm T} = \mathbf{T}\_{\rm S} - \mathbf{T}\_{\rm E} \tag{5}$$

where the greenhouse strength GF is expressed in flux (W/m2) units, and GT is expressed as a temperature (TK) difference. The two greenhouse expressions are equivalent. GF expressed in flux units is more directly comparable to radiative modeling results and

 TE = [So (1 – A)/4]1/4 (3) Also needed is the information about the planet's surface temperature and the up-welling thermal flux that is emitted by the ground surface, both of which are measurable quantities. Direct measurements have been made by space probes that landed on the surfaces of Mars and Venus. Space based measurements are also possible by using selected spectral windows in the long-wave spectrum where radiation from the ground surface can escape directly out to space. Such measurements are a topic in remote sensing. They rely on associating the intensity of the measured spectral radiance with the equivalent spectral brightness of Planck radiation. By such means, microwave measurements of Venus were the first indication that the surface temperature of Venus was exceedingly hot, before verification by direct space probe measurements. A summary of planetary greenhouse parameters is listed in Table 1.

> **Parameter Mars Earth Venus**  TS(K) 215 288 730 TE(K) 210 255 230 TS4 (W/m2) 121 390 16,100 TE4 (W/m2) 111 240 157 GT(K) 5 33 500 GF(W/m2) 10 150 16,000 PS(bar) 0.01 1 100

The greenhouse parameters listed in Table 1 are all measurable quantities so that, at least in principle, the strength of the greenhouse effect of a planet is empirically determinable. As explained above, the effective temperature, TE(K), is actually a measure of either the emitted long-wave radiation at the top of the atmosphere, or the absorbed solar radiation by the planet, as given by Equation (3). Direct measurements of the globally averaged radiative fluxes at the top of the atmosphere and at the ground surface are actually quite difficult to obtain, particularly in view of the large variability of the solar and thermal radiative fluxes both geographically and with time, and also as a function of wavelength and viewing geometry. In practice, it is a combination of observational data analyses and theoretical modeling that is required to refine our knowledge of the planetary greenhouse parameters. Basically, it is the long-wave radiative fluxes at the top of the atmosphere and at the ground surface that define the greenhouse effect. The strength of the planetary greenhouse effect is expressed as the difference between the up-welling long-wave flux emitted by the ground

GF = TS4 – TE4 (4)

 GT = TS –TE (5) where the greenhouse strength GF is expressed in flux (W/m2) units, and GT is expressed as a temperature (TK) difference. The two greenhouse expressions are equivalent. GF expressed in flux units is more directly comparable to radiative modeling results and

Table 1. Planetary greenhouse parameters. (After Lacis et al., 2011).

surface, and the outgoing long-wave flux at the top of the atmosphere. Thus,

and

observational data. GT expresses the strength of the greenhouse effect directly in terms of the temperature difference by which the global mean surface temperature would be increased in comparison to the temperature that would exist if there were no greenhouse effect. The numbers given in Table 1 represent annually-averaged global-mean values. Obviously, the strength of the greenhouse effect varies with geographic location and with season. While the well-mixed non-condensing greenhouse gas contribution to the greenhouse effect may tend to be globally uniform and steady in time, the water vapor and the cloud feedback contribution to the greenhouse effect is strongly variable, geographically, seasonally, and even inter-annually.

It is interesting that of the numbers in Table 1, the outgoing long-wave fluxes of the three planets, as represented by the effective temperature, are so similar. More noteworthy are the factors of a 100 difference in the massiveness of the respective atmospheres, as represented by the surface pressure PS. In all three atmospheres, carbon dioxide is the key greenhouse gas. But the Earth is unique in that its greenhouse effect includes feedback amplification.

## **3. Geological perspective**

It is the combination of solar radiative heating and the strength of the greenhouse effect that determines the surface temperature of a planet. In the early days of the solar system some 4.5 billion years ago, the Sun had only about 75% of its present luminosity. But, as hydrogen in the interior of the Sun gets transformed into heavier elements, this causes the temperature and luminosity of the Sun increase, which in turn produces a slow and steady rise in the energy input to the terrestrial climate system. This is sometimes called the astrophysical climate forcing, and this type of solar radiative forcing will continue unabated. In several billion years time it will make the Earth too hot to be habitable.

While changes in solar insolation act to regulate the inflow of energy to the atmosphere and ground, it is the strength of the greenhouse effect that exerts its influence on the terrestrial climate system by regulating the energy outflow, and by generating additional heating of the ground surface by down-welling long-wave radiation. Here, carbon dioxide plays the leading role. In the early days of the Earth when the Sun was much less luminous, high concentrations of carbon dioxide where able to keep the temperature habitable for life forms to get established and to take root.

Over million-year time scales, volcanoes are the principal source of atmospheric carbon dioxide, and weathering of rocks is the principal sink (Berner, 2004), with the biosphere of the Earth also participating both as a source and as a sink. Because the principal sources operate independently of each other, the atmospheric level of carbon dioxide can fluctuate unchecked. There is geological evidence that shows glaciation was taking place at tropical latitudes 650 to 750 million years ago (Kirschvink, 1992), suggesting that snowball Earth conditions might have existed at that time. If, over millions of years, rock weathering were to exceed volcanic replenishment, atmospheric carbon dioxide could fall below its critical value, thus sending the Earth to snowball conditions. Eventually, with the ocean covered by ice, and thus prevented from taking up carbon dioxide, sufficient buildup of atmospheric carbon dioxide from volcanic eruptions would gradually bring the Earth out of its snowball state climate.

Greenhouse Effect 281

it was photolyzed by the intense sunlight and the liberated hydrogen was lost to space. There is thus the lingering suspicion that the current climate on Venus might well be attributed to the absence of having had appropriate life forms that could have prevented the critical buildup of carbon dioxide and thus prevented the runaway greenhouse effect that

The basic understanding of the greenhouse effect mechanism was first described in 1824 by the French mathematician and physicist Joseph Fourier (Fourier, 1824). This was also that period in time when conservation of energy, the most basic and fundamental concept of physics, was being formulated and quantified. Solar energy from the Sun is absorbed, and warms the Earth. An equal amount of thermal energy must then be radiated to space by the Earth in order to maintain its equilibrium energy balance. Otherwise the temperature of Earth would either keep rising indefinitely, or decrease indefinitely. In this spirit, Fourier performed experiments on atmospheric heat flow and pondered the question of how the Earth stays warm enough for plant and animal life to thrive. Fourier realized that much of the thermal radiation emitted by the Earth's surface was being absorbed within the Earth's atmosphere, and that some of this absorbed radiation was then being re-emitted downward, providing additional warming of the ground surface over and above that attributable just to

In 1863, the Irish physicist John Tyndall (Fleming, 1998) provided experimental support for Fourier's greenhouse idea, demonstrating by means of quantitative spectroscopy that the common atmospheric trace gases, such as water vapor, ozone, and carbon dioxide, are strong absorbers and emitters of thermal radiant energy, but are essentially transparent to visible sunlight. It was clear to Tyndall from his measurements that water vapor was the strongest absorber of thermal radiation, and therefore, the most influential atmospheric gas controlling the Earth's surface temperature. Meanwhile on the other hand, the principal components of the atmosphere, nitrogen and oxygen, were found to be radiatively inactive, providing only the atmospheric framework and temperature structure within which water vapor and carbon dioxide can exert their radiative influence. With this understanding of the radiative properties of the absorbing gases in the atmosphere, Tyndall speculated in 1861 that a reduction in atmospheric carbon dioxide could in fact induce an ice-age climate.

Utilizing the work of Tyndall and the careful measurements of heat transmission through the atmosphere compiled by the American astronomer Samuel Langley, the Swedish chemist and physicist Svante Arrhenius was the first to develop a quantitative mathematical framework of the terrestrial greenhouse effect (Arrhenius, 1896). Arrhenius published his heat-balance calculations of the Earth's sensitivity to carbon dioxide change in 1896. His radiative modeling results were remarkably similar to our current understanding of how the terrestrial greenhouse effect keeps the surface temperature some 33C (or 60F) warmer than it otherwise would be. Given that Arrhenius' basic interest was to explain the likely causes of ice-age climate, his greenhouse model was quite successful. He showed that reducing atmospheric carbon dioxide by a third would cool the global surface temperatures by −3°C (−5.5°F), and that doubling carbon dioxide would cause the tropical latitudes to warm by 5°C (9°F), with somewhat larger warming in polar regions, results that are in surprisingly

has rendered Venus inhospitable to life.

**4. Historical understanding** 

the direct absorption of solar energy.

The Cenozoic era, comprising the past 65.5 million years, marks the Cretaceous extinction of dinosaurs and the auspicious rise of mammals. During this era, the level of atmospheric concentration of carbon dioxide peaked near 3000 parts per million roughly 50 million years ago (Royer, 2006). This coincided with peak global warming when the global mean ocean temperature was some 12C warmer than present (Hansen et al., 2008). The information on carbon dioxide concentrations comes from carbon isotope analysis obtained from ocean core data (Fletcher, 2008), and temperature information is deduced from oxygen-18 variations in the ocean core data. It so happens that about two tenths of one percent of atmospheric oxygen is oxygen-18, or heavy oxygen with ten neutrons instead of the nominal eight. Since oxygen-18 concentration relative to oxygen-16 is systematically temperature dependent, this ratio can therefore be used as a geological thermometer to estimate global temperature changes in the geological record.

It is generally believed that it was enhanced weathering of rocks associated with the formation of the Himalayas that brought the atmospheric carbon dioxide down to a critical 450 ppm level, at which point the Antarctic polar ice cap began to form about 34 million years ago. The geological record clearly shows the close coupling between the level of atmospheric carbon dioxide and the global temperature of Earth, suggesting a thermostat-type control by atmospheric carbon dioxide in regulating the surface temperature of Earth. All along, the terrestrial biosphere has also played a very critical role in the time evolution of the global carbon cycle by having sequestered most of Earth's carbon in the form of limestone, and not allowed it to accumulate in the atmosphere as was apparently the case on Venus.

On thousand-year time scales, both the Antarctic and Greenland ice core data show that atmospheric carbon dioxide has fluctuated between 180 to 300 ppm over the glacialinterglacial cycles during the past 650,000 years (Jansen et al., 2007), with close coupling between the carbon dioxide concentration, ice volume, sea level, and global temperature. The relevant physical processes that control atmospheric carbon dioxide on thousand-year timescales between the glacial and interglacial extremes are not yet fully under-stood, but appear to involve the biosphere, ocean chemistry, and ocean circulation, with a significant role for Milankovitch variations in the Earth-orbital parameters which alter the seasonal distribution of incident solar radiation in the polar regions. While in the context of current climate, carbon dioxide is clearly an externally imposed radiative forcing on the climate system (as the direct result of fossil fuel burning) in the context of recorded human history, within the longer geological record, carbon dioxide is seen to be a variable quantity, albeit on a very slow time scale.

In some ways Venus is the twin sister of Earth. Both planets have similar mass and size, similar composition, similar amounts of carbon, an a generally similar location within the solar system, Based on this, there is well-grounded speculation, supported by isotope evidence, that originally Venus may have supported an ocean and an atmosphere similar to that of Earth. The suggestion is that for some reason Venus was not able to sequester its store of carbon in the form of limestone as Earth was able to accomplish. In this scenario, carbon dioxide arising from volcanic eruptions continued to accumulate within the Venusian atmosphere, increasing its greenhouse effect to the point where eventually all ocean water was vaporized. Water vapor then was carried high into the atmosphere where

The Cenozoic era, comprising the past 65.5 million years, marks the Cretaceous extinction of dinosaurs and the auspicious rise of mammals. During this era, the level of atmospheric concentration of carbon dioxide peaked near 3000 parts per million roughly 50 million years ago (Royer, 2006). This coincided with peak global warming when the global mean ocean temperature was some 12C warmer than present (Hansen et al., 2008). The information on carbon dioxide concentrations comes from carbon isotope analysis obtained from ocean core data (Fletcher, 2008), and temperature information is deduced from oxygen-18 variations in the ocean core data. It so happens that about two tenths of one percent of atmospheric oxygen is oxygen-18, or heavy oxygen with ten neutrons instead of the nominal eight. Since oxygen-18 concentration relative to oxygen-16 is systematically temperature dependent, this ratio can therefore be used as a geological thermometer to estimate global temperature

It is generally believed that it was enhanced weathering of rocks associated with the formation of the Himalayas that brought the atmospheric carbon dioxide down to a critical 450 ppm level, at which point the Antarctic polar ice cap began to form about 34 million years ago. The geological record clearly shows the close coupling between the level of atmospheric carbon dioxide and the global temperature of Earth, suggesting a thermostat-type control by atmospheric carbon dioxide in regulating the surface temperature of Earth. All along, the terrestrial biosphere has also played a very critical role in the time evolution of the global carbon cycle by having sequestered most of Earth's carbon in the form of limestone, and not allowed it to accumulate in the atmosphere as

On thousand-year time scales, both the Antarctic and Greenland ice core data show that atmospheric carbon dioxide has fluctuated between 180 to 300 ppm over the glacialinterglacial cycles during the past 650,000 years (Jansen et al., 2007), with close coupling between the carbon dioxide concentration, ice volume, sea level, and global temperature. The relevant physical processes that control atmospheric carbon dioxide on thousand-year timescales between the glacial and interglacial extremes are not yet fully under-stood, but appear to involve the biosphere, ocean chemistry, and ocean circulation, with a significant role for Milankovitch variations in the Earth-orbital parameters which alter the seasonal distribution of incident solar radiation in the polar regions. While in the context of current climate, carbon dioxide is clearly an externally imposed radiative forcing on the climate system (as the direct result of fossil fuel burning) in the context of recorded human history, within the longer geological record, carbon dioxide is seen to be a variable

In some ways Venus is the twin sister of Earth. Both planets have similar mass and size, similar composition, similar amounts of carbon, an a generally similar location within the solar system, Based on this, there is well-grounded speculation, supported by isotope evidence, that originally Venus may have supported an ocean and an atmosphere similar to that of Earth. The suggestion is that for some reason Venus was not able to sequester its store of carbon in the form of limestone as Earth was able to accomplish. In this scenario, carbon dioxide arising from volcanic eruptions continued to accumulate within the Venusian atmosphere, increasing its greenhouse effect to the point where eventually all ocean water was vaporized. Water vapor then was carried high into the atmosphere where

changes in the geological record.

was apparently the case on Venus.

quantity, albeit on a very slow time scale.

it was photolyzed by the intense sunlight and the liberated hydrogen was lost to space. There is thus the lingering suspicion that the current climate on Venus might well be attributed to the absence of having had appropriate life forms that could have prevented the critical buildup of carbon dioxide and thus prevented the runaway greenhouse effect that has rendered Venus inhospitable to life.

#### **4. Historical understanding**

The basic understanding of the greenhouse effect mechanism was first described in 1824 by the French mathematician and physicist Joseph Fourier (Fourier, 1824). This was also that period in time when conservation of energy, the most basic and fundamental concept of physics, was being formulated and quantified. Solar energy from the Sun is absorbed, and warms the Earth. An equal amount of thermal energy must then be radiated to space by the Earth in order to maintain its equilibrium energy balance. Otherwise the temperature of Earth would either keep rising indefinitely, or decrease indefinitely. In this spirit, Fourier performed experiments on atmospheric heat flow and pondered the question of how the Earth stays warm enough for plant and animal life to thrive. Fourier realized that much of the thermal radiation emitted by the Earth's surface was being absorbed within the Earth's atmosphere, and that some of this absorbed radiation was then being re-emitted downward, providing additional warming of the ground surface over and above that attributable just to the direct absorption of solar energy.

In 1863, the Irish physicist John Tyndall (Fleming, 1998) provided experimental support for Fourier's greenhouse idea, demonstrating by means of quantitative spectroscopy that the common atmospheric trace gases, such as water vapor, ozone, and carbon dioxide, are strong absorbers and emitters of thermal radiant energy, but are essentially transparent to visible sunlight. It was clear to Tyndall from his measurements that water vapor was the strongest absorber of thermal radiation, and therefore, the most influential atmospheric gas controlling the Earth's surface temperature. Meanwhile on the other hand, the principal components of the atmosphere, nitrogen and oxygen, were found to be radiatively inactive, providing only the atmospheric framework and temperature structure within which water vapor and carbon dioxide can exert their radiative influence. With this understanding of the radiative properties of the absorbing gases in the atmosphere, Tyndall speculated in 1861 that a reduction in atmospheric carbon dioxide could in fact induce an ice-age climate.

Utilizing the work of Tyndall and the careful measurements of heat transmission through the atmosphere compiled by the American astronomer Samuel Langley, the Swedish chemist and physicist Svante Arrhenius was the first to develop a quantitative mathematical framework of the terrestrial greenhouse effect (Arrhenius, 1896). Arrhenius published his heat-balance calculations of the Earth's sensitivity to carbon dioxide change in 1896. His radiative modeling results were remarkably similar to our current understanding of how the terrestrial greenhouse effect keeps the surface temperature some 33C (or 60F) warmer than it otherwise would be. Given that Arrhenius' basic interest was to explain the likely causes of ice-age climate, his greenhouse model was quite successful. He showed that reducing atmospheric carbon dioxide by a third would cool the global surface temperatures by −3°C (−5.5°F), and that doubling carbon dioxide would cause the tropical latitudes to warm by 5°C (9°F), with somewhat larger warming in polar regions, results that are in surprisingly

Greenhouse Effect 283

radiative flux measurements of the Earth's energy balance. The flux difference of 150 W/m2 between the up-welling surface flux and the long-wave flux at the top of the atmosphere that is emitted out to space, is a simple measure of the strength of the terrestrial greenhouse effect. As a further point, in terms of black body Planck radiation, the 240 W/m2 that is radiated out to space corresponds to a Planck temperature of 255 K (−18°C, −1°F), which is sometimes referred to as the effective temperature of the Earth. Thus, the temperature difference between the surface temperature and the effective temperature (33 K, 33°C, 60°F) provides another equivalent measure of the strength of the terrestrial greenhouse, one that is

The relative efficacy of any particular atmospheric greenhouse contributor is determined by how strongly it will absorb thermal radiation, and on how weakly the absorbed thermal radiation is re-emitted to space. According to the principles of radiative physics, the emissivity of a substance must be equal to its absorptivity. It is notable that the absorptivity of most absorbing materials has very little dependence on temperature, while the emission of thermal radiation is very strongly dependent on the temperature of the emitting body (varying approximately as the fourth power of the temperature of the radiating substance). As a result, clouds that are near the ground are poor contributors to the greenhouse effect because they absorb and re-emit thermal radiation at nearly the same temperature. Cirrus clouds, on the other hand, are very strong greenhouse contributors because they absorb the relatively high-temperature radiation emitted from the ground surface, but because they are located at a cold temperature, they emit to space only a small fraction of the radiation they have absorbed. This is the basic principle of efficient radiative operation of the greenhouse effect: absorb the up-welling high-temperature radiation, and emit radiation at a much colder temperature. This kind of radiative interaction makes for efficient trapping of heat

The modern approach to the study of Earth's climate and of the human impact on climate began in the 1930s with the work of Guy Callendar (Fleming, 1998), a British engineer who very systematically documented the time trend of anthropogenic fossil fuel use and was able to make the connection between fossil fuel burning and the corresponding increase in atmospheric carbon dioxide. However, realistic modeling of the Earth's climate system did not become feasible until the 1960s and 1970s when the availability of high-speed largememory computing systems began to be applied to the analysis of what is really a very complex problem in atmospheric physics. At present time, capable climate models are being run at dozens of climate centers worldwide. Their modeling task is to simulate the workings of current climate system in great detail. The model generated climate is represented by the global maps and vertical profiles of evolving atmospheric temperature, wind, water vapor, clouds, aerosols, and greenhouse gases. The objective of the modeling effort is to understand how the climate system works. By demonstrating the ability to model the geographic and seasonal patterns and variability of temperature, winds, cloudiness and precipitation, this will establish confidence to model the changes in global climate that are occurring as the result of global warming due to the increase in atmospheric greenhouse gases produced by

expressed in terms of temperature.

within the atmosphere below.

**5.1 Modeling realism** 

human industrial activity.

close agreement with current climate modeling simulations for the expected change in global surface temperature in response to the doubling of carbon dioxide forcing.

In 1905, the American geologist Thomas Chamberlin made the important finding that the greenhouse contribution by atmospheric water vapor behaved as a positive feedback mechanism (Christianson, 1999). In his thinking, Chamberlin noted that surface heating by solar radiation, or surface heating due to other agents such as carbon dioxide, raises the atmospheric temperature and leads to evaporation of more water vapor. This extra amount of water vapor produces additional heating and the further evaporation of additional water vapor. When the added heat source is taken away, any water vapor that is in excess of its equilibrium amount will precipitate from the atmosphere. Clearly, this is not a runaway situation. Rather, the change in the externally applied heating simply results in the establishment of a new equilibrium for the amount of water vapor that the atmosphere can hold. Since water vapor is an efficient greenhouse gas, the net effect of this interaction produces a significantly larger temperature change than would otherwise be the case if the amount of atmospheric water vapor did not change, demonstrating that it is carbon dioxide (and the other non-condensing greenhouse gases) that are the controlling factor of the terrestrial greenhouse effect. As a result, atmospheric water vapor and clouds act as feedback effects that magnify the greenhouse heating that is initially supplied by the noncondensing greenhouse gases, or by other radiative forcings.

## **5. Terrestrial greenhouse effect**

In the terrestrial greenhouse, carbon dioxide accounts for approximately 7°C (13°F) of the total 33°C (60°F) greenhouse effect. The other non-condensing greenhouse gases, i.e., ozone, methane, nitrous oxide, and anthropogenic chlorofluorocarbons, add another 3–4°C (5–7°F) to the overall global greenhouse strength. Because clouds absorb thermal radiation at all wavelengths of the thermal spectrum, this makes clouds into important contributors to the terrestrial greenhouse effect. Even aerosols, in particular the larger sized desert dust particles, with their substantial absorption and scattering at thermal wavelengths, make a small but non-negligible, contribution to the terrestrial greenhouse effect. Also, the cooling efficiency of volcanic aerosols is reduced by about 10% because of their non-negligible greenhouse effect (Lacis and Mishchenko, 1995).

The greenhouse effect is intimately an integral part of the global energy balance within the climate system. Basically, it is the greenhouse effect that determines how the long-wave thermal energy is transformed from its initial emission from the ground surface to its final destination out to space from the top of the atmosphere. In the process, this effectively defines the strength of the terrestrial greenhouse effect and determines the magnitude of the greenhouse warming over and above what the surface temperature would be if the ground surface were determined by simply being in equilibrium with the absorbed sunlight. In general terms, the nominal global mean surface temperature of the Earth is about 288 K (15°C, 59°F). The corresponding thermal radiation emitted by the ground surface is very closely described by the 'black body', or Planck emission, which is proportional to the fourth power of the surface temperature. This yields 390 W/m2 for the up-welling long-wave flux emitted by the ground surface. However, the long-wave flux to space leaving the top of the atmosphere is about 240 W/m2, which has been verified by satellite-based long-wave

close agreement with current climate modeling simulations for the expected change in

In 1905, the American geologist Thomas Chamberlin made the important finding that the greenhouse contribution by atmospheric water vapor behaved as a positive feedback mechanism (Christianson, 1999). In his thinking, Chamberlin noted that surface heating by solar radiation, or surface heating due to other agents such as carbon dioxide, raises the atmospheric temperature and leads to evaporation of more water vapor. This extra amount of water vapor produces additional heating and the further evaporation of additional water vapor. When the added heat source is taken away, any water vapor that is in excess of its equilibrium amount will precipitate from the atmosphere. Clearly, this is not a runaway situation. Rather, the change in the externally applied heating simply results in the establishment of a new equilibrium for the amount of water vapor that the atmosphere can hold. Since water vapor is an efficient greenhouse gas, the net effect of this interaction produces a significantly larger temperature change than would otherwise be the case if the amount of atmospheric water vapor did not change, demonstrating that it is carbon dioxide (and the other non-condensing greenhouse gases) that are the controlling factor of the terrestrial greenhouse effect. As a result, atmospheric water vapor and clouds act as feedback effects that magnify the greenhouse heating that is initially supplied by the non-

In the terrestrial greenhouse, carbon dioxide accounts for approximately 7°C (13°F) of the total 33°C (60°F) greenhouse effect. The other non-condensing greenhouse gases, i.e., ozone, methane, nitrous oxide, and anthropogenic chlorofluorocarbons, add another 3–4°C (5–7°F) to the overall global greenhouse strength. Because clouds absorb thermal radiation at all wavelengths of the thermal spectrum, this makes clouds into important contributors to the terrestrial greenhouse effect. Even aerosols, in particular the larger sized desert dust particles, with their substantial absorption and scattering at thermal wavelengths, make a small but non-negligible, contribution to the terrestrial greenhouse effect. Also, the cooling efficiency of volcanic aerosols is reduced by about 10% because of their non-negligible

The greenhouse effect is intimately an integral part of the global energy balance within the climate system. Basically, it is the greenhouse effect that determines how the long-wave thermal energy is transformed from its initial emission from the ground surface to its final destination out to space from the top of the atmosphere. In the process, this effectively defines the strength of the terrestrial greenhouse effect and determines the magnitude of the greenhouse warming over and above what the surface temperature would be if the ground surface were determined by simply being in equilibrium with the absorbed sunlight. In general terms, the nominal global mean surface temperature of the Earth is about 288 K (15°C, 59°F). The corresponding thermal radiation emitted by the ground surface is very closely described by the 'black body', or Planck emission, which is proportional to the fourth power of the surface temperature. This yields 390 W/m2 for the up-welling long-wave flux emitted by the ground surface. However, the long-wave flux to space leaving the top of the atmosphere is about 240 W/m2, which has been verified by satellite-based long-wave

global surface temperature in response to the doubling of carbon dioxide forcing.

condensing greenhouse gases, or by other radiative forcings.

**5. Terrestrial greenhouse effect** 

greenhouse effect (Lacis and Mishchenko, 1995).

radiative flux measurements of the Earth's energy balance. The flux difference of 150 W/m2 between the up-welling surface flux and the long-wave flux at the top of the atmosphere that is emitted out to space, is a simple measure of the strength of the terrestrial greenhouse effect. As a further point, in terms of black body Planck radiation, the 240 W/m2 that is radiated out to space corresponds to a Planck temperature of 255 K (−18°C, −1°F), which is sometimes referred to as the effective temperature of the Earth. Thus, the temperature difference between the surface temperature and the effective temperature (33 K, 33°C, 60°F) provides another equivalent measure of the strength of the terrestrial greenhouse, one that is expressed in terms of temperature.

The relative efficacy of any particular atmospheric greenhouse contributor is determined by how strongly it will absorb thermal radiation, and on how weakly the absorbed thermal radiation is re-emitted to space. According to the principles of radiative physics, the emissivity of a substance must be equal to its absorptivity. It is notable that the absorptivity of most absorbing materials has very little dependence on temperature, while the emission of thermal radiation is very strongly dependent on the temperature of the emitting body (varying approximately as the fourth power of the temperature of the radiating substance). As a result, clouds that are near the ground are poor contributors to the greenhouse effect because they absorb and re-emit thermal radiation at nearly the same temperature. Cirrus clouds, on the other hand, are very strong greenhouse contributors because they absorb the relatively high-temperature radiation emitted from the ground surface, but because they are located at a cold temperature, they emit to space only a small fraction of the radiation they have absorbed. This is the basic principle of efficient radiative operation of the greenhouse effect: absorb the up-welling high-temperature radiation, and emit radiation at a much colder temperature. This kind of radiative interaction makes for efficient trapping of heat within the atmosphere below.

#### **5.1 Modeling realism**

The modern approach to the study of Earth's climate and of the human impact on climate began in the 1930s with the work of Guy Callendar (Fleming, 1998), a British engineer who very systematically documented the time trend of anthropogenic fossil fuel use and was able to make the connection between fossil fuel burning and the corresponding increase in atmospheric carbon dioxide. However, realistic modeling of the Earth's climate system did not become feasible until the 1960s and 1970s when the availability of high-speed largememory computing systems began to be applied to the analysis of what is really a very complex problem in atmospheric physics. At present time, capable climate models are being run at dozens of climate centers worldwide. Their modeling task is to simulate the workings of current climate system in great detail. The model generated climate is represented by the global maps and vertical profiles of evolving atmospheric temperature, wind, water vapor, clouds, aerosols, and greenhouse gases. The objective of the modeling effort is to understand how the climate system works. By demonstrating the ability to model the geographic and seasonal patterns and variability of temperature, winds, cloudiness and precipitation, this will establish confidence to model the changes in global climate that are occurring as the result of global warming due to the increase in atmospheric greenhouse gases produced by human industrial activity.

Greenhouse Effect 285

thermal equilibrium of the atmosphere, more realistic temperature profiles could be

The need to have a realistic atmospheric structure in order to be able to calculate accurate radiative fluxes is self-evident. It is also clear that the terrestrial atmosphere is very complex in both the radiative properties of the atmospheric constituents as well as the atmospheric dynamics and thermodynamics that are involved in producing the resultant temperature structure. In the early models, improvisation and approximation were the rule, primarily because the computer resources to do better were not available. In current state-of-the-art climate models the problems of determining the atmospheric structure, including as well the vertical distribution of water vapor, clouds, greenhouse gases, and temperature, all of which vary geographically and with time, are gradually beginning to

A rather full description of our current climate modeling capabilities, including the success in our ability to model the terrestrial climate system, and our ability to address the looming problems of global warming and climate change that confront the future of humanity is contained in the *IPCC AR4 (2007) Report of the Intergovernmental Panel on Climate Change*  (Solomon et al, 2007). The overall objective is to be able to simulate the basic behavior of the climate system in terms of fundamental physics principles. This means that the evaporation and condensation of water vapor is computed based on local temperature, pressure, wind speed, relative humidity, and surface conditions. Atmospheric winds are calculated from hydrodynamics principles and include dependence on pressure, temperature, topography, and surface roughness. The incident solar radiation is calculated at each geographic point depending on its latitude and longitude, and also includes the effects of changing Earth-Sun distance and variation in orbital parameters. All this is being done with the highest spatial, vertical and time resolution that is supported by current computer speed and memory. This typically means that current climate models might employ a horizontal resolution of 50-500 km, support 20-100 vertical layers, and employ an 0.5-3 hour time resolution. Such climate models generate a lot of simulated 'climate' data which then can be compared to similarly voluminous observational data of spatial and seasonal maps of temperature, water vapor, cloud, and precipitation variations to critically evaluate the climate modeling performance. If a climate model can reproduce the observational data with reasonable fidelity, there is then reason to believe that such a model might be useful for analyzing past trends of climate

One example of such a model is the GISS ModelE climate model (Schmidt et al., 2006). This climate model can be operated at different resolutions for the horizontal, vertical, and time dimensions. It is coupled to a fully interacting ocean model to provide an accurate rendering of the temporal response of the climate system to changes in radiative forcing. The model produces realistic simulations of current climate temperature, precipitation, water vapor, and clouds. More specifically, the GISS ModelE produces realistic atmospheric temperature profiles, and realistic horizontal and vertical distributions of water vapor and clouds – all the information that is necessary to calculate accurate radiative fluxes and atmospheric

obtained (Manabe and Wetherald, 1967).

be addressed in realistic fashion.

change, and also in making prediction for the future.

heating and cooling rates.

**5.3 Current climate models** 

Since the greenhouse problem is largely radiative in nature, it follows that effective radiative modeling is the key. Unfortunately, the spectral absorption properties of greenhouse gases is very complex, consisting of many thousands of individual spectral lines of widely varying strengths, grouped in different sized often overlapping bunches, spread across the entire long-wave spectrum. Moreover, the width of each line is strongly pressure dependent, and there is a temperature dependence of the absorption strength as well. Before the advent of large computers, it was common practice then to rely on empirical formulas and statistical band models to represent the spectral absorption properties of the different atmospheric gases. The atmospheric temperature profile is the one additional, and absolutely essential factor that is required in order to perform the radiative calculations. Given a multi-layered model of the atmosphere, the radiative calculations yield heating rate profiles of solar radiation and cooling rate profiles of thermal radiation. By means of brute force time marching, the results of the heating and cooling rate calculations are applied over a short time interval to recalculate the atmospheric temperature profile until the radiative heating and cooling rates balance each other. Such atmospheric heat balancing calculation (Manabe and Moller, 1961) produce an atmospheric temperature profile that is in radiative equilibrium.

Radiative equilibrium might be adequate to represent the Martian atmosphere, but not that of Earth. The atmospheric lapse rate, or the rate that the atmospheric temperature decreases with height, has a defining role in determining the strength of the terrestrial greenhouse effect. The temperature structure of the atmosphere becomes established as the direct result of radiative heating and cooling, including also the effects of atmospheric dynamics and thermodynamics. If radiation were the only means of energy transport in the atmosphere, given the large thermal opacity of the atmosphere, the resulting temperature gradient needs to be very steep in order to export all of the energy radiated by the ground surface out to space. The greenhouse effect corresponding to such a 'radiation only' atmosphere would be about 66°C (120°F), or about twice its actual value of 33°C (60°F). But such a purely radiative temperature gradient would be dynamically unstable against convection. Convection would set in, and heat energy would be transported upward by a newly enabled parallel means of energy transport to establish a less-steep temperature gradient that would be stable against further convection. This problem was resolved by including a 'convective adjustment' to the atmospheric temperature gradient (Manabe and Strickler, 1964) to produce a realistic profile for the atmospheric temperature, and thus provide a realistic atmospheric structure for the calculation of the radiative fluxes.

#### **5.2 Temperature lapse rate**

In the Earth's atmosphere, water vapor plays a very important role in energy transport by convection. This is because water vapor contains latent heat energy which is released to the atmosphere when water vapor condenses, thus making energy transport by convection that much more energetic than by dry gas alone. The net effect of including water vapor in the convective energy transport establishes what is called the 'moist adiabatic' temperature gradient, a very persistent characteristic of the terrestrial atmosphere. It had been noticed from meteorological observations that as temperatures changed, the moisture content in the atmosphere changed by large amounts, but that the relative humidity profile tended to be more stable. By imposing a condition of 'fixed relative humidity' when calculating the

Since the greenhouse problem is largely radiative in nature, it follows that effective radiative modeling is the key. Unfortunately, the spectral absorption properties of greenhouse gases is very complex, consisting of many thousands of individual spectral lines of widely varying strengths, grouped in different sized often overlapping bunches, spread across the entire long-wave spectrum. Moreover, the width of each line is strongly pressure dependent, and there is a temperature dependence of the absorption strength as well. Before the advent of large computers, it was common practice then to rely on empirical formulas and statistical band models to represent the spectral absorption properties of the different atmospheric gases. The atmospheric temperature profile is the one additional, and absolutely essential factor that is required in order to perform the radiative calculations. Given a multi-layered model of the atmosphere, the radiative calculations yield heating rate profiles of solar radiation and cooling rate profiles of thermal radiation. By means of brute force time marching, the results of the heating and cooling rate calculations are applied over a short time interval to recalculate the atmospheric temperature profile until the radiative heating and cooling rates balance each other. Such atmospheric heat balancing calculation (Manabe and Moller, 1961) produce an

Radiative equilibrium might be adequate to represent the Martian atmosphere, but not that of Earth. The atmospheric lapse rate, or the rate that the atmospheric temperature decreases with height, has a defining role in determining the strength of the terrestrial greenhouse effect. The temperature structure of the atmosphere becomes established as the direct result of radiative heating and cooling, including also the effects of atmospheric dynamics and thermodynamics. If radiation were the only means of energy transport in the atmosphere, given the large thermal opacity of the atmosphere, the resulting temperature gradient needs to be very steep in order to export all of the energy radiated by the ground surface out to space. The greenhouse effect corresponding to such a 'radiation only' atmosphere would be about 66°C (120°F), or about twice its actual value of 33°C (60°F). But such a purely radiative temperature gradient would be dynamically unstable against convection. Convection would set in, and heat energy would be transported upward by a newly enabled parallel means of energy transport to establish a less-steep temperature gradient that would be stable against further convection. This problem was resolved by including a 'convective adjustment' to the atmospheric temperature gradient (Manabe and Strickler, 1964) to produce a realistic profile for the atmospheric temperature, and thus provide a realistic atmospheric structure for the

In the Earth's atmosphere, water vapor plays a very important role in energy transport by convection. This is because water vapor contains latent heat energy which is released to the atmosphere when water vapor condenses, thus making energy transport by convection that much more energetic than by dry gas alone. The net effect of including water vapor in the convective energy transport establishes what is called the 'moist adiabatic' temperature gradient, a very persistent characteristic of the terrestrial atmosphere. It had been noticed from meteorological observations that as temperatures changed, the moisture content in the atmosphere changed by large amounts, but that the relative humidity profile tended to be more stable. By imposing a condition of 'fixed relative humidity' when calculating the

atmospheric temperature profile that is in radiative equilibrium.

calculation of the radiative fluxes.

**5.2 Temperature lapse rate** 

thermal equilibrium of the atmosphere, more realistic temperature profiles could be obtained (Manabe and Wetherald, 1967).

The need to have a realistic atmospheric structure in order to be able to calculate accurate radiative fluxes is self-evident. It is also clear that the terrestrial atmosphere is very complex in both the radiative properties of the atmospheric constituents as well as the atmospheric dynamics and thermodynamics that are involved in producing the resultant temperature structure. In the early models, improvisation and approximation were the rule, primarily because the computer resources to do better were not available. In current state-of-the-art climate models the problems of determining the atmospheric structure, including as well the vertical distribution of water vapor, clouds, greenhouse gases, and temperature, all of which vary geographically and with time, are gradually beginning to be addressed in realistic fashion.

#### **5.3 Current climate models**

A rather full description of our current climate modeling capabilities, including the success in our ability to model the terrestrial climate system, and our ability to address the looming problems of global warming and climate change that confront the future of humanity is contained in the *IPCC AR4 (2007) Report of the Intergovernmental Panel on Climate Change*  (Solomon et al, 2007). The overall objective is to be able to simulate the basic behavior of the climate system in terms of fundamental physics principles. This means that the evaporation and condensation of water vapor is computed based on local temperature, pressure, wind speed, relative humidity, and surface conditions. Atmospheric winds are calculated from hydrodynamics principles and include dependence on pressure, temperature, topography, and surface roughness. The incident solar radiation is calculated at each geographic point depending on its latitude and longitude, and also includes the effects of changing Earth-Sun distance and variation in orbital parameters. All this is being done with the highest spatial, vertical and time resolution that is supported by current computer speed and memory. This typically means that current climate models might employ a horizontal resolution of 50-500 km, support 20-100 vertical layers, and employ an 0.5-3 hour time resolution. Such climate models generate a lot of simulated 'climate' data which then can be compared to similarly voluminous observational data of spatial and seasonal maps of temperature, water vapor, cloud, and precipitation variations to critically evaluate the climate modeling performance. If a climate model can reproduce the observational data with reasonable fidelity, there is then reason to believe that such a model might be useful for analyzing past trends of climate change, and also in making prediction for the future.

One example of such a model is the GISS ModelE climate model (Schmidt et al., 2006). This climate model can be operated at different resolutions for the horizontal, vertical, and time dimensions. It is coupled to a fully interacting ocean model to provide an accurate rendering of the temporal response of the climate system to changes in radiative forcing. The model produces realistic simulations of current climate temperature, precipitation, water vapor, and clouds. More specifically, the GISS ModelE produces realistic atmospheric temperature profiles, and realistic horizontal and vertical distributions of water vapor and clouds – all the information that is necessary to calculate accurate radiative fluxes and atmospheric heating and cooling rates.

Greenhouse Effect 287

H2O 94.7 .458 59.8 .534 74.7 .490 cloud 56.2 .272 23.2 .207 37.3 .244 CO2 40.3 .195 24.0 .214 31.0 .203 O3 4.0 .019 1.7 .015 2.7 .018 N2O 4.1 .020 1.5 .013 2.6 .017 CH4 3.5 .017 1.2 .011 2.2 .014 CFCs 1.0 .005 0.2 .002 0.6 .004 aerosol 3.0 .014 0.4 .004 1.5 .010 sum 206.8 1.000 112.0 1.000 152.6 1.000

Table 2. Single-addition & single-subtraction LW flux changes. (After Lacis et al., 2010)

contributions have been normalized to unity.

are similarly normalized to unity.

The results in Table 2 were obtained by starting with a one-year's worth of ModelE climate data which had been run to equilibrium for 1980 atmospheric conditions. For this model, the globally averaged annual-mean long-wave up-welling flux at the ground surface was 393.9 W/m2 (TS = 288.7 K), and the long-wave flux at the top of the atmosphere emitted to space was 241.3 W/m2 (TE = 255.4 K), with the total strength of the greenhouse effect GF = 152.6 W/m2 (GT = 33.3 K). Upon zeroing out all atmospheric absorbers, the outgoing long-wave flux to space was that emitted by the ground surface, 393.9 W/m2. The absorbers were then added in individually one at a time, and the reduction in outgoing flux re-evaluated. These are the global annual mean flux changes listed in the left hand single-addition column. Thus, water vapor by itself reduced the outgoing flux by 94.7 W/m2, cloud alone by 56.2 W/m2, carbon dioxide alone by 40.3 W/m2, with smaller amounts for the minor greenhouse gases. The sum of these singleaddition contributions adds up to 206.8 W/m2, indicative of the substantial effect of nonlinear overlapping absorption since the sum total flux reduction with all greenhouse gases acting together is 152.6 W/m2. In the second column these fractional single-addition

Similarly, the single-subtraction columns were obtained by subtracting each individual atmospheric absorber one at a time from the full compliment of absorbers. In this case, the reference top of the atmosphere outgoing long-wave flux was 241.3 W/m2. By removing water vapor alone, the flux emitted to space increased by 59.8 W/m2, 23.2 W/m2 for cloud alone, 24.0 W/m2 for carbon dioxide alone, with smaller amounts for the minor contributors. Again, because of non-linear overlapping absorption effects, the individual flux changes only sum to 112.0 W/m2, instead of 152.6 W/m2. The single-subtraction results

The single-addition and single-subtraction results are then weighted by their closeness to the equilibrium GF value. Thus, by taking 0.572 times the group-subtraction values plus 0.428 times the single-addition values, we arrive at the normalized average results for the radiative flux and corresponding fractional contributions to the total greenhouse effect by

single-addition single-subtraction normalized average W/m2 fraction W/m2 fraction W/m2 fraction

LW absorber

### **5.4 Radiative transfer modeling**

The GISS ModelE radiative transfer model has been specifically designed to render accurate results for the radiative fluxes and the radiative heating and cooling rates throughout the atmosphere for any specified atmospheric temperature structure and absorbing gas and cloud distribution. The starting point for modeling the absorption by atmospheric gases is the HITRAN line atlas (Rothman et al., 2009) which contains absorption line strengths, line positions, widths, and energy levels for several hundred thousand lines for water vapor, carbon dioxide, ozone, and other atmospheric gases. This information is used in line-by-line models to calculate the spectral transmission and absorption by the different greenhouse gases, including the effects of overlapping absorption, with very high precision. The line-by-line results are time consuming, but they serve as the standard of reference to which much faster methods of calculation can be compared and validated for use in climate modeling applications. The correlated *k*distribution method (Lacis and Oinas, 1991) is a widely used approach for reproducing the line-by-line accuracy to well within 1% for all the greenhouse gases, including an accurate rendering of the non-linearities due to pressure broadening and the overlapping absorption for widely varying absorber amounts. For thermal radiation the effects of multiple scattering are accounted for as parameterized corrections, but they are modeled explicitly for solar radiation. All in all, the GISS modelE radiative transfer modeling is performed with high accuracy, as verified by a wide range of line-byline calculations.

The ModelE radiation treatment has been designed to be both comprehensive and flexible. At each grid point of ModelE, accurate radiative fluxes are calculated at each model time step in response to ever changing temperature profiles, clouds, aerosols, and greenhouse gases. These calculations are performed globally at thousands of grid points, and thousands of times annually. In this way, accurate values of globally and annually averaged fluxes at the ground surface and at the top of the atmosphere are obtained for a more convenient analysis of the terrestrial greenhouse effect and of the global energy balance, results that include a detailed and realistic rendering of the climate system and its seasonal variability. Moreover, with the current computer speed and memory capabilities, all of this data can be stored and reprocessed over and over again. This is where the radiation model flexibility comes into play. All of the different greenhouse gases, clouds, and aerosols can individually be selected or zeroed out, and the radiative calculations repeated to obtain a quantitative measure of the non-linear effects of overlapping absorption by different combinations of greenhouse gases.

#### **5.5 Greenhouse warming attribution**

While it is important to have an accurate measure of the total greenhouse strength for our current-climate atmosphere, it is equally important to have a clear understanding of the relative contribution that each of the atmospheric constituents makes to the total strength of the greenhouse effect. This task was performed with the GISS ModelE (Schmidt, et al., 2010) whereby the radiative effects of the different atmospheric constituents were evaluated individually one by one, so as to determine their relative importance. The results of these calculations are summarized in Table 2.

The GISS ModelE radiative transfer model has been specifically designed to render accurate results for the radiative fluxes and the radiative heating and cooling rates throughout the atmosphere for any specified atmospheric temperature structure and absorbing gas and cloud distribution. The starting point for modeling the absorption by atmospheric gases is the HITRAN line atlas (Rothman et al., 2009) which contains absorption line strengths, line positions, widths, and energy levels for several hundred thousand lines for water vapor, carbon dioxide, ozone, and other atmospheric gases. This information is used in line-by-line models to calculate the spectral transmission and absorption by the different greenhouse gases, including the effects of overlapping absorption, with very high precision. The line-by-line results are time consuming, but they serve as the standard of reference to which much faster methods of calculation can be compared and validated for use in climate modeling applications. The correlated *k*distribution method (Lacis and Oinas, 1991) is a widely used approach for reproducing the line-by-line accuracy to well within 1% for all the greenhouse gases, including an accurate rendering of the non-linearities due to pressure broadening and the overlapping absorption for widely varying absorber amounts. For thermal radiation the effects of multiple scattering are accounted for as parameterized corrections, but they are modeled explicitly for solar radiation. All in all, the GISS modelE radiative transfer modeling is performed with high accuracy, as verified by a wide range of line-byline calculations.

The ModelE radiation treatment has been designed to be both comprehensive and flexible. At each grid point of ModelE, accurate radiative fluxes are calculated at each model time step in response to ever changing temperature profiles, clouds, aerosols, and greenhouse gases. These calculations are performed globally at thousands of grid points, and thousands of times annually. In this way, accurate values of globally and annually averaged fluxes at the ground surface and at the top of the atmosphere are obtained for a more convenient analysis of the terrestrial greenhouse effect and of the global energy balance, results that include a detailed and realistic rendering of the climate system and its seasonal variability. Moreover, with the current computer speed and memory capabilities, all of this data can be stored and reprocessed over and over again. This is where the radiation model flexibility comes into play. All of the different greenhouse gases, clouds, and aerosols can individually be selected or zeroed out, and the radiative calculations repeated to obtain a quantitative measure of the non-linear effects of overlapping absorption by different combinations of

While it is important to have an accurate measure of the total greenhouse strength for our current-climate atmosphere, it is equally important to have a clear understanding of the relative contribution that each of the atmospheric constituents makes to the total strength of the greenhouse effect. This task was performed with the GISS ModelE (Schmidt, et al., 2010) whereby the radiative effects of the different atmospheric constituents were evaluated individually one by one, so as to determine their relative importance. The results of these

**5.4 Radiative transfer modeling** 

greenhouse gases.

**5.5 Greenhouse warming attribution** 

calculations are summarized in Table 2.


Table 2. Single-addition & single-subtraction LW flux changes. (After Lacis et al., 2010)

The results in Table 2 were obtained by starting with a one-year's worth of ModelE climate data which had been run to equilibrium for 1980 atmospheric conditions. For this model, the globally averaged annual-mean long-wave up-welling flux at the ground surface was 393.9 W/m2 (TS = 288.7 K), and the long-wave flux at the top of the atmosphere emitted to space was 241.3 W/m2 (TE = 255.4 K), with the total strength of the greenhouse effect GF = 152.6 W/m2 (GT = 33.3 K). Upon zeroing out all atmospheric absorbers, the outgoing long-wave flux to space was that emitted by the ground surface, 393.9 W/m2. The absorbers were then added in individually one at a time, and the reduction in outgoing flux re-evaluated. These are the global annual mean flux changes listed in the left hand single-addition column. Thus, water vapor by itself reduced the outgoing flux by 94.7 W/m2, cloud alone by 56.2 W/m2, carbon dioxide alone by 40.3 W/m2, with smaller amounts for the minor greenhouse gases. The sum of these singleaddition contributions adds up to 206.8 W/m2, indicative of the substantial effect of nonlinear overlapping absorption since the sum total flux reduction with all greenhouse gases acting together is 152.6 W/m2. In the second column these fractional single-addition contributions have been normalized to unity.

Similarly, the single-subtraction columns were obtained by subtracting each individual atmospheric absorber one at a time from the full compliment of absorbers. In this case, the reference top of the atmosphere outgoing long-wave flux was 241.3 W/m2. By removing water vapor alone, the flux emitted to space increased by 59.8 W/m2, 23.2 W/m2 for cloud alone, 24.0 W/m2 for carbon dioxide alone, with smaller amounts for the minor contributors. Again, because of non-linear overlapping absorption effects, the individual flux changes only sum to 112.0 W/m2, instead of 152.6 W/m2. The single-subtraction results are similarly normalized to unity.

The single-addition and single-subtraction results are then weighted by their closeness to the equilibrium GF value. Thus, by taking 0.572 times the group-subtraction values plus 0.428 times the single-addition values, we arrive at the normalized average results for the radiative flux and corresponding fractional contributions to the total greenhouse effect by

Greenhouse Effect 289

strongest of the non-condensing greenhouse gases in the atmosphere, the argument can be effectively made that carbon dioxide is indeed the principal control knob that governs the temperature of Earth. This has important consequences and bearing on our understanding of current climate change as well as the changes in climate that have taken place over

It is illuminating to examine the historically recent change in atmospheric carbon dioxide in the broader geological context. The ice core record shows that during the last ice age, when the global temperature was roughly 5°C (9°F) colder than current climate, the atmospheric carbon dioxide was at levels near 180 ppm. In fact, the Vostoc ice core data show four major ice-age cycles during the 420,000 year time span with the carbon dioxide level peaking near 280 ppm during the relatively brief interglacial warm interludes when global temperatures were similar to present day conditions. During the coldest extremes of the ice-age cycle, the carbon dioxide level is seen to drop to its lowest levels near 180 ppm. Hansen has used this geological data to establish an empirically based sensitivity to radiative forcings for the fastfeedback processes of the climate system. In his ice-core data analysis, Hansen finds the terrestrial climate sensitivity to be such that the global surface temperature warms about

The significance of all this information is becoming ever more clear, pointing to a growing global warming climate crisis. The geological record and climate modeling analyses show that carbon dioxide is indeed the principal control knob, or thermostat, that is instrumental in governing the temperature of Earth through its controlling influence on the terrestrial

The inescapable conclusion is that the anthropogenic forcings continue to add onto the growing strength of the terrestrial greenhouse effect. The continuing high rate of increase in atmospheric carbon dioxide is particularly worrisome because the present levels are near 390 ppm, far in excess of the more typical 280 ppm for the interglacial maxima. This raises concern of an approaching a 'tipping point', thought to be about 450 ppm, beyond which the radiative equilibrium of the Earth would no longer be capable of sustaining polar icecaps (Hansen et al., 2007). Moreover, since the atmospheric residence time of carbon dioxide is very long, and measured in centuries, this makes the reduction of atmospheric carbon dioxide a difficult undertaking once a decision is eventually reached to limit or reverse the

The present biosphere has not experienced climate under such levels of atmospheric carbon dioxide, so there is cause for concern regarding the survival rate of many species under the impending changes in Earth's climate that are being forced by human industrial activity. Because of the very large heat capacity of the ocean, full impact of these radiative forcings is slow to materialize, raising the false hope that climate change is not serious, and may not even be happening. Such opinions stem from the significant uncertainties in modeling the rate at which the climate is changing. Because of the non-linearities in feedback interactions, there is additional uncertainty in regard to sensitivity of the climate feedbacks that act to determine the magnitude of the eventual change in climate. Still

geological time scales.

**5.7 Present-day perspective** 

3°C per 4 W/m2 of radiative forcing.

growth of carbon dioxide emissions.

greenhouse effect.

the individual greenhouse contributors. In round numbers, of the total greenhouse effect for the current climate atmosphere, about 50% is directly attributable to water vapor, 25% to clouds, 20% to carbon dioxide, and the remaining 5% due to the minor greenhouse gases, i.e., methane, nitrous oxide, ozone, and chlorofluorocarbons.

#### **5.6 Forcing/feedback significance**

The most important conclusion to be drawn from this greenhouse attribution is that the individual radiative flux contributions to the total greenhouse effect can be gainfully subdivided into two separate groups: (1) the thermodynamically inactive non-condensing greenhouse gases (carbon dioxide and the other minor trace gases), which basically remain in the atmosphere once they enter there; and (2) the thermodynamically active condensing species (water vapor and clouds), which evaporate, condense, and precipitate, and thus respond quickly to changes in local temperature and other meteorological conditions. To be sure, the non-condensing greenhouse gases do interact chemically, but these chemical reactions are slow acting processes that result in atmospheric life times ranging from a decade to longer than several centuries. Because of this, the noncondensing greenhouse gases effectively behave as permanent fixtures of the atmosphere, and as such, provide a steady amount of radiative heating that is largely independent of varying meteorological conditions. The condensing species, on the other hand, adjust rapidly to changes in meteorological conditions, behaving as feedback effects that seek to establish the maximum allowable equilibrium concentration in accordance with the background atmospheric temperature structure that is provided by the non-condensing greenhouse gases.

Accordingly, the amount of water vapor that the atmosphere can sustain is temperature dependent, as dictated by the Clausius-Clapeyron equation. Basically, this means that a warmer atmosphere can sustain more water vapor, and a colder atmosphere less. Possession of strong long-wave absorption bands by water vapor make water vapor a strong positive feedback in response to radiative forcing produced by the non-condensing greenhouse gases. Since clouds are closely dependent on the amount of available water vapor, clouds are likewise classified as part of the fast feedback processes, with the added complication that clouds also reflect solar radiation, a cooling effect that offsets the cloud greenhouse effect. From these climate modeling analyses it is clear that the noncondensing greenhouse gases provide the sustaining radiative forcing for the terrestrial greenhouse effect, but that water vapor and clouds, in their role as climate feedbacks, act as amplifiers (by about a factor of four) of the radiative forcing provided by the noncondensing gases.

As further evidence to emphasize this very point, recent climate modeling experiments were performed to demonstrate even more clearly the feedback role of water vapor and clouds in the current climate system (Lacis et al., 2010). In these experiments all of the non-condensing greenhouse gases were set to zero. Without the radiative forcing temperature support by the non-condensing greenhouse gases, excess atmospheric water vapor rapidly condensed and precipitated from the atmosphere, plunging the Earth inevitably toward an ice bound state. This was a clear-cut demonstration that the non-condensing greenhouse gases do indeed control the magnitude of the terrestrial greenhouse effect. Since carbon dioxide is by far the

the individual greenhouse contributors. In round numbers, of the total greenhouse effect for the current climate atmosphere, about 50% is directly attributable to water vapor, 25% to clouds, 20% to carbon dioxide, and the remaining 5% due to the minor greenhouse gases,

The most important conclusion to be drawn from this greenhouse attribution is that the individual radiative flux contributions to the total greenhouse effect can be gainfully subdivided into two separate groups: (1) the thermodynamically inactive non-condensing greenhouse gases (carbon dioxide and the other minor trace gases), which basically remain in the atmosphere once they enter there; and (2) the thermodynamically active condensing species (water vapor and clouds), which evaporate, condense, and precipitate, and thus respond quickly to changes in local temperature and other meteorological conditions. To be sure, the non-condensing greenhouse gases do interact chemically, but these chemical reactions are slow acting processes that result in atmospheric life times ranging from a decade to longer than several centuries. Because of this, the noncondensing greenhouse gases effectively behave as permanent fixtures of the atmosphere, and as such, provide a steady amount of radiative heating that is largely independent of varying meteorological conditions. The condensing species, on the other hand, adjust rapidly to changes in meteorological conditions, behaving as feedback effects that seek to establish the maximum allowable equilibrium concentration in accordance with the background atmospheric temperature structure that is provided by the non-condensing

Accordingly, the amount of water vapor that the atmosphere can sustain is temperature dependent, as dictated by the Clausius-Clapeyron equation. Basically, this means that a warmer atmosphere can sustain more water vapor, and a colder atmosphere less. Possession of strong long-wave absorption bands by water vapor make water vapor a strong positive feedback in response to radiative forcing produced by the non-condensing greenhouse gases. Since clouds are closely dependent on the amount of available water vapor, clouds are likewise classified as part of the fast feedback processes, with the added complication that clouds also reflect solar radiation, a cooling effect that offsets the cloud greenhouse effect. From these climate modeling analyses it is clear that the noncondensing greenhouse gases provide the sustaining radiative forcing for the terrestrial greenhouse effect, but that water vapor and clouds, in their role as climate feedbacks, act as amplifiers (by about a factor of four) of the radiative forcing provided by the non-

As further evidence to emphasize this very point, recent climate modeling experiments were performed to demonstrate even more clearly the feedback role of water vapor and clouds in the current climate system (Lacis et al., 2010). In these experiments all of the non-condensing greenhouse gases were set to zero. Without the radiative forcing temperature support by the non-condensing greenhouse gases, excess atmospheric water vapor rapidly condensed and precipitated from the atmosphere, plunging the Earth inevitably toward an ice bound state. This was a clear-cut demonstration that the non-condensing greenhouse gases do indeed control the magnitude of the terrestrial greenhouse effect. Since carbon dioxide is by far the

i.e., methane, nitrous oxide, ozone, and chlorofluorocarbons.

**5.6 Forcing/feedback significance** 

greenhouse gases.

condensing gases.

strongest of the non-condensing greenhouse gases in the atmosphere, the argument can be effectively made that carbon dioxide is indeed the principal control knob that governs the temperature of Earth. This has important consequences and bearing on our understanding of current climate change as well as the changes in climate that have taken place over geological time scales.

#### **5.7 Present-day perspective**

It is illuminating to examine the historically recent change in atmospheric carbon dioxide in the broader geological context. The ice core record shows that during the last ice age, when the global temperature was roughly 5°C (9°F) colder than current climate, the atmospheric carbon dioxide was at levels near 180 ppm. In fact, the Vostoc ice core data show four major ice-age cycles during the 420,000 year time span with the carbon dioxide level peaking near 280 ppm during the relatively brief interglacial warm interludes when global temperatures were similar to present day conditions. During the coldest extremes of the ice-age cycle, the carbon dioxide level is seen to drop to its lowest levels near 180 ppm. Hansen has used this geological data to establish an empirically based sensitivity to radiative forcings for the fastfeedback processes of the climate system. In his ice-core data analysis, Hansen finds the terrestrial climate sensitivity to be such that the global surface temperature warms about 3°C per 4 W/m2 of radiative forcing.

The significance of all this information is becoming ever more clear, pointing to a growing global warming climate crisis. The geological record and climate modeling analyses show that carbon dioxide is indeed the principal control knob, or thermostat, that is instrumental in governing the temperature of Earth through its controlling influence on the terrestrial greenhouse effect.

The inescapable conclusion is that the anthropogenic forcings continue to add onto the growing strength of the terrestrial greenhouse effect. The continuing high rate of increase in atmospheric carbon dioxide is particularly worrisome because the present levels are near 390 ppm, far in excess of the more typical 280 ppm for the interglacial maxima. This raises concern of an approaching a 'tipping point', thought to be about 450 ppm, beyond which the radiative equilibrium of the Earth would no longer be capable of sustaining polar icecaps (Hansen et al., 2007). Moreover, since the atmospheric residence time of carbon dioxide is very long, and measured in centuries, this makes the reduction of atmospheric carbon dioxide a difficult undertaking once a decision is eventually reached to limit or reverse the growth of carbon dioxide emissions.

The present biosphere has not experienced climate under such levels of atmospheric carbon dioxide, so there is cause for concern regarding the survival rate of many species under the impending changes in Earth's climate that are being forced by human industrial activity. Because of the very large heat capacity of the ocean, full impact of these radiative forcings is slow to materialize, raising the false hope that climate change is not serious, and may not even be happening. Such opinions stem from the significant uncertainties in modeling the rate at which the climate is changing. Because of the non-linearities in feedback interactions, there is additional uncertainty in regard to sensitivity of the climate feedbacks that act to determine the magnitude of the eventual change in climate. Still

Greenhouse Effect 291

enhancement of the greenhouse effect by water vapor and cloud feedbacks (the fast feedback processes) is not directly in response to the amount of carbon dioxide in the atmosphere per se, but rather to the change in temperature that is generated by the carbon

There is also substantial natural variability within the climate system that produces global temperature fluctuations by several tenths of a degree on inter-annual and decadal time scales, including climate system response to the 11-year sunspot cycle. However, all of these temperature fluctuations are relative to a zero reference point, while the steady increase in the strength of the terrestrial greenhouse effect continues unabated. These fluctuations produce random looking changes in global temperature that tend to mask and obscure the steadily increasing global warming signal of carbon dioxide. Perhaps of more immediate concern are the likely prospects for an increase in the weather extremes such as more severe droughts and more severe floods, all fueled by the greater energy that is being added to the hydrological cycle as the global temperature rises and atmospheric water vapor increases.

There is no mistaking the fact that the level of atmospheric carbon dioxide is increasing, and this continued increase causes the strength of the terrestrial greenhouse effect to keep on increasing. Accordingly, there is no viable alternative to counteract the impending effects of global warming except through direct human action to reduce the level of atmospheric carbon dioxide, to control the strength of the terrestrial greenhouse effect which ultimately determines the surface temperature of Earth and the global climate that

Assessing the real cause and the perceived reality of global climate change has become a heated topic in the public discourse, particularly in the United States. Global warming has been denounced as the "greatest hoax ever" by prominent Senators and Congressmen. There are well organized efforts funded by fossil fuel interests to discredit climate science and to promote discord and mistrust in the public mind. The presumed rationale for this anti-climate science crusade by fossil fuel interests is perhaps similar to the reaction of the tobacco industry in response to the linking of deleterious health impacts to smoking. In their thinking, if global warming is determined to be harmful to human interests, the pressure to limit carbon dioxide emissions will be harmful to fossil fuel financial interests. On the other hand, there are environmental groups who already perceive global warming as a brewing environmental disaster that is already happening. They are demanding immediate action to curtail the use of fossil fuels. This brawling has attracted the attention of a broad spectrum of individuals from many different disciplines weighing in with opinions and suggestions as to how best understand what is really happening with global

The terrestrial climate system however is very complex, and is not really amenable to simple analyses and interpretations. So, for example, performing a 'more sophisticated' statistical analysis of the global temperature record will not yield quantitative attribution as the cause of global warming because the existing temperature record is much too short, and there are too many causative factors that are not adequately accounted for that impact the global

The more menacing future concern is the inevitable rise in sea level.

dioxide increase.

results therefrom.

climate.

**5.9 Public perception** 

larger uncertainties abound in modeling the space-time response of the climate system because horizontal energy transports tend to be several orders of magnitude larger than corresponding greenhouse gas radiative forcings that are the ultimate instigators of global climate change.

While there remain legitimate uncertainties regarding the details of time specific, local, and regional global climate change, these uncertainties should not be projected onto the solid evidence that human industrial activity is busy driving Earth's climate into uncharted territory. The atmospheric carbon dioxide control knob is currently being turned at a much faster rate than ever before.

In view of the pivotal role that carbon dioxide plays in governing the global temperature of Earth, a great deal of attention has being focused on obtaining accurate measurements of atmospheric carbon dioxide, and how carbon dioxide has been changing with time. In 1958, Charles Keeling, a research chemist at Scripps Institute in California, began making such high-precision measurements of carbon dioxide and of the other greenhouse gases such as methane and nitrous oxide. These very high precision measurement techniques have been subsequently applied to measuring the gaseous composition of air bubbles trapped in glacial ice, thereby extending knowledge of atmospheric composition over time scales that go back more than 420,000 years, as in the Antarctic Vostoc ice-core data. These ice-core measurements show that the atmospheric concentration of carbon dioxide has increased from its pre-industrial level of 280 parts per million in 1850 to the present value near 390 ppm, and exhibits a steady rate of increase by about 2 ppm per year.

## **5.8 Future prospects**

The problem that we face today, has been described by James Hansen in his book *Storms of my Grandchildren*. The problem is that carbon dioxide is not only the principal thermostat that governs the global temperature of Earth, carbon is also the principal fuel that powers the human civilization. It is expected that at the current rate of fossil fuel use, the critical level of 450 ppm for carbon dioxide will be reached in about thirty years. Clearly, the thermal inertia of the ocean and the polar ice caps is very large, and it would take many centuries for the polar ice to melt. But upon reaching 450 ppm for atmospheric carbon dioxide, the terrestrial climate will be set on a collision course where the Earth will no longer be able to support polar ice caps. And the obvious implication of this is the eventual rise in sea level by some 70 meters, which would inundate most of the land areas where civilization resides.

Fortunately, because of the very large heat capacity of the ocean, there is still some time for humans to react and bring the global warming problem under control. While the radiative forcing attributable to the carbon dioxide that is being added to the atmosphere is felt instantaneously, the surface temperature change associated with this increase in radiative forcing materializes more slowly. Thus, about 40% of the global equilibrium temperature response materializes in about 5 years as the ocean mixed layer warms in response to the applied forcing. As the heat begins to diffuse downward into the deep ocean, about 60% of the total temperature response is reached in about 100 years, with approach to the 100% level taking several thousand years. It should also be noted that the

larger uncertainties abound in modeling the space-time response of the climate system because horizontal energy transports tend to be several orders of magnitude larger than corresponding greenhouse gas radiative forcings that are the ultimate instigators of global

While there remain legitimate uncertainties regarding the details of time specific, local, and regional global climate change, these uncertainties should not be projected onto the solid evidence that human industrial activity is busy driving Earth's climate into uncharted territory. The atmospheric carbon dioxide control knob is currently being turned at a much

In view of the pivotal role that carbon dioxide plays in governing the global temperature of Earth, a great deal of attention has being focused on obtaining accurate measurements of atmospheric carbon dioxide, and how carbon dioxide has been changing with time. In 1958, Charles Keeling, a research chemist at Scripps Institute in California, began making such high-precision measurements of carbon dioxide and of the other greenhouse gases such as methane and nitrous oxide. These very high precision measurement techniques have been subsequently applied to measuring the gaseous composition of air bubbles trapped in glacial ice, thereby extending knowledge of atmospheric composition over time scales that go back more than 420,000 years, as in the Antarctic Vostoc ice-core data. These ice-core measurements show that the atmospheric concentration of carbon dioxide has increased from its pre-industrial level of 280 parts per million in 1850 to the present value near 390

The problem that we face today, has been described by James Hansen in his book *Storms of my Grandchildren*. The problem is that carbon dioxide is not only the principal thermostat that governs the global temperature of Earth, carbon is also the principal fuel that powers the human civilization. It is expected that at the current rate of fossil fuel use, the critical level of 450 ppm for carbon dioxide will be reached in about thirty years. Clearly, the thermal inertia of the ocean and the polar ice caps is very large, and it would take many centuries for the polar ice to melt. But upon reaching 450 ppm for atmospheric carbon dioxide, the terrestrial climate will be set on a collision course where the Earth will no longer be able to support polar ice caps. And the obvious implication of this is the eventual rise in sea level by some 70 meters, which would inundate most of the land areas

Fortunately, because of the very large heat capacity of the ocean, there is still some time for humans to react and bring the global warming problem under control. While the radiative forcing attributable to the carbon dioxide that is being added to the atmosphere is felt instantaneously, the surface temperature change associated with this increase in radiative forcing materializes more slowly. Thus, about 40% of the global equilibrium temperature response materializes in about 5 years as the ocean mixed layer warms in response to the applied forcing. As the heat begins to diffuse downward into the deep ocean, about 60% of the total temperature response is reached in about 100 years, with approach to the 100% level taking several thousand years. It should also be noted that the

ppm, and exhibits a steady rate of increase by about 2 ppm per year.

climate change.

faster rate than ever before.

**5.8 Future prospects** 

where civilization resides.

enhancement of the greenhouse effect by water vapor and cloud feedbacks (the fast feedback processes) is not directly in response to the amount of carbon dioxide in the atmosphere per se, but rather to the change in temperature that is generated by the carbon dioxide increase.

There is also substantial natural variability within the climate system that produces global temperature fluctuations by several tenths of a degree on inter-annual and decadal time scales, including climate system response to the 11-year sunspot cycle. However, all of these temperature fluctuations are relative to a zero reference point, while the steady increase in the strength of the terrestrial greenhouse effect continues unabated. These fluctuations produce random looking changes in global temperature that tend to mask and obscure the steadily increasing global warming signal of carbon dioxide. Perhaps of more immediate concern are the likely prospects for an increase in the weather extremes such as more severe droughts and more severe floods, all fueled by the greater energy that is being added to the hydrological cycle as the global temperature rises and atmospheric water vapor increases. The more menacing future concern is the inevitable rise in sea level.

There is no mistaking the fact that the level of atmospheric carbon dioxide is increasing, and this continued increase causes the strength of the terrestrial greenhouse effect to keep on increasing. Accordingly, there is no viable alternative to counteract the impending effects of global warming except through direct human action to reduce the level of atmospheric carbon dioxide, to control the strength of the terrestrial greenhouse effect which ultimately determines the surface temperature of Earth and the global climate that results therefrom.

#### **5.9 Public perception**

Assessing the real cause and the perceived reality of global climate change has become a heated topic in the public discourse, particularly in the United States. Global warming has been denounced as the "greatest hoax ever" by prominent Senators and Congressmen. There are well organized efforts funded by fossil fuel interests to discredit climate science and to promote discord and mistrust in the public mind. The presumed rationale for this anti-climate science crusade by fossil fuel interests is perhaps similar to the reaction of the tobacco industry in response to the linking of deleterious health impacts to smoking. In their thinking, if global warming is determined to be harmful to human interests, the pressure to limit carbon dioxide emissions will be harmful to fossil fuel financial interests. On the other hand, there are environmental groups who already perceive global warming as a brewing environmental disaster that is already happening. They are demanding immediate action to curtail the use of fossil fuels. This brawling has attracted the attention of a broad spectrum of individuals from many different disciplines weighing in with opinions and suggestions as to how best understand what is really happening with global climate.

The terrestrial climate system however is very complex, and is not really amenable to simple analyses and interpretations. So, for example, performing a 'more sophisticated' statistical analysis of the global temperature record will not yield quantitative attribution as the cause of global warming because the existing temperature record is much too short, and there are too many causative factors that are not adequately accounted for that impact the global

Greenhouse Effect 293

The bottom line is that atmospheric carbon dioxide has been a thermostat in regulating the temperature of Earth throughout the entire geological history of Earth. The presentday exceedingly rapid increase in atmospheric carbon dioxide due to human industrial activity is unprecedented in the geological record. It is setting the course for continued inevitable global warming that will eventually melt the polar ice caps to produce a catastrophic sea level rise if left unchecked. The large heat capacity of the ocean provides ample time for humans to forestall a climate disaster. Since humans have been responsible for changing the level of atmospheric carbon dioxide, they are then also in control over the thermostat that controls the global temperature of the Earth. Humans are at a difficult crossroad. Carbon dioxide is the lifeblood of civilization as we know it. It is also the direct cause fueling an impending climate disaster. There is no other viable alternative that will counteract global warming except direct human effort to reduce the atmospheric carbon

Arrhenius, S. (1896). On the influence of carbonic acid in the air upon the temperature of the

Fourier, J. (1824). Remarques générales sur les températures du globe terrestre et des spaces

Hansen, J., et al., (2007). Dangerous human-made interference with climate: A GISS modelE study. *Atmos. Chem. Phys.*, 7, 2287-2312, doi:10.5194/acp-7-2287-2007. Hansen et al., J.E. (2008). Target atmospheric CO2: Where should humanity aim? *Open* 

Kirschvink, J. L. (1992). In *The Proterozoic Biosphere: A Multidisciplinary Study*. D. Des Maris,

Lacis et al., A. A. (2010). Atmospheric CO2: Principal control knob governing Earth's

Lacis, A.A., and M.I. Mishchenko, (1995). Climate forcing, climate sensitivity, and climate

*1994, April 24-29*. R.J. Charlson and J. Heintzenberg, Eds. John Wiley Sons. Lacis, A.A. and V. Oinas (1991) . A description of the correlated k distributed method for

in vertically inhomogeneous atmospheres. *J. Geophys. Res.*, 96: 9027-9063. Manabe S. and F. Moller (1961). On the radiative equilibrium and heat balance of the

Manabe S. and R.F. Strickler (1964). Thermal equilibrium of the atmosphere with a

Manabe S. and R.T. Wetherals (1967). Thermal equilibrium of the atmosphere with a given

response: A radiative modeling perspective on atmospheric aerosols. In *Aerosol Forcing of Climate: Report of the Dahlem Workshop on Aerosol Forcing of Climate, Berlin* 

modeling nongray gaseous absorption, thermal emission, and multiple scattering

Berner, R.A. (2004). *The Phanerozoic Carbon Cycle: CO2 and O2*. Oxford U Press, New York.

Fleming, J.R. (1998). *Historical Perspectives on Climate Change*, Oxford University Press.

planétaires, *Annales de Chemie et de Physique*, 27: 136-167.

Hansen, J.E. (2009). *Storms of my Grandchildren*, Bloomsbury USA, New York.

temperature, *Science*, 300: 356-359, doi:10.1126/science.1190653.

ed., Cambridge U Press, Cambridge, U.K. 51.

atmosphere, *Mon. Wea. Rev.* 89: 503-532.

convective adjustment, *J. Atmos. Sci.* 21: 361-376.

distribution of relative humidity*, J. Atmos. Sci*. 24: 241-259.

ground, *Philosophical Mag*., 41: 237-276.

Christianson, G.E. (1999). *Greenhouse*, Walker & Company.

*Atmos. Sci. J.*, 2: 217-231.

dioxide level.

**7. References** 

temperature. Also, sufficiently precise measurements that would definitively describe what is happening in the climate system are not available. And, there are aspects of the climate system, particularly interactions of the atmosphere with ocean dynamics that give rise to the so-called natural variability that is still not fully understood. All of this provides plenty of room for rampant skepticism to flourish.

There are, nevertheless, some important aspects of the ongoing global climate change that are understandable in fairly simple basic physics terms. Global warming, the steadily increasing component of global climate change, is specifically linked to the growing strength of the greenhouse effect. Thus, it is the greenhouse effect that is at the very center of the global warming controversy, and it is also the component that is most readily understandable in basic physics terms.

## **6. Conclusion**

There is a close analogy to be drawn between the way an ordinary thermostat maintains the temperature of a house, and the way that atmospheric carbon dioxide (and the other minor non-condensing greenhouse gases) control the global temperature of Earth. Atmospheric carbon dioxide performs a role that is similar to that of the house thermostat in setting the equilibrium temperature of the Earth. It differs from the house thermostat in that carbon dioxide itself is a potent greenhouse gas, warming the ground surface by means of the greenhouse effect. It is this sustained warming that enables water vapor and clouds to maintain their atmospheric distributions as the so-called feedback effects that amplify the initial warming provided by the non-condensing greenhouse gases. It is also important to keep in mind that the house thermostat merely turns the furnace on and off, and that the heat capacity of the house requires minutes before the house temperature responds to the thermostat setting. The climate system similarly has a delayed response to forcing.

Within only the past century, the carbon dioxide control knob has been turned sharply upward toward a much hotter global climate. The pre-industrial level of atmospheric carbon dioxide was about 280 ppm, which is representative of the inter-glacial maximum level of atmospheric carbon dioxide. During ice age extremes, the level of carbon dioxide drops to near 180 ppm, for which the global temperature is about 5 C colder. The rapid recent increase of atmospheric carbon dioxide has been attributed to human industrial activity, primarily due to the burning of fossil fuels has pushed the carbon dioxide level to near the 390 ppm, far beyond the inter-glacial maximum. The climate system is trying to respond to the new setting of the global thermostat, and this response has been the rise in global surface temperature by about 0.2 C per decade for the past three decades.

It has been suggested that we are well past the 300 to 350 ppm target level for atmospheric carbon dioxide beyond which dangerous anthropogenic interference in the climate system would be expected to exceed the 25% risk tolerance for impending degradation of land and ocean ecosystems, sea level rise, and inevitable disruption of the socio-economic and foodproducing infrastructure (Hansen et al., 2008). This prospect of a rising risk of triggering unacceptable environmental consequences makes reduction and control of atmospheric carbon dioxide a serious and pressing issue for humanity, one that is worthy of real-time attention.

temperature. Also, sufficiently precise measurements that would definitively describe what is happening in the climate system are not available. And, there are aspects of the climate system, particularly interactions of the atmosphere with ocean dynamics that give rise to the so-called natural variability that is still not fully understood. All of this provides plenty of

There are, nevertheless, some important aspects of the ongoing global climate change that are understandable in fairly simple basic physics terms. Global warming, the steadily increasing component of global climate change, is specifically linked to the growing strength of the greenhouse effect. Thus, it is the greenhouse effect that is at the very center of the global warming controversy, and it is also the component that is most readily

There is a close analogy to be drawn between the way an ordinary thermostat maintains the temperature of a house, and the way that atmospheric carbon dioxide (and the other minor non-condensing greenhouse gases) control the global temperature of Earth. Atmospheric carbon dioxide performs a role that is similar to that of the house thermostat in setting the equilibrium temperature of the Earth. It differs from the house thermostat in that carbon dioxide itself is a potent greenhouse gas, warming the ground surface by means of the greenhouse effect. It is this sustained warming that enables water vapor and clouds to maintain their atmospheric distributions as the so-called feedback effects that amplify the initial warming provided by the non-condensing greenhouse gases. It is also important to keep in mind that the house thermostat merely turns the furnace on and off, and that the heat capacity of the house requires minutes before the house temperature responds to the

thermostat setting. The climate system similarly has a delayed response to forcing.

temperature by about 0.2 C per decade for the past three decades.

Within only the past century, the carbon dioxide control knob has been turned sharply upward toward a much hotter global climate. The pre-industrial level of atmospheric carbon dioxide was about 280 ppm, which is representative of the inter-glacial maximum level of atmospheric carbon dioxide. During ice age extremes, the level of carbon dioxide drops to near 180 ppm, for which the global temperature is about 5 C colder. The rapid recent increase of atmospheric carbon dioxide has been attributed to human industrial activity, primarily due to the burning of fossil fuels has pushed the carbon dioxide level to near the 390 ppm, far beyond the inter-glacial maximum. The climate system is trying to respond to the new setting of the global thermostat, and this response has been the rise in global surface

It has been suggested that we are well past the 300 to 350 ppm target level for atmospheric carbon dioxide beyond which dangerous anthropogenic interference in the climate system would be expected to exceed the 25% risk tolerance for impending degradation of land and ocean ecosystems, sea level rise, and inevitable disruption of the socio-economic and foodproducing infrastructure (Hansen et al., 2008). This prospect of a rising risk of triggering unacceptable environmental consequences makes reduction and control of atmospheric carbon dioxide a serious and pressing issue for humanity, one that is worthy of real-time

room for rampant skepticism to flourish.

understandable in basic physics terms.

**6. Conclusion** 

attention.

The bottom line is that atmospheric carbon dioxide has been a thermostat in regulating the temperature of Earth throughout the entire geological history of Earth. The presentday exceedingly rapid increase in atmospheric carbon dioxide due to human industrial activity is unprecedented in the geological record. It is setting the course for continued inevitable global warming that will eventually melt the polar ice caps to produce a catastrophic sea level rise if left unchecked. The large heat capacity of the ocean provides ample time for humans to forestall a climate disaster. Since humans have been responsible for changing the level of atmospheric carbon dioxide, they are then also in control over the thermostat that controls the global temperature of the Earth. Humans are at a difficult crossroad. Carbon dioxide is the lifeblood of civilization as we know it. It is also the direct cause fueling an impending climate disaster. There is no other viable alternative that will counteract global warming except direct human effort to reduce the atmospheric carbon dioxide level.

#### **7. References**


**14**

*Germany* 

**Regional-Scale Assessment of the Climatic Role** 

Several natural and anthropogenic processes influence the climate of the Earth. Human affect climate through increasing greenhouse gas concentrations, changing aerosol compositions as well as by land surface changes (IPCC, 2007). There are several recent EUprojects carried out in the last decade, to provide high-resolution climate change projections with focus on future climate changes and their impacts in Europe (Christensen et al., 2007; Jacob et al., 2008; van der Linden & Mitchell, 2009). These studies are based on the results of regional climate model simulations driven by different predefined greenhouse gas emission scenarios. The difference between the simulated climatic conditions for the future and for the present time period is the climate change signal. For the 21st century, projected climate change signals for temperature and precipitation show seasonal and spatial differences in

Europe and also vary depending on the applied greenhouse gas emission scenario.

For the period 2021-2050 all regional climate models predict a quite robust (i.e. above the noise generated by the internal model variability and consistent across multiple climate models – Hagemann et al., 2009) surface warming in central and eastern Europe. The annual precipitation shows an increase in the northeastern and a decrease in the southwestern regions. The transition (neutral) zone (e.g. Hungary, Rumania) can be characterized by the largest spread between models (Christensen & Christensen, 2007; Kjellström et al., 2011),

At the end of the 21st century, a warming is expected in all seasons over Europe, which is stronger than in the first half of the 21st century. All models agree that the largest warming for summer is projected to occur in the Mediterranean area, southern France and over the Iberian Peninsula. Less warming is projected over the Scandinavian regions. For winter the maximum warming occurs in eastern Europe (Giorgi et al., 2004; Christensen & Christensen, 2007). For precipitation, the largest increase is projected in winter, whereas the decrease is the strongest in summer. Changes in the intermediate seasons (spring and autumn) are less pronounced. Results of the regional model simulations show a north-south gradient of annual precipitation changes over Europe, with positive changes in the north (especially in winter) and negative changes in the south (especially over the Mediterranean area in summer). The line of zero change moves with the seasons (Kjellström et al., 2011). For

**1. Introduction** 

**1.1 Regional climate projections for Europe** 

where precipitation changes are quite small.

**of Forests Under Future Climate Conditions** 

*2Climate Service Center–eine Einrichtung am Helmholtz-Zentrum Geesthacht* 

Borbála Gálos1 and Daniela Jacob1,2 *1Max Planck Institute for Meteorology, Hamburg* 


## **Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions**

Borbála Gálos1 and Daniela Jacob1,2 *1Max Planck Institute for Meteorology, Hamburg 2Climate Service Center–eine Einrichtung am Helmholtz-Zentrum Geesthacht Germany* 

## **1. Introduction**

294 Greenhouse Gases – Emission, Measurement and Management

McKay, C.P., J.B. Pollack and R. Courtin (1991). The greenhouse and anti-greenhouse effects

Rothman, L.S. et al., (2009). The HITRAN 2008 molecular spectroscopic database. *J. Quant.* 

Schmidt et al., G.A. (2010). Attribution of the present-day total greenhouse effect, *J. Geophys.* 

Solomon, S. et al. (2007). *Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change*, Cambridge University Press.

on Titan, Science, 253: 118-1121.

*Spectrosc. Radiat. Transfer*, 110: 533572.

*Res*., 115: D20106, doi:10.1029/2010JD014287.

#### **1.1 Regional climate projections for Europe**

Several natural and anthropogenic processes influence the climate of the Earth. Human affect climate through increasing greenhouse gas concentrations, changing aerosol compositions as well as by land surface changes (IPCC, 2007). There are several recent EUprojects carried out in the last decade, to provide high-resolution climate change projections with focus on future climate changes and their impacts in Europe (Christensen et al., 2007; Jacob et al., 2008; van der Linden & Mitchell, 2009). These studies are based on the results of regional climate model simulations driven by different predefined greenhouse gas emission scenarios. The difference between the simulated climatic conditions for the future and for the present time period is the climate change signal. For the 21st century, projected climate change signals for temperature and precipitation show seasonal and spatial differences in Europe and also vary depending on the applied greenhouse gas emission scenario.

For the period 2021-2050 all regional climate models predict a quite robust (i.e. above the noise generated by the internal model variability and consistent across multiple climate models – Hagemann et al., 2009) surface warming in central and eastern Europe. The annual precipitation shows an increase in the northeastern and a decrease in the southwestern regions. The transition (neutral) zone (e.g. Hungary, Rumania) can be characterized by the largest spread between models (Christensen & Christensen, 2007; Kjellström et al., 2011), where precipitation changes are quite small.

At the end of the 21st century, a warming is expected in all seasons over Europe, which is stronger than in the first half of the 21st century. All models agree that the largest warming for summer is projected to occur in the Mediterranean area, southern France and over the Iberian Peninsula. Less warming is projected over the Scandinavian regions. For winter the maximum warming occurs in eastern Europe (Giorgi et al., 2004; Christensen & Christensen, 2007). For precipitation, the largest increase is projected in winter, whereas the decrease is the strongest in summer. Changes in the intermediate seasons (spring and autumn) are less pronounced. Results of the regional model simulations show a north-south gradient of annual precipitation changes over Europe, with positive changes in the north (especially in winter) and negative changes in the south (especially over the Mediterranean area in summer). The line of zero change moves with the seasons (Kjellström et al., 2011). For

Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions 297

Consequently, the change of vegetation from tundra to taiga under future climate conditions amplifies the global warming. Tropical forests maintain high rates of evapotranspiration. In this region, surface warming arising from the low albedo of forests is offset by the strong

In temperate forests the albedo and evaporative forcings are moderate compared to boreal and tropical forests (Bala et al., 2007; Bonan, 2008; Jackson et al., 2008). Climate model studies for the temperate regions showed that replacing forests with agriculture or grasslands reduces the surface air temperatures (Bonan, 1997; Bounoua et al., 2002; Oleson et al., 2004) and the number of summer hot days (Anav et al., 2010). Other studies show opposite results, where temperate forests cool the air compared to grasslands and croplands and contribute to higher precipitation rates in the growing season (Copeland et al., 1996; Hogg et al., 2000; Sánchez et al., 2007). In the Mediterranean region climatic effects of forest cover change can also vary during the summer months (Heck et al., 2001). In the period from April until mid-July potential vegetation cover conditions led to cooler and moister conditions due to the increase of evapotranspiration. In mid-July soil moisture dropped below the critical value and transpiration was almost completely inhibited. It resulted in dryer and warmer summer accelerating the projected climate change. Teuling et al. (2010) pointed out that the role of the forests in the surface energy and water budget is depending on the selected time scale: in the short term, forests contribute to the increase of temperature, but on longer time scales they can reduce the impact of extreme heat weaves. These studies indicate that forests can enhance or dampen the climate change signal depending on various contrasting vegetation feedbacks, which can diminish or counteract each other. Furthermore the variability of the climatic, soil and vegetation characteristics as well as the description of the land surface processes in the applied climate model also have

The climatic feedbacks of land cover changes due to climate change and the regional land use politics as well as the role of the forests in the climate change mitigation on country scale are still unknown. Fine scale studies are essential not only for the assessment of the climate protecting effects of forests, but also for the development of adaptation strategies in forestry,

In order to address this topic, our sensitivity study is focusing on the climatic effects of afforestation in Europe under future climate conditions based on the following research

In which regions does the increase of forest cover enhance/reduce the projected climate

How big are the effects of forest cover change on the summer precipitation and

Which are the regions, where afforestation is the most beneficial from a climatic point of

On country scale, a more detailed case study has been carried out for Hungary. For the end of the 21st century, regional climate model simulations project a significant increase of summer temperature and a decrease of summer precipitation (Bartholy et al., 2007; Gálos et al., 2007;

evaporative cooling that reduces global temperature increase (Bonan, 2008).

an influence on the simulated vegetation-atmosphere interactions.

agriculture and water management for the next decades.

temperature relative to the climate change signal?

**1.3 Research foci** 

questions:

change?

view?

southern and central Europe the spatial distribution of the projected temperature and precipitation changes in summer refer to a marked shift towards a warmer and drier climate (Vidale et al., 2007; Beniston, 2009). Recent results from enhanced greenhouse gas emission scenario simulation over Europe suggest that not only the climatic means are changing, but there is also an increase in the inter-annual variability of the future temperature and precipitation values, which leads to higher probability of extremes compared to the presentday conditions (Schär et al., 2004; Giorgi et al., 2004; Seneviratne et al., 2006; Beniston et al., 2007; Kjellström et al., 2007; Vidale et al., 2007; Fischer & Schär 2010).

The most commonly used IPCC-SRES (Intergovernmental Panel on Climate Change – Special Report on Emission Scenarios) emission scenarios for these studies are the B1, A1B, A2. In the first half of the 21st century, the climatic impact of these greenhouse gas emission scenarios are quite similar (van der Linden & Mitchell, 2009). But the temperature difference increases by mid-century with greatest warming in A2 and least warming in B1 (IPCC, 2007).

## **1.2 Climatic effects of land use and land cover change**

Temperature and precipitation play an important role in determining the distribution of the terrestrial ecosystems that in turn interact with the atmosphere through biogeophysical and biogeochemical processes. Vegetation affects the physical characteristics of the land surface, which control the surface energy fluxes and hydrological cycle (biogeophysical feedbacks; Pielke et al., 1998; Brovkin, 2002; Pitman, 2003; Betts, 2007; Anderson et al., 2010). Through biogeochemical effects, ecosystems alter the biogeochemical cycles, thereby change the chemical composition of the atmosphere (Betts, 2001; Bonan, 2002; Pitman, 2003; Bonan, 2004; Feddema et al., 2005). Forests have larger leaf areas and aerodynamic roughness lengths, lower albedo and deeper roots compared to other vegetated surfaces. They sequester carbon thereby alter the carbon storage of land.

Depending on the region, biogeophysical and biogeochemical feedbacks of land cover on climate can amplify or dampen each other (Arora & Montenegro, 2011). Through these landatmosphere interactions changes of the land cover and land use due to natural and human influence alter climate and hence can lead to the enhancement or reduction of the projected climate change signals expected from increased atmospheric CO2 concentration (Feddema et al., 2005; Bonan, 2008). Long term studies show that land use and land cover changes have much weaker influence on the atmospheric circulation compared to greenhouse-gas forcings (Betts, 2007; Göttel et al., 2008; Wramneby et al., 2010; Arora & Montenegro, 2011). However, in smaller areas and in regions with strong land-atmosphere interactions the feedback processes can significantly affect and modify the weather and climate, the temperature and precipitation variability (Seneviratne et al., 2006; Seneviratne et al., 2010).

This section focuses on studies of the biogeophysical processes, which represent the contrasting climatic effects of forest cover changes on different regions, seasons and time scales. Changes of vegetation cover under future climate conditions enhance the warming trend in Scandinavian Mountains as well as the drying trend in southern Europe, but mitigate the projected increase of temperature in central Europe (Wramneby et al., 2010). Boreal forests have the greatest biogeophysical effect of all biomes on annual mean global temperature, which is larger than their effect on the carbon cycle (Bonan, 2008). If snow is present, the darker coniferous forest masks the snow cover. It is resulting in lower surface albedo compared to tundra vegetation or bare ground, which leads to higher winter and spring air temperatures (Bonan et al., 1992; Brovkin, 2002; Kleidon et al., 2007; Göttel et al., 2008). Consequently, the change of vegetation from tundra to taiga under future climate conditions amplifies the global warming. Tropical forests maintain high rates of evapotranspiration. In this region, surface warming arising from the low albedo of forests is offset by the strong evaporative cooling that reduces global temperature increase (Bonan, 2008).

In temperate forests the albedo and evaporative forcings are moderate compared to boreal and tropical forests (Bala et al., 2007; Bonan, 2008; Jackson et al., 2008). Climate model studies for the temperate regions showed that replacing forests with agriculture or grasslands reduces the surface air temperatures (Bonan, 1997; Bounoua et al., 2002; Oleson et al., 2004) and the number of summer hot days (Anav et al., 2010). Other studies show opposite results, where temperate forests cool the air compared to grasslands and croplands and contribute to higher precipitation rates in the growing season (Copeland et al., 1996; Hogg et al., 2000; Sánchez et al., 2007). In the Mediterranean region climatic effects of forest cover change can also vary during the summer months (Heck et al., 2001). In the period from April until mid-July potential vegetation cover conditions led to cooler and moister conditions due to the increase of evapotranspiration. In mid-July soil moisture dropped below the critical value and transpiration was almost completely inhibited. It resulted in dryer and warmer summer accelerating the projected climate change. Teuling et al. (2010) pointed out that the role of the forests in the surface energy and water budget is depending on the selected time scale: in the short term, forests contribute to the increase of temperature, but on longer time scales they can reduce the impact of extreme heat weaves.

These studies indicate that forests can enhance or dampen the climate change signal depending on various contrasting vegetation feedbacks, which can diminish or counteract each other. Furthermore the variability of the climatic, soil and vegetation characteristics as well as the description of the land surface processes in the applied climate model also have an influence on the simulated vegetation-atmosphere interactions.

## **1.3 Research foci**

296 Greenhouse Gases – Emission, Measurement and Management

southern and central Europe the spatial distribution of the projected temperature and precipitation changes in summer refer to a marked shift towards a warmer and drier climate (Vidale et al., 2007; Beniston, 2009). Recent results from enhanced greenhouse gas emission scenario simulation over Europe suggest that not only the climatic means are changing, but there is also an increase in the inter-annual variability of the future temperature and precipitation values, which leads to higher probability of extremes compared to the presentday conditions (Schär et al., 2004; Giorgi et al., 2004; Seneviratne et al., 2006; Beniston et al.,

The most commonly used IPCC-SRES (Intergovernmental Panel on Climate Change – Special Report on Emission Scenarios) emission scenarios for these studies are the B1, A1B, A2. In the first half of the 21st century, the climatic impact of these greenhouse gas emission scenarios are quite similar (van der Linden & Mitchell, 2009). But the temperature difference increases by

Temperature and precipitation play an important role in determining the distribution of the terrestrial ecosystems that in turn interact with the atmosphere through biogeophysical and biogeochemical processes. Vegetation affects the physical characteristics of the land surface, which control the surface energy fluxes and hydrological cycle (biogeophysical feedbacks; Pielke et al., 1998; Brovkin, 2002; Pitman, 2003; Betts, 2007; Anderson et al., 2010). Through biogeochemical effects, ecosystems alter the biogeochemical cycles, thereby change the chemical composition of the atmosphere (Betts, 2001; Bonan, 2002; Pitman, 2003; Bonan, 2004; Feddema et al., 2005). Forests have larger leaf areas and aerodynamic roughness lengths, lower albedo and deeper roots compared to other vegetated surfaces. They

Depending on the region, biogeophysical and biogeochemical feedbacks of land cover on climate can amplify or dampen each other (Arora & Montenegro, 2011). Through these landatmosphere interactions changes of the land cover and land use due to natural and human influence alter climate and hence can lead to the enhancement or reduction of the projected climate change signals expected from increased atmospheric CO2 concentration (Feddema et al., 2005; Bonan, 2008). Long term studies show that land use and land cover changes have much weaker influence on the atmospheric circulation compared to greenhouse-gas forcings (Betts, 2007; Göttel et al., 2008; Wramneby et al., 2010; Arora & Montenegro, 2011). However, in smaller areas and in regions with strong land-atmosphere interactions the feedback processes can significantly affect and modify the weather and climate, the temperature and

This section focuses on studies of the biogeophysical processes, which represent the contrasting climatic effects of forest cover changes on different regions, seasons and time scales. Changes of vegetation cover under future climate conditions enhance the warming trend in Scandinavian Mountains as well as the drying trend in southern Europe, but mitigate the projected increase of temperature in central Europe (Wramneby et al., 2010). Boreal forests have the greatest biogeophysical effect of all biomes on annual mean global temperature, which is larger than their effect on the carbon cycle (Bonan, 2008). If snow is present, the darker coniferous forest masks the snow cover. It is resulting in lower surface albedo compared to tundra vegetation or bare ground, which leads to higher winter and spring air temperatures (Bonan et al., 1992; Brovkin, 2002; Kleidon et al., 2007; Göttel et al., 2008).

2007; Kjellström et al., 2007; Vidale et al., 2007; Fischer & Schär 2010).

**1.2 Climatic effects of land use and land cover change** 

sequester carbon thereby alter the carbon storage of land.

precipitation variability (Seneviratne et al., 2006; Seneviratne et al., 2010).

mid-century with greatest warming in A2 and least warming in B1 (IPCC, 2007).

The climatic feedbacks of land cover changes due to climate change and the regional land use politics as well as the role of the forests in the climate change mitigation on country scale are still unknown. Fine scale studies are essential not only for the assessment of the climate protecting effects of forests, but also for the development of adaptation strategies in forestry, agriculture and water management for the next decades.

In order to address this topic, our sensitivity study is focusing on the climatic effects of afforestation in Europe under future climate conditions based on the following research questions:


On country scale, a more detailed case study has been carried out for Hungary. For the end of the 21st century, regional climate model simulations project a significant increase of summer temperature and a decrease of summer precipitation (Bartholy et al., 2007; Gálos et al., 2007;

Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions 299

hydrostatic approximation. The physical parameterizations from the global climate model ECHAM4 are implemented at the Max Planck Institute for Meteorology in Hamburg

Land surface processes in REMO are controlled by physical vegetation properties. For each land cover type the parameter values of leaf area index and fractional vegetation cover for the growing and dormancy season, background albedo, surface roughness length due to vegetation, forest ratio, plant-available soil water holding capacity and volumetric wilting point are allocated in the global dataset of land surface parameters (Hagemann et al., 1999; Hagemann, 2002). In the current model version the vegetation phenology is represented by the monthly varying values of the leaf area index and vegetation ratio. The mean climatology of the annual cycle of background albedo is also implemented (Rechid & Jacob, 2006, Rechid et al., 2008a, 2008b). The other land surface parameters remain constant throughout the year. Land cover change in REMO can be implemented by modification of

The simulations have been carried out for Europe (figure 1), with 0.22° horizontal grid resolution. REMO was driven with lateral boundary conditions from the coupled atmosphere–ocean GCM ECHAM5/MPI-OM (Roeckner et al., 2006; Jungclaus et al., 2006).

[m]

Fig. 1. Simulation domain. The smaller domain applied for the Hungarian case study is

 *Reference simulation* for the past (1971-1990) with present (unchanged) forest cover. *Emission scenario simulation* for the future (2071-2090) with present (unchanged) forest cover applying the A2 IPCC-SRES emission scenario1 (Houghton et al., 2001; Nakicenovic et al., 2000). This experiment was the reference simulation to the land

1 A2: continuously increasing global population and regionally oriented economic growth that is more

The following experiments have been performed (table 1):

(Roeckner et al., 1996) in the regional model.

the characteristic land surface parameters.

**2.2 Experimental setup** 

marked with rectangle

cover change study.

fragmented and slower than in other storylines.

Jacob et al., 2008; Szépszó, 2008; Radvánszky & Jacob, 2009). From ecological point of view Hungary (in the southeastern part of central Europe) has been selected as study region because here, many of the zonal tree species have their lower limit of distribution (Mátyás et al., 2009; Mátyás, 2010; Mátyás et al., 2010), which are especially sensitive and vulnerable to the increase of the frequency of climatic extremes, primarily droughts. In these forests the more frequent and severe droughts at the end of the 20th century already resulted in growth decline, loss of vitality and the decrease of the macroclimatically suitable area of distribution (Berki et al., 2009). Under the projected climate conditions these species may disappear from this region (Berki et al., 2009; Mátyás et al., 2010; Czúcz et al., 2011).

In the last 50 years, large scale afforestation was carried out in Hungary, which is planned to continue also in the near future. The influence of the historical land cover change on weather and climate has been investigated by Drüszler et al. (2011). For the future, forests can also have an important role in the adaptation to climate change. Therefore this case study is concentrating on the possible mitigation of the strong warming and drying of summers projected for the second half of the 21st century. The assessment of the maximal climatic effects of afforestation is addressing the following research questions:


In order to answer the scientific questions the chapter is organized as follows: in section 2 the applied model and experimental set up and the main steps of the analyses are described. Results are presented in section 3: in 3.1 the sign of the climatic effect of emission change and potential afforestation has been studied for precipitation and temperature. In section 3.2 the most climate change affected regions as well as the areas characterized by the largest climatic effects of forest cover increase have been determined. In section 3.3 the magnitude of the climatic feedbacks of afforestation is analyzed relative to the magnitude of the climate change signal. Section 3.4 introduces the country scale effects of maximal afforestation in Hungary. Results are summarized, conclusions are drawn and the possibilities for the practical application are stressed in section 4.

## **2. Methods**

Regional climate models have the potential to provide detailed information about the future climate on fine horizontal resolution. For studying the climatic feedbacks of land cover change in Europe regional scale analyses are essential because of the differences in the climate sensitivity among regions and the large spatial variability of the land surface properties and the related processes.

#### **2.1 The regional climate model REMO – General characteristics and land surface parameterization**

In this study the REgional climate MOdel (REMO) has been applied for Europe, with horizontal resolution 0.22°. REMO (Jacob, 2001; Jacob et al., 2001; Jacob et al., 2007) is a regional three-dimensional numerical model of the atmosphere. It is based on the 'Europamodell', the former numerical weather prediction model of the German Weather Service, DWD (Majewski, 1991). The calculation of the prognostic variables is based on the hydrostatic approximation. The physical parameterizations from the global climate model ECHAM4 are implemented at the Max Planck Institute for Meteorology in Hamburg (Roeckner et al., 1996) in the regional model.

Land surface processes in REMO are controlled by physical vegetation properties. For each land cover type the parameter values of leaf area index and fractional vegetation cover for the growing and dormancy season, background albedo, surface roughness length due to vegetation, forest ratio, plant-available soil water holding capacity and volumetric wilting point are allocated in the global dataset of land surface parameters (Hagemann et al., 1999; Hagemann, 2002). In the current model version the vegetation phenology is represented by the monthly varying values of the leaf area index and vegetation ratio. The mean climatology of the annual cycle of background albedo is also implemented (Rechid & Jacob, 2006, Rechid et al., 2008a, 2008b). The other land surface parameters remain constant throughout the year. Land cover change in REMO can be implemented by modification of the characteristic land surface parameters.

## **2.2 Experimental setup**

298 Greenhouse Gases – Emission, Measurement and Management

Jacob et al., 2008; Szépszó, 2008; Radvánszky & Jacob, 2009). From ecological point of view Hungary (in the southeastern part of central Europe) has been selected as study region because here, many of the zonal tree species have their lower limit of distribution (Mátyás et al., 2009; Mátyás, 2010; Mátyás et al., 2010), which are especially sensitive and vulnerable to the increase of the frequency of climatic extremes, primarily droughts. In these forests the more frequent and severe droughts at the end of the 20th century already resulted in growth decline, loss of vitality and the decrease of the macroclimatically suitable area of distribution (Berki et al., 2009). Under the projected climate conditions these species may disappear from this region

In the last 50 years, large scale afforestation was carried out in Hungary, which is planned to continue also in the near future. The influence of the historical land cover change on weather and climate has been investigated by Drüszler et al. (2011). For the future, forests can also have an important role in the adaptation to climate change. Therefore this case study is concentrating on the possible mitigation of the strong warming and drying of summers projected for the second half of the 21st century. The assessment of the maximal

 Which regions are the most affected by climate change for the end of the 21st century? In which part of the country can forests play an important role in reducing the projected

In order to answer the scientific questions the chapter is organized as follows: in section 2 the applied model and experimental set up and the main steps of the analyses are described. Results are presented in section 3: in 3.1 the sign of the climatic effect of emission change and potential afforestation has been studied for precipitation and temperature. In section 3.2 the most climate change affected regions as well as the areas characterized by the largest climatic effects of forest cover increase have been determined. In section 3.3 the magnitude of the climatic feedbacks of afforestation is analyzed relative to the magnitude of the climate change signal. Section 3.4 introduces the country scale effects of maximal afforestation in Hungary. Results are summarized, conclusions are drawn and the possibilities for the

Regional climate models have the potential to provide detailed information about the future climate on fine horizontal resolution. For studying the climatic feedbacks of land cover change in Europe regional scale analyses are essential because of the differences in the climate sensitivity among regions and the large spatial variability of the land surface

**2.1 The regional climate model REMO – General characteristics and land surface** 

In this study the REgional climate MOdel (REMO) has been applied for Europe, with horizontal resolution 0.22°. REMO (Jacob, 2001; Jacob et al., 2001; Jacob et al., 2007) is a regional three-dimensional numerical model of the atmosphere. It is based on the 'Europamodell', the former numerical weather prediction model of the German Weather Service, DWD (Majewski, 1991). The calculation of the prognostic variables is based on the

climatic effects of afforestation is addressing the following research questions:

(Berki et al., 2009; Mátyás et al., 2010; Czúcz et al., 2011).

tendency of drying?

**2. Methods** 

**parameterization** 

practical application are stressed in section 4.

properties and the related processes.

The simulations have been carried out for Europe (figure 1), with 0.22° horizontal grid resolution. REMO was driven with lateral boundary conditions from the coupled atmosphere–ocean GCM ECHAM5/MPI-OM (Roeckner et al., 2006; Jungclaus et al., 2006).

Fig. 1. Simulation domain. The smaller domain applied for the Hungarian case study is marked with rectangle

The following experiments have been performed (table 1):


<sup>1</sup> A2: continuously increasing global population and regionally oriented economic growth that is more fragmented and slower than in other storylines.

Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions 301

*Climate change due to potential afforestation* has been calculated comparing the simulation results with- and without forest cover increase for the future time period (2071-2090).

*Climate changes due to emission changes and potential afforestation* has been determined comparing the results of the potential afforestation experiment (2071-2090) to the reference study in the past (1971-1990). The sign and the magnitude climatic effects of potential afforestation have been analyzed relative to the climate change signal. Regional differences in the climate change altering effect of afforestation have been determined and investigated

For more detailed analyses, a case study has been prepared for Hungary over a smaller simulation domain (figure 1), applying the same regional climate model and the same steps for data analyses. Climate change due to emission change has been calculated for 2071-2100 compared to the 30-year time period in the past (1961-1990). The A1B IPCC-SRES emission scenario2 (Houghton et al., 2001; Nakicenovic et al., 2000) has been applied, which represents the average estimation for the analyzed region. To get information about the maximum climatic effects of afforestation and its regional differences, the whole vegetated area of Hungary was assumed to be forest and the new afforested areas are all deciduous. Over the simulation domain (marked in figure 1), forest cover has been changed only in Hungary. The assumed maximal afforestation takes approximately 75 % increase of forest

cover in country mean additionally to the existing 20 % forested area (figure 3).

Fig. 3. Forest cover in Hungary in the reference simulation (a) and in the maximal afforestation experiment (b). Increase of forest cover in the maximal afforestation experiment compared to the reference (c). Adapted from: Gálos et al., 2011

**3.1 Sign of the climatic effects of emission change and potential afforestation** 

A Mann-Whitney–U-Test (Mann & Whitney, 1947) was applied to test the significance of the climatic effects of afforestation and emission change. This is a ranking test, which does not

The climate change signals without any land cover changes have been analyzed for summer temperature means and precipitation sums in the time period 2071-2090 compared to 1971- 1990. Based on the simulation results a positive temperature signal is expected in whole Europe (figure 4a). Increase of temperature is projected to occur with precipitation decrease

2 A1: very rapid economic growth, global population peaks in mid-century and declines thereafter, and rapid introduction of new and more efficient technologies. The A1 scenario family develops into three groups that describe alternative directions of technological change in the energy system; A1B means a

for two selected regions.

assume a normal distribution.

balance across all sources.

**3. Results** 

 *Potential afforestation* experiment for 2071-2090. The forest cover increase (figure 2) is based on the net primary production map for Europe derived from remotely sensed MODIS (Moderate-Resolution Imaging Spectroradiometer) products, precipitation and temperature conditions from the Wordclim database and soil conditions from the International Institute for Applied Systems Analysis (Kindermann, pers. comm.). The new afforested areas were assumed to be deciduous.


a based on Kindermann (pers. comm.)

<sup>b</sup> Roeckner et al., 2006; Jungclaus et al., 2006

Table 1. Analyzed data and time periods

Fig. 2. Increase of the forest cover in the potential afforestation simulation compared to the present forested area in the model

## **2.3 Main steps of the analyses**

These analyses are focusing on the biogeophysical feedbacks of afforestation on the climate. Corresponding to the forest cover increase in all grid boxes the new distribution of the land cover categories has been determined and a new land surface parameter set has been calculated. Simulation results have been analyzed for May, June, July and August. In this study the mean of this period is considered '*summer*' (MJJA), because in these months water availability is especially important for the vegetation growth. The leaf area index of the deciduous forests reaches its maximum, which has a strong control on the land-atmosphere interactions.

*Climate change due to emission change* has been investigated analyzing the summer precipitation sums and 2m-temperature means for 2071-2090 (without any land cover changes) compared to 1971-1990.

*Climate change due to potential afforestation* has been calculated comparing the simulation results with- and without forest cover increase for the future time period (2071-2090).

*Climate changes due to emission changes and potential afforestation* has been determined comparing the results of the potential afforestation experiment (2071-2090) to the reference study in the past (1971-1990). The sign and the magnitude climatic effects of potential afforestation have been analyzed relative to the climate change signal. Regional differences in the climate change altering effect of afforestation have been determined and investigated for two selected regions.

For more detailed analyses, a case study has been prepared for Hungary over a smaller simulation domain (figure 1), applying the same regional climate model and the same steps for data analyses. Climate change due to emission change has been calculated for 2071-2100 compared to the 30-year time period in the past (1961-1990). The A1B IPCC-SRES emission scenario2 (Houghton et al., 2001; Nakicenovic et al., 2000) has been applied, which represents the average estimation for the analyzed region. To get information about the maximum climatic effects of afforestation and its regional differences, the whole vegetated area of Hungary was assumed to be forest and the new afforested areas are all deciduous. Over the simulation domain (marked in figure 1), forest cover has been changed only in Hungary. The assumed maximal afforestation takes approximately 75 % increase of forest cover in country mean additionally to the existing 20 % forested area (figure 3).

Fig. 3. Forest cover in Hungary in the reference simulation (a) and in the maximal afforestation experiment (b). Increase of forest cover in the maximal afforestation experiment compared to the reference (c). Adapted from: Gálos et al., 2011

A Mann-Whitney–U-Test (Mann & Whitney, 1947) was applied to test the significance of the climatic effects of afforestation and emission change. This is a ranking test, which does not assume a normal distribution.

## **3. Results**

300 Greenhouse Gases – Emission, Measurement and Management

 *Potential afforestation* experiment for 2071-2090. The forest cover increase (figure 2) is based on the net primary production map for Europe derived from remotely sensed MODIS (Moderate-Resolution Imaging Spectroradiometer) products, precipitation and temperature conditions from the Wordclim database and soil conditions from the International Institute for Applied Systems Analysis (Kindermann, pers. comm.). The

Experiment *Reference simulation Potential afforestation simulation*  **Characteristics** Present forest cover Afforestation over all vegetated area a

Fig. 2. Increase of the forest cover in the potential afforestation simulation compared to the

These analyses are focusing on the biogeophysical feedbacks of afforestation on the climate. Corresponding to the forest cover increase in all grid boxes the new distribution of the land cover categories has been determined and a new land surface parameter set has been calculated. Simulation results have been analyzed for May, June, July and August. In this study the mean of this period is considered '*summer*' (MJJA), because in these months water availability is especially important for the vegetation growth. The leaf area index of the deciduous forests reaches its maximum, which has a strong control on the land-atmosphere

*Climate change due to emission change* has been investigated analyzing the summer precipitation sums and 2m-temperature means for 2071-2090 (without any land cover

2071-2090 2071-2090

[%]

new afforested areas were assumed to be deciduous.

**Horizontal resolution** 0.22°

**Greenhouse gas forcing** IPCC-SRES emission scenario A2

**Lateral boundaries** ECHAM5/MPI-OMb

**Time period** 1971-1990

a based on Kindermann (pers. comm.) <sup>b</sup> Roeckner et al., 2006; Jungclaus et al., 2006 Table 1. Analyzed data and time periods

present forested area in the model

**2.3 Main steps of the analyses** 

changes) compared to 1971-1990.

interactions.

## **3.1 Sign of the climatic effects of emission change and potential afforestation**

The climate change signals without any land cover changes have been analyzed for summer temperature means and precipitation sums in the time period 2071-2090 compared to 1971- 1990. Based on the simulation results a positive temperature signal is expected in whole Europe (figure 4a). Increase of temperature is projected to occur with precipitation decrease

<sup>2</sup> A1: very rapid economic growth, global population peaks in mid-century and declines thereafter, and rapid introduction of new and more efficient technologies. The A1 scenario family develops into three groups that describe alternative directions of technological change in the energy system; A1B means a balance across all sources.

Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions 303

From the figure 4 it can be concluded that there are regions, where temperature and/or precipitation changes due to emission change and afforestation show the same sign so that they can enhance each other. In other areas they have the opposite sign, thus depending on their magnitude, afforestation can reduce or fully compensate the effects of the emission

**3.2 Magnitude of the climatic effect of emission change and potential afforestation**  First, those regions have been determined, which are the most affected by climate change for the end of the 21st century. Without any forest cover changes the strongest warming and drying are projected to the Mediterranean area, southern France and over the Iberian Peninsula (figure 5). These signals are in a good agreement with the simulation results of

Fig. 5. The regions the most affected by climate change due to emission change

Fig. 6. The regions with the largest effects of forest cover increase on temperature and

Second, regions, where forest cover increase (without any emission change) has the largest effects on temperature and precipitation have been also identified. For the northern part of central and western Europe temperature decrease due to afforestation can exceed 0.3 °C additionally to more than 10 % increase of the summer precipitation sum (figure 6). The

precipitation (dT: temperature change, dP: precipitation change)

other regional climate models for the same region.

(dT: temperature change, dP: precipitation change)

change.

in southern and central Europe and in the southern part of Scandinavia, whereas Northeast Europe can be characterized with warmer and wetter conditions (figure 4a).

Figure 4b represents the direction of the temperature and precipitation changes due to afforestation only. Comparing the simulation results with- and without forest cover changes for the future (2071-2090), the cooling and moistening effects of afforestation are dominant in most parts of the temperate zone. Here, the largest amount of forest cover increase has been assumed. Forests have large leaf area and they are aerodynamically rough, that support the more intense vertical mixing compared to other vegetated surfaces. It leads to enhanced ability of evapotranspiration and thereby to the decrease of temperature compared to the reference simulation.

Fig. 4. Climate change signal for temperature (dT) and precipitation (dP) due to emission change (a) and due to potential afforestation (b)

Smaller patches in central and southeast Europe can be characterized also by colder but dryer conditions. Portugal, the Mediterranean coasts and the southern part of the boreal zone show a shift into the warmer and wetter direction (figure 4b). For the Mediterranean region a possible reason for it can be that in this dry area vegetation has deeper roots in the reference simulation than forests in the afforestation experiment. It means in the model that less water is available for cooling through evapotranspiration. Increase of forest cover resulted in higher temperatures and less precipitation in the northern part of Scandinavia and Russia as well as in smaller areas in Spain and around the Black Sea (figure 4b).

in southern and central Europe and in the southern part of Scandinavia, whereas Northeast

Figure 4b represents the direction of the temperature and precipitation changes due to afforestation only. Comparing the simulation results with- and without forest cover changes for the future (2071-2090), the cooling and moistening effects of afforestation are dominant in most parts of the temperate zone. Here, the largest amount of forest cover increase has been assumed. Forests have large leaf area and they are aerodynamically rough, that support the more intense vertical mixing compared to other vegetated surfaces. It leads to enhanced ability of evapotranspiration and thereby to the decrease of temperature

Fig. 4. Climate change signal for temperature (dT) and precipitation (dP) due to emission

and Russia as well as in smaller areas in Spain and around the Black Sea (figure 4b).

Smaller patches in central and southeast Europe can be characterized also by colder but dryer conditions. Portugal, the Mediterranean coasts and the southern part of the boreal zone show a shift into the warmer and wetter direction (figure 4b). For the Mediterranean region a possible reason for it can be that in this dry area vegetation has deeper roots in the reference simulation than forests in the afforestation experiment. It means in the model that less water is available for cooling through evapotranspiration. Increase of forest cover resulted in higher temperatures and less precipitation in the northern part of Scandinavia

Europe can be characterized with warmer and wetter conditions (figure 4a).

compared to the reference simulation.

(b)

(a)

change (a) and due to potential afforestation (b)

From the figure 4 it can be concluded that there are regions, where temperature and/or precipitation changes due to emission change and afforestation show the same sign so that they can enhance each other. In other areas they have the opposite sign, thus depending on their magnitude, afforestation can reduce or fully compensate the effects of the emission change.

## **3.2 Magnitude of the climatic effect of emission change and potential afforestation**

First, those regions have been determined, which are the most affected by climate change for the end of the 21st century. Without any forest cover changes the strongest warming and drying are projected to the Mediterranean area, southern France and over the Iberian Peninsula (figure 5). These signals are in a good agreement with the simulation results of other regional climate models for the same region.

Fig. 5. The regions the most affected by climate change due to emission change (dT: temperature change, dP: precipitation change)

Fig. 6. The regions with the largest effects of forest cover increase on temperature and precipitation (dT: temperature change, dP: precipitation change)

Second, regions, where forest cover increase (without any emission change) has the largest effects on temperature and precipitation have been also identified. For the northern part of central and western Europe temperature decrease due to afforestation can exceed 0.3 °C additionally to more than 10 % increase of the summer precipitation sum (figure 6). The

Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions 305

relatively small compared to the effect of the emission change. In the temperate zone afforestation can reduce the projected warming. In northern part of central Europe and

The increase of forest cover can amplify the projected precipitation change in Sweden, Spain, Belarus and Russia (figure 7b). Whereas in the northern part of central Europe, Ukraine and eastern Finland the precipitation change signal due to emission change can be reduced by afforestation in more than 50 %. Figure 4 shows a border area between the projected precipitation increase and decrease due to emission change. Here the projected precipitation change is relatively small and not significant. In this zone the magnitude of the precipitation change due to afforestation can exceed the magnitude of the climate change

Fig. 8. Change of the summer precipitation sum (dP) due to emission change (2071-2090 vs. 1961-1990), due to potential afforestation (2071-2090) and due to emission change + potential afforestation for the two selected 30 box-regions in France (left) and in northern Poland

Two smaller regions with similar amount of forest cover increase have been selected to represent the regional differences. In the northern part of Poland the changes of the physical properties of the land cover resulted in an increase of evapotranspiration. Due to the increase of latent heat of vaporization, the simulated climate change signal for temperature (2 °C) may decrease by 0.3 °C during the summer months (not shown). This effect of the afforestation is 17 % of climate change signal. Potential afforestation may result in

Ukraine the effect of afforestation can be 15-20 % of the climate change signal.

signal (figure 7b).

(right)

regions characterized by largest cooling and moistening effects of afforestation do not correspond to the areas with the strongest warming and drying due to emission change (figure 5). In many southern European areas afforestation can result in larger than 10 % increase/decrease of summer precipitation sums but these are not statistically significant.

#### **3.3 Sign and magnitude of the climatic feedbacks of potential afforestation relative to the magnitude of the climate change signal**

The magnitude of the climatic effects of afforestation has also been analyzed relative to the effect of the enhanced greenhouse gas emissions in order to determine the regions, where forests can play a major role in altering the climate change signal. In figure 7 the values represent the temperature as well as the precipitation signals for afforestation divided by the climate change signal. The reddish colours are referring to the areas, where the changes of the analyzed climatic variables have the same sign for both afforestation and emission change. Whereas in the regions marked with bluish colours they show opposite sign.

Fig. 7. Climate change signal due to potential afforestation divided by the climate change signal due to emission change for temperature (a) and precipitation (b)

The temperature change signal for potential afforestation is smaller than for emission change in the entire continent (figure 7a). Increase of the forest cover can enhance the climate change signal in the boreal and in the Mediterranean regions but its magnitude is

regions characterized by largest cooling and moistening effects of afforestation do not correspond to the areas with the strongest warming and drying due to emission change (figure 5). In many southern European areas afforestation can result in larger than 10 % increase/decrease of summer precipitation sums but these are not statistically significant.

**3.3 Sign and magnitude of the climatic feedbacks of potential afforestation relative to** 

The magnitude of the climatic effects of afforestation has also been analyzed relative to the effect of the enhanced greenhouse gas emissions in order to determine the regions, where forests can play a major role in altering the climate change signal. In figure 7 the values represent the temperature as well as the precipitation signals for afforestation divided by the climate change signal. The reddish colours are referring to the areas, where the changes of the analyzed climatic variables have the same sign for both afforestation and emission

change. Whereas in the regions marked with bluish colours they show opposite sign.

Fig. 7. Climate change signal due to potential afforestation divided by the climate change

The temperature change signal for potential afforestation is smaller than for emission change in the entire continent (figure 7a). Increase of the forest cover can enhance the climate change signal in the boreal and in the Mediterranean regions but its magnitude is

signal due to emission change for temperature (a) and precipitation (b)

**the magnitude of the climate change signal** 

(b)

(a)

relatively small compared to the effect of the emission change. In the temperate zone afforestation can reduce the projected warming. In northern part of central Europe and Ukraine the effect of afforestation can be 15-20 % of the climate change signal.

The increase of forest cover can amplify the projected precipitation change in Sweden, Spain, Belarus and Russia (figure 7b). Whereas in the northern part of central Europe, Ukraine and eastern Finland the precipitation change signal due to emission change can be reduced by afforestation in more than 50 %. Figure 4 shows a border area between the projected precipitation increase and decrease due to emission change. Here the projected precipitation change is relatively small and not significant. In this zone the magnitude of the precipitation change due to afforestation can exceed the magnitude of the climate change signal (figure 7b).

Fig. 8. Change of the summer precipitation sum (dP) due to emission change (2071-2090 vs. 1961-1990), due to potential afforestation (2071-2090) and due to emission change + potential afforestation for the two selected 30 box-regions in France (left) and in northern Poland (right)

Two smaller regions with similar amount of forest cover increase have been selected to represent the regional differences. In the northern part of Poland the changes of the physical properties of the land cover resulted in an increase of evapotranspiration. Due to the increase of latent heat of vaporization, the simulated climate change signal for temperature (2 °C) may decrease by 0.3 °C during the summer months (not shown). This effect of the afforestation is 17 % of climate change signal. Potential afforestation may result in

Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions 307

Figure 9 shows that the climatic effects of the maximal afforestation are the largest in the northeastern part of Hungary (NEH), which does not correspond to the area with the largest

Fig. 10. Change of the summer precipitation sum (dP) due to emission change (2071-2100 vs. 1961-1990), due to maximal afforestation (2071-2100) and due to emission change + maximal afforestation in Hungary and in the three investigated regions (SWH: southwestern part of Hungary, SEH: southeastern part of Hungary, NEH: northeastern part of Hungary).

For the regions SWH and NEH as well as for the area with the largest increase of forest cover (SEH) the magnitude of the effects of forest cover increase on the summer precipitation sum has been compared to the magnitude of the projected climate change signal. For all three regions and for the whole area of Hungary the precipitation change due to emission changes and the precipitation change due to maximal afforestation have opposite effect (figure 10). This means that the projected climate change signal for precipitation can be reduced by the increase of the forest cover. The magnitude of the climate change altering effect of the maximal afforestation differs among regions. The area of SWH can be characterized by 30 % relative precipitation decrease due to emission change (2071-2100 vs. 1961-1990), which could be hardly compensated by the forest cover increase. In SEH, the significant decrease of summer precipitation can be weakened through the afforestation. In the mountainous region NEH, the projected drying in summer is the mildest (17 %). But also this is the area, where the largest precipitation-increasing (9 %) effect of maximal afforestation is observable. Here, for precipitation, more than half of the projected climate change signal can be relieved with enhanced forest cover (figure 10). These effects can play a major role in mitigating of the probability and severity of droughts in this

A regional scale sensitivity study has been carried out to assess the climatic effects of forest cover increase for Europe, for the end of the 21st century. Applying the regional climate

amount of afforestation.

Adapted from: Gálos et al., 2011

region (Gálos et al., 2011).

**4. Conclusions** 

significant increase of the summer precipitation sum by 16 % (47 mm) compared to the reference simulation (figure 8). Due to the enhanced greenhouse gas emissions, a relatively small (-7 %; -21 mm) decrease of precipitation is projected for the period 2071-2090 compared to 1971-1990. Thus the combined effect of afforestation and emission changes led to 8 % (25 mm) increase of precipitation compared to the past time period without any land cover changes (figure 8).

In France, the precipitation change signal for emission change is projected to be larger (-26% -91 mm). Here, the 16 % (42 mm) increase of precipitation due to afforestation can diminish but not fully compensate the significant decrease of precipitation due to emission change (figure 8). If emission changes occur together with potential afforestation, the summer precipitation sum might decrease by 14 % (-49 mm).

## **3.4 Country scale effects of maximal afforestation in Hungary**

A more detailed case study has been carried out for Hungary to assess the regional differences of the climatic role of afforestation within the country. These analyses concentrated on the possible mitigation of the strong warming and drying tendency of summers projected for this region.

Fig. 9. The areas the most affected by climate change (left) and the spatial distribution of the precipitation-increasing (dP) and temperature-decreasing (dT) effect of maximal afforestation compared to the reference (right). The three investigated regions are hatched: the area the most affected by climate change (southwestern part of Hungary: SWH), the region with the largest amount of afforestation (southeastern part of Hungary: SEH) and the area, where the effect of maximal afforestation on precipitation is the largest (northeastern part of Hungary: NEH)

Based on the applied threshold values for temperature and relative precipitation the southwestern part of Hungary (SWH) can be the most affected by warming and drying. In this area the projected increase of summer temperature can exceed the 3.5 °C and the decrease of the summer precipitation sum the 25 % for the time period 2071-2100 compared to 1961-1990. The least affected are the northeastern areas.

Similarly the region has been determined, which can be characterized by the largest precipitation increase and temperature decrease due to maximal afforestation (figure 9). Figure 9 shows that the climatic effects of the maximal afforestation are the largest in the northeastern part of Hungary (NEH), which does not correspond to the area with the largest amount of afforestation.

Fig. 10. Change of the summer precipitation sum (dP) due to emission change (2071-2100 vs. 1961-1990), due to maximal afforestation (2071-2100) and due to emission change + maximal afforestation in Hungary and in the three investigated regions (SWH: southwestern part of Hungary, SEH: southeastern part of Hungary, NEH: northeastern part of Hungary). Adapted from: Gálos et al., 2011

For the regions SWH and NEH as well as for the area with the largest increase of forest cover (SEH) the magnitude of the effects of forest cover increase on the summer precipitation sum has been compared to the magnitude of the projected climate change signal. For all three regions and for the whole area of Hungary the precipitation change due to emission changes and the precipitation change due to maximal afforestation have opposite effect (figure 10). This means that the projected climate change signal for precipitation can be reduced by the increase of the forest cover. The magnitude of the climate change altering effect of the maximal afforestation differs among regions. The area of SWH can be characterized by 30 % relative precipitation decrease due to emission change (2071-2100 vs. 1961-1990), which could be hardly compensated by the forest cover increase. In SEH, the significant decrease of summer precipitation can be weakened through the afforestation. In the mountainous region NEH, the projected drying in summer is the mildest (17 %). But also this is the area, where the largest precipitation-increasing (9 %) effect of maximal afforestation is observable. Here, for precipitation, more than half of the projected climate change signal can be relieved with enhanced forest cover (figure 10). These effects can play a major role in mitigating of the probability and severity of droughts in this region (Gálos et al., 2011).

## **4. Conclusions**

306 Greenhouse Gases – Emission, Measurement and Management

significant increase of the summer precipitation sum by 16 % (47 mm) compared to the reference simulation (figure 8). Due to the enhanced greenhouse gas emissions, a relatively small (-7 %; -21 mm) decrease of precipitation is projected for the period 2071-2090 compared to 1971-1990. Thus the combined effect of afforestation and emission changes led to 8 % (25 mm) increase of precipitation compared to the past time period without any land

In France, the precipitation change signal for emission change is projected to be larger (-26% -91 mm). Here, the 16 % (42 mm) increase of precipitation due to afforestation can diminish but not fully compensate the significant decrease of precipitation due to emission change (figure 8). If emission changes occur together with potential afforestation, the summer

A more detailed case study has been carried out for Hungary to assess the regional differences of the climatic role of afforestation within the country. These analyses concentrated on the possible mitigation of the strong warming and drying tendency of

Fig. 9. The areas the most affected by climate change (left) and the spatial distribution of the precipitation-increasing (dP) and temperature-decreasing (dT) effect of maximal afforestation compared to the reference (right). The three investigated regions are hatched: the area the most affected by climate change (southwestern part of Hungary: SWH), the region with the largest amount of afforestation (southeastern part of Hungary: SEH) and the area, where the effect of maximal afforestation on precipitation is the largest (northeastern part of Hungary: NEH)

Based on the applied threshold values for temperature and relative precipitation the southwestern part of Hungary (SWH) can be the most affected by warming and drying. In this area the projected increase of summer temperature can exceed the 3.5 °C and the decrease of the summer precipitation sum the 25 % for the time period 2071-2100 compared

Similarly the region has been determined, which can be characterized by the largest precipitation increase and temperature decrease due to maximal afforestation (figure 9).

to 1961-1990. The least affected are the northeastern areas.

cover changes (figure 8).

summers projected for this region.

precipitation sum might decrease by 14 % (-49 mm).

**3.4 Country scale effects of maximal afforestation in Hungary** 

A regional scale sensitivity study has been carried out to assess the climatic effects of forest cover increase for Europe, for the end of the 21st century. Applying the regional climate

Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions 309

Based on the simulation results it can be concluded that only large, contiguous forest blocks have robust effect on the climate on regional scale. The climate change altering effects of forest cover change show large spatial differences. In the most climate change affected regions climatic effects of afforestation are relatively small. But in other regions they can play an important role in reducing the probability and severity of climatic extremes (Gálos et al., 2011). These sensitivity studies confirm, that at the end of the 21st century in the regions, which are the most affected by climate change, vegetation feedbacks have weaker influence on the atmospheric circulation in comparison to the greenhouse gas forcing (Betts, 2007; Wramneby et al., 2010) and afforestation is not a substitute for reduced greenhouse gas

For each of the introduced sensitivity studies, one regional climate model has been applied driven by one emission scenario. For the climate change signal, results are in general agreement with earlier studies from the EU-projects PRUDENCE and ENSEMBLES. For the land cover change experiments the simulated sign and magnitude of the climatic effects can largely depend on the description of the land surface properties and processes in the applied climate model as well as the variability of the climatic and soil conditions and vegetation characteristics of the studied region. Therefore more similar fine-scale pilot studies are essential to reduce the uncertainties and to draw appropriate conclusions for decision makers about the role of the forests in the climate change mitigation on country scale.

From practical point of view, the investigation of the role of the land surface in the climate system gets even more important with the expected land cover change due to climate change and land use politics that differ among regions. Research into forestry's effect on climate is still relatively new and requires a major expansion to support policy development (Anderson et al., 2010). Our results also pointed out that the changes of forest cover could enhance and reduce the projected regional climate change. Therefore regional scale information is substantial about the climatic feedbacks of the future land cover and land use for the adaptation to the climate change in agriculture, forestry and water management.

Results of these sensitivity experiments contribute to the assessment of the climate change altering effects of forest cover change. It helps to identify the areas, where forest cover increase is climatically the most beneficial and should be supported to reduce the projected climate change. Here, the existing forests should be maintained. Thus the present study

Further research is essential to get information about the internal model variability and for studying the robustness of the results applying an ensemble of regional climate model simulations. In these sensitivity studies the spatial and temporal changes of vegetation cover due to climate change was not considered. Although the projected precipitation and temperature changes may have severe impacts on the spatial distribution of forests. Especially in the case-study region Hungary the drastic reduction of the macroclimatically suitable area for forests is expected at the forest/steppe limit (Berki et al., 2009; Mátyás et al., 2010; Czúcz et al., 2011). Therefore for the long-term investigation of forest-climate

provides an important basis for the future adaptation strategies.

emissions (Arora & Montenegro, 2011).

**Practical application of the results** 

**Outlook** 

model REMO, the projected temperature and precipitation tendencies have been analyzed for summer, based on the results of the A2 IPCC-SRES emission scenario simulation. For the end of the 21st century it has been studied, whether the effects of the emission changes could be reduced or enhanced by the increased forest cover. The magnitude of the biogeophysical effects of afforestation on temperature and precipitation has been determined relative to the magnitude of the climate change signal due to emission change. The regions have been determined, in which forests play a major role in altering of the projected climate change.

Simulation results of this sensitivity study can be summarized as follows:

	- *Sign of the effects:* For the A2 emission scenario, a positive temperature signal is expected in whole Europe, which is projected to occur together with precipitation decrease in southern and central Europe and in the southern part of Scandinavia.
	- *Magnitude of the effects:* The strongest warming and drying are projected for the Mediterranean area, southern France and over the Iberian Peninsula.
	- *Sign of the effects:* In most parts of the temperate zone the cooling and moistening effect of afforestation dominates. Portugal, the Mediterranean coasts and the southern part of the boreal zone show a shift into the warmer and wetter direction. Warmer and dryer conditions over larger areas may occur in the boreal region.
	- *Magnitude of the effects:* Afforestation has the largest climatic effects in the northern part of central and western Europe, where the temperature decrease due to afforestation may exceed 0.3 °C additionally to more than 10 % increase of the summer precipitation sum.
	- *Sign of the effects:* In the largest part of the temperate zone precipitation and temperature anomalies due to forest cover increase show the opposite sign than due to emission change, which means that the climate change signal can be reduced by afforestation. Whereas in Sweden, in Spain and in some regions in the eastern part of the continent afforestation can amplify the climate change signal for both investigated variables.
	- *Magnitude of the effects:* The largest climate change mitigating effects of afforestation can be expected in northern Germany, Poland and Ukraine, which is 15-20 % of the climate change signal for temperature and more than 50 % for precipitation. These changes are significant at the 90 % confidence level.
	- *Sign of the effects:* The projected climate change signal for precipitation can be reduced assuming maximal afforestation.
	- *Magnitude of the effects:* The strong warming and drying tendency projected for southwest Hungary could be hardly compensated by forest cover increase. But in the northwestern region more than half of the projected climate change signal for precipitation can be relieved with enhanced forest cover.

For Hungary results represent the first regional scale assessment of the climatic role of forests for a long future time period and its role in adapting to climate change.

Based on the simulation results it can be concluded that only large, contiguous forest blocks have robust effect on the climate on regional scale. The climate change altering effects of forest cover change show large spatial differences. In the most climate change affected regions climatic effects of afforestation are relatively small. But in other regions they can play an important role in reducing the probability and severity of climatic extremes (Gálos et al., 2011). These sensitivity studies confirm, that at the end of the 21st century in the regions, which are the most affected by climate change, vegetation feedbacks have weaker influence on the atmospheric circulation in comparison to the greenhouse gas forcing (Betts, 2007; Wramneby et al., 2010) and afforestation is not a substitute for reduced greenhouse gas emissions (Arora & Montenegro, 2011).

For each of the introduced sensitivity studies, one regional climate model has been applied driven by one emission scenario. For the climate change signal, results are in general agreement with earlier studies from the EU-projects PRUDENCE and ENSEMBLES. For the land cover change experiments the simulated sign and magnitude of the climatic effects can largely depend on the description of the land surface properties and processes in the applied climate model as well as the variability of the climatic and soil conditions and vegetation characteristics of the studied region. Therefore more similar fine-scale pilot studies are essential to reduce the uncertainties and to draw appropriate conclusions for decision makers about the role of the forests in the climate change mitigation on country scale.

### **Practical application of the results**

From practical point of view, the investigation of the role of the land surface in the climate system gets even more important with the expected land cover change due to climate change and land use politics that differ among regions. Research into forestry's effect on climate is still relatively new and requires a major expansion to support policy development (Anderson et al., 2010). Our results also pointed out that the changes of forest cover could enhance and reduce the projected regional climate change. Therefore regional scale information is substantial about the climatic feedbacks of the future land cover and land use for the adaptation to the climate change in agriculture, forestry and water management.

Results of these sensitivity experiments contribute to the assessment of the climate change altering effects of forest cover change. It helps to identify the areas, where forest cover increase is climatically the most beneficial and should be supported to reduce the projected climate change. Here, the existing forests should be maintained. Thus the present study provides an important basis for the future adaptation strategies.

### **Outlook**

308 Greenhouse Gases – Emission, Measurement and Management

model REMO, the projected temperature and precipitation tendencies have been analyzed for summer, based on the results of the A2 IPCC-SRES emission scenario simulation. For the end of the 21st century it has been studied, whether the effects of the emission changes could be reduced or enhanced by the increased forest cover. The magnitude of the biogeophysical effects of afforestation on temperature and precipitation has been determined relative to the magnitude of the climate change signal due to emission change. The regions have been determined, in which forests play a major role in altering of the

 *Sign of the effects:* For the A2 emission scenario, a positive temperature signal is expected in whole Europe, which is projected to occur together with precipitation decrease in southern and central Europe and in the southern part of Scandinavia. *Magnitude of the effects:* The strongest warming and drying are projected for the

 *Sign of the effects:* In most parts of the temperate zone the cooling and moistening effect of afforestation dominates. Portugal, the Mediterranean coasts and the southern part of the boreal zone show a shift into the warmer and wetter direction. Warmer and dryer conditions over larger areas may occur in the boreal region. *Magnitude of the effects:* Afforestation has the largest climatic effects in the northern part of central and western Europe, where the temperature decrease due to afforestation may exceed 0.3 °C additionally to more than 10 % increase of the

 *Sign of the effects:* In the largest part of the temperate zone precipitation and temperature anomalies due to forest cover increase show the opposite sign than due to emission change, which means that the climate change signal can be reduced by afforestation. Whereas in Sweden, in Spain and in some regions in the eastern part of the continent afforestation can amplify the climate change signal for

 *Magnitude of the effects:* The largest climate change mitigating effects of afforestation can be expected in northern Germany, Poland and Ukraine, which is 15-20 % of the climate change signal for temperature and more than 50 % for precipitation. These

*Sign of the effects:* The projected climate change signal for precipitation can be

 *Magnitude of the effects:* The strong warming and drying tendency projected for southwest Hungary could be hardly compensated by forest cover increase. But in the northwestern region more than half of the projected climate change signal for

For Hungary results represent the first regional scale assessment of the climatic role of

Mediterranean area, southern France and over the Iberian Peninsula.

c. *Climate change signal due to potential afforestation relative to the emission change:* 

changes are significant at the 90 % confidence level.

precipitation can be relieved with enhanced forest cover.

forests for a long future time period and its role in adapting to climate change.

reduced assuming maximal afforestation.

Simulation results of this sensitivity study can be summarized as follows:

projected climate change.

a. *Climate change signal due to emission change:* 

b. *Climate change signal due to potential afforestation:* 

summer precipitation sum.

both investigated variables.

d. *Based on the detailed case study for Hungary:* 

Further research is essential to get information about the internal model variability and for studying the robustness of the results applying an ensemble of regional climate model simulations. In these sensitivity studies the spatial and temporal changes of vegetation cover due to climate change was not considered. Although the projected precipitation and temperature changes may have severe impacts on the spatial distribution of forests. Especially in the case-study region Hungary the drastic reduction of the macroclimatically suitable area for forests is expected at the forest/steppe limit (Berki et al., 2009; Mátyás et al., 2010; Czúcz et al., 2011). Therefore for the long-term investigation of forest-climate

Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions 311

Betts R. (2007). Implications of land ecosystem-atmosphere interactions for strategies for

Bonan G.B., Pollard D. & Thompson S.L. (1992). Effects of boreal forests on global climate.

Bonan G.B. (1997). Effects of land use on the climate of the United States. *Climatic Change,* 37,

Bonan G.B., Levis S., Kergoat L. & Oleson K.W. (2002). Landscapes as patches of plant

Bonan G.B. (2004). Biogeophysical feedbacks between land cover and climate. In: *Ecosystems* 

Geophysical Monograph American Geophysical Union, Washington, D.C. Bonan G.B. 2008. Forests and climate change: forcings, feedbacks, and the climate benefits of

Bounoua L., Defries R., Collatz G.J., Sellers P., & Khan H. (2002). Effects of land cover

Brovkin V. (2002). Climate-vegetation interaction. *J. Phys.* IV France 12, 57-82, doi:

Christensen J.H., Carter T.R., Rummukainen M., & Amanatidis G. (2007). Evaluating the

Christensen J.H. & Christensen O.B. (2007). A summary of the PRUDENCE model

Copeland J.H., Pielke R.A. & Kittel T.G.F. (1996). Potential climatic impacts of vegetation

Czúcz B., Gálhidy L. & Mátyás Cs. (2011). Present and forecasted xeric climatic limits of

Drüszler Á., Vig P. & Csirmaz K. (2011). Effects of Historical Land Cover Changes on the

Feddema J.J., Oleson K.W., Bonan G.B., Mearns L.O., Buja L.E., Meehl G.A. & Washington

Fischer E.M. & Schär C. (2010). Consistent geographical patterns of changes in high-impact European heatwaves. *Nature Geoscience,* 3, 398-403, doi: 10.1038/NGEO866 Gálos B., Lorenz Ph., & Jacob D. (2007). Will dry events occur more often in Hungary in the *future? Environ. Res. Lett.,* 2, 034006 (9pp), doi: 10.1088/1748-9326/2/3/034006 Gálos B., Mátyás Cs. & Jacob D. (2011). Regional characteristics of climate change altering

Climates. *Science,* 310, 1674 –1678, doi: 10.1126/science.1118160

performance and utility of regional climate models: the PRUDENCE project. *Clim.* 

projections of changes in European climate by the end of this century. *Clim. Change,* 

change: A regional modelig study. *Journal of Geophysical Research,* 101, 7409-7418

beech and sessile oak distribution at low altitudes in Central Europe. *Ann. For. Sci.,*

Precipitation Distribution in Hungary. *Riscuri Si Catastrofe (Risks and Disasters)*,

W.M. (2005). The Importance of Land-Cover Change in Simulating Future

effects of afforestation. *Environ. Res. Lett.* 6, 044010 (9pp), doi:10.1088/1748-

0889.2007.00284.x

10.1051/jp4:20020452

doi: 10.1029/95JD02676

ISSN: 15845273; I/2011 (in print)

68(1), 99-108

9326/6/4/044010

449–486

*Nature,* 359, 716-718, doi: 10.1038/359716a0

*Biogeochemical Cycles,* 16, 30pp,10.1029/2000GB001360

forests. *Science,* 320, 1444–1449, doi: 10.1126/science.1155121

conversion on surface climate. *Climatic Change,* 52, 29-64

*Change*, 81, 1-6, DOI 10.1007/s10584-006-9211-6

81, 7-30, doi: 10.1007/s10584-006-9210-7

climate change adaptation and mitigation. *Tellus,* 59B, 602-615, doi: 10.1111/j.1600-

functional types: an integrating concept for climate and ecosystem models. *Global* 

*and Land Use Change.*, DeFries R.S., Asner G.P., & Houghton R.A. (eds), 153, 61-72,

interactions in the future, regional climate modelling is essential with dynamic vegetation scheme and more detailed and improved description of the forest related processes and parameters combining the biogeophysical and biogeochemical effects.

## **5. Acknowledgements**

The authors give special thanks to the Regional Modelling Group of the Max Planck Institute for Meteorology, Hamburg for the fruitful scientific discussions about the simulation results. We thank to Prof. Dr. Csaba Mátyás for his expertise and suggestions regarding to the practical importance of this topic. The authors would like to thank to Georg Kindermann (IIASA, International Institute for Applied Systems Analysis Austria) for providing the forest cover database as well as to Kevin Sieck and Claas Teichmann for the technical guidance during the simulations. The REMO simulations without land cover changes have been carried out in the frame of the EU-project ENSEMBLES. This research was supported by the EC-FP7 project CC-TAME (www.cctame.eu; grant agreement n° 212535) and the TÁMOP 4.2.2-08 joint EU-national research project.

## **6. References**


interactions in the future, regional climate modelling is essential with dynamic vegetation scheme and more detailed and improved description of the forest related processes and

The authors give special thanks to the Regional Modelling Group of the Max Planck Institute for Meteorology, Hamburg for the fruitful scientific discussions about the simulation results. We thank to Prof. Dr. Csaba Mátyás for his expertise and suggestions regarding to the practical importance of this topic. The authors would like to thank to Georg Kindermann (IIASA, International Institute for Applied Systems Analysis Austria) for providing the forest cover database as well as to Kevin Sieck and Claas Teichmann for the technical guidance during the simulations. The REMO simulations without land cover changes have been carried out in the frame of the EU-project ENSEMBLES. This research was supported by the EC-FP7 project CC-TAME (www.cctame.eu; grant agreement n°

Anav A., Ruti P.M., Artale V. & Valentini R. (2010). Modelling the effects of land-cover changes on surface climate in the Mediterranean region. *Clim. Res.,* 41, 91-104 Anderson R.G., Canadel J.G., Randerson J.T., Jackson R.B., Hungate B.A., Baldocchi D.D.,

Arora V.K. & Montenegro A. 2011. Small temperature benefits provided by realistic afforestation efforts. *Nature Geoscience* ,doi: 10.1038/NGEO1182, 5pp Bala G., Caldeira K., Wickett M., Phillips T.J., Lobell D.B. , Delire C. & Mirin A. (2007).

Bartholy J., Pongrácz R. & Gelybó Gy. (2007). Regional climate change expected in Hungary

Berki I., Rasztovits E., Móricz N. & Mátyás Cs. (2009). Determination of the drought

Beniston M., Stephenson D.B., Christensen O.B., Ferro C.A.T., Frei C., Goyette S., Halsnaes

model projections. *Clim. Change,* 81, 71–95, doi:10.1007/s10584-006-9226-z Beniston M. (2009). Trends in joint quantiles of temperature and precipitation in Europe

Betts R.A. (2001). Biogeophysical impacts of land use on present-day climate: near-surface

for 2071-2100. *Applied Ecology and Environmental Research*, 5, 1-17

Ban-Weiss G.A., Bonan G.B., Caldeira K., Cao L., Diffenbaugh N.S., Gurney K.R., Kueppers L.M., Law B.E., Luyssaert S., & O'Halloran T.L. (2010). Biophysical considerations in forestry for climate protection. *Front Ecol Environ*;

Combined climate and carbon-cycle effects of large-scale deforestation. *Proc Natl* 

tolerance limit of beech forests and forecasting their future distribution in Hungary.

K., Holt T., Jylhä K., Koffi B., Palutikof J., Schöll R., Semmler T. & Woth K. (2007). Future extreme events in European climate: an exploration of regional climate

since 1901 and projected for 2100. *Geophys. Res. Lett.*, 36, L07707,

temperature change and radiative forcing. *Atmospheric Science Letters* 1, 23pp, doi:

parameters combining the biogeophysical and biogeochemical effects.

212535) and the TÁMOP 4.2.2-08 joint EU-national research project.

**5. Acknowledgements** 

**6. References** 

doi:10.1890/090179

*Acad Sci USA* 104, 6550–6555

doi:10.1029/2008GL037119

10.1006/asle.2001.0023

*Cereal Research Communations,* 37, 613-616


Regional-Scale Assessment of the Climatic Role of Forests Under Future Climate Conditions 313

Jungclaus J.H., Keenlyside N., Botzet M., Haak H., Luo J-J., Latif M., Marotzke J.,

Kjellström E., Bärring L., Jacob D., Jones R., Lenderink G. & Schär C. (2007). Modelling daily

Kjellström E., Nikulin G., Hansson U., Strandberg G. & Ullerstig A. (2011). 21st century

van der Linden P. & Mitchell J.F.B. (eds.) (2009). *ENSEMBLES: Climate Change and its Impacts:* 

Majewski D. (1991). The Europa-Modell of the Deutscher Wetterdienst ECMWF. *Seminar on* 

Mátyás Cs. (2009). Ecological perspectives of climate change in Europe's continental,

Mátyás Cs. (2010). Forecasts needed for retreating forests (Opinion). *Nature*, 464, 1271 doi:

Mátyás Cs., Berki I., Czúcz B., Gálos B., Móricz N. & Rasztovits E. (2010). Future of beech in

Nakicenovic N. et al. (2000). IPCC Special Report on Emission Scenarios. Cambridge:

Oleson K.W., Bonan G.B., Levis S. & Vertenstein M. (2004). Effects of land use change on

Pielke R.A., Avissar Sr R., Raupach M., Dolman A.J., Zeng X. & Denning A.S. (1998).

Pitman A.J. (2003). The evolution off, and revolution in, land surface schemes designed for

Radvánszky B. & Jacob D. (2009). The Changing Annual Distribution of Rainfall in the

Rechid D. & Jacob D. (2006). Influence of monthly varying vegetation on the simulated

Rechid D., Raddatz T.J. & Jacob D. (2008a). Parameterization of snow-free land surface

Rechid D., Hagemann S. & Jacob D. (2008b). Sensitivity of climate models to seasonal

*Climate Dynamics,* 23, 117–132, doi: 10.1007/s00382-004-0426-9

climate model. *Int. J. of Climatol.* 23, 479-510, doi: 10.1002/joc.893

weather and climate. *Global Change Biology,* 4, 461–475

*Zeitschrift für Geomorphologie*, 53, 171195(25)

climate in Europe. *Meteorol Z.,* 15, 99-116

10.1007/s00704-007-0371-8

the coupled model ECHAM5/MPI-OM. *J. Climate,* 19, 3952–3972

*Change*, 81, 249-265, doi: 10.1007/s105800692205

Hadley Centre, FitzRoy Road, Exeter EX1 3PB, UK

*numerical methods in atmospheric models,* 2, 147-191

NATO Science Series, Springer Verl.

Cambridge University Press, p 599

0870.2010.00475

10.1038/4641271a

*Hung.*, 6, 91-110

Mikolajewicz U. & Roeckner E. (2006). Ocean circulation and tropical variability in

temperature extremes: Recent climate and future changes over Europe. *Climatic* 

changes in the European climate: uncertainties derived from an ensemble of regional climate model simulations. *Tellus,* 63A(1), 24-40, doi:10.1111/j.1600-

*Summary of research and results from the ENSEMBLES project*. 160pp, Met Office

drought-threatened Southeast In: *Regional aspects of climate-terrestrial-hydrologic interactions in non-boreal Eastern Europe,* Groisman P.Y., Ivanov S.V. eds, pp 31-42,

Southeast Europe from the perspective of evolutionary ecology. *Acta Silv. & Lign.* 

North American climate: impact of surface datasets and model Biogeophysics.

Interactions between the atmosphere and terrestrial ecosystems: influence on

Drainage Area of the River Tisza during the Second Half of the 21st Century.

albedo as a function of vegetation phenology based on MODIS data and applied in climate modelling. *Theor. Appl. Climatol.*, 95, 245-255 doi: 10.1007/s00704-008-0003-y

variability of snow-free land surface albedo *Theor. Appl. Climatol.,* 95, 197-221 doi:


Giorgi F., Bi X. & Pal J.S. (2004). Mean, interannual variability and trends in a regional

Göttel H., Alexander J., Keup-Thiel E., Rechid D., Hagemann S., Blome T., Wolf A. & Jacob

the Barents Sea Region Climatic Change. *BALANCE Special Issues,* 87, 35-50 Hagemann S., Botzet M., Dümenil L. & Machenhauer M. (1999). Derivation of global GCM

Hagemann S. (2002). An improved land surface parameter dataset for global and regional climate models. *Report 336*, Max-Planck-Institute for Meteorology, Hamburg Hagemann S., Göttel H., Jacob D., Lorenz P. & Roeckner E. (2009). Improved regional scale

Heck P., Lüthi D., Wernli H. & Schär Ch. (2001). Climate impacts of European-scale

Houghton J. et al. (ed) (2001). *Climate change 2001 The Scientific Basis.* Intergovernmental Panel on Climate Change (Cambridge: Cambridge University Press) IPCC (2007). Climate change 2007: Synthesis Report. Contribution of Working Groups I, II

Jackson R.B., Randerson J.T., Canadell J.G., Anderson R.G., Avissar R., Baldocchi D.D.,

Jacob D. (2001). A note to the simulation of the annual and inter-annual variability of the

Jacob D., Bärring L., Christensen O.B., Christensen J.H., de Castro M., Déqué M., Giorgi F.,

*Environ. Res. Lett.,* 3, 044006 (5pp), doi: 10.1088/1748-9326/3/4/044006 Jacob D., Andrae U., Elgered G., Fortelius C., Graham L.P., Jackson S.D., Karstens U.,

*Meteorology and Atmospheric Physics,* 77(1-4), 19-43

*meteorological Service (Időjárás),* 112, 141-153

*Climate Dynamics,* 23, 839–858, doi: 10.1007/s00382-004-0467-0

Institute for Meteorology, Hamburg

*Climate* ,13, 4229-4243

*Physics,* 77, 61-73

Change http//:www.ipcc.ch

catchments. *Climate Dynamics,* 32, 767–781

climate change experiment over Europe. II: climate change scenarios (2071–2100).

D. (2008). Influence of changed vegetation fields on regional climate simulations in

boundary conditions from 1 km land use satellite data. *Report 289* Max-Planck-

processes reflected in projected hydrological changes over large European

anthropogenic vegetation changes: A sensitivity study using a regional climate model. *Journal of Geophysical Research,* 106, 7817-7835, doi: 10.1029/2000JD900673 Hogg E.H., Price D.T., & Black T.A. (2000). Postulated feedbacks of deciduous forest

phenology on seasonal climate patterns in the Western Canadian interior. *Journal of* 

and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate

Bonan G.B., Caldeira K., Diffenbaugh N.S., Field C.B., Hungate B.A., Jobbágy E.G., Kueppers L.M., Nosetto M.D., & Pataki D.E. (2008). Protecting climate with forests

Koepken Chr., Lindau R., Podzun R., Rockel B., Rubel F., Sass H.B., Smith R.N.D., van den Hurk B.J.J.M. & Yang X. (2001). A Comprehensive Model Intercomparison Study Investigating the Water Budget during the BALTEX-PIDCAP Period.

water budget over the Baltic Sea drainage basin. *Meteorology and Atmospheric* 

Hagemann S., Hirschi M., Jones R., Kjellström E., Lenderink G., Rockel B., Sánchez E., Schär C., Seneviratne S.I., Sommot S., van Ulden A. & van den Hurk B. (2007). An inter-comparison of regional climate models for Europe: model performance in present-day climate. *Climatic Change*, 81, 31-52, doi: 10.1007/s10584-006-9213-4 Jacob D., Kotova L., Lorenz P., Moseley Ch. & Pfeifer S. (2008). Regional climate modeling

activities in relation to the CLAVIER project. *Quarterly Journal of the Hungarian* 


**15**

*USA* 

Ingrid Cnossen

**Climate Change in the Upper Atmosphere**

Studies of climate change traditionally focus on long-term changes taking place in the troposphere. This is understandable, as this is the part of the atmosphere that most directly affects life on the surface. However, long-term trends higher in the atmosphere can have important consequences as well. The middle and upper atmosphere consist of the stratosphere (~15-50 km), the mesosphere (50-90 km), the thermosphere (90-800 km), and

Many satellites operate within the thermosphere. Since the drag exerted upon them is proportional to the ambient atmospheric density, long-term changes in the density in the thermosphere can affect their trajectories and orbital lifetimes. Also, long-term trends in the ionosphere can affect radio wave propagation, and therefore the performance of the Global Positioning System (GPS) and other space-based navigation and communication systems that rely on radio waves (Laštovička et al., 2006a). In addition to these in-situ effects, there is evidence that perturbations in the upper atmosphere may propagate downward (e.g. Haynes et al., 1991), and that the state of the middle and upper atmosphere has an influence on the troposphere (e.g. Baldwin & Dunkerton, 2001; Sassi et al., 2010). There is thus a possibility that long-term trends in the upper atmosphere could play a role in tropospheric climate too. For these reasons it is important to gain a good understanding of the processes

As in the lower atmosphere, the increase in carbon-dioxide concentration is thought to be one of the main causes for climatic changes in the upper atmosphere (Laštovička et al., 2006a). However, it has become clear in recent years that several other mechanisms for longterm change must be considered as well (e.g. Qian et al. 2011). These include changes in ozone, methane and water vapour concentration, the secular variation of the Earth's magnetic field, and long-term changes in solar and geomagnetic activity. Also climate

The aim of this chapter is to review our current knowledge of the processes mentioned above and their role in causing climate change in the upper atmosphere. A brief overview of observed trends in several variables is given first (section 2), followed by a discussion of the possible responsible mechanisms (sections 3-6). Estimates of trends caused by each of the mechanisms as determined from model simulations are compared to observed trends where

the embedded ionosphere, the ionized part of the upper atmosphere (see figure 1).

that cause long-term trends in this part of the atmosphere.

change in the lower atmosphere could have an effect.

**1. Introduction** 

possible.

*High Altitude Observatory, National Center for Atmospheric Research* 


## **Climate Change in the Upper Atmosphere**

## Ingrid Cnossen

*High Altitude Observatory, National Center for Atmospheric Research USA* 

#### **1. Introduction**

314 Greenhouse Gases – Emission, Measurement and Management

Roeckner E., Arpe K., Bengtsson L., Christoph M., Claussen M., Dümenil L., Esch M.,

Roeckner E., Brokopf R., Esch M., Giorgetta M., Hagemann S., Kornblueh L., Manzini E.,

and future climates. *Clim. Dyn.* 29, 319–332, doi: 10.1007/s00382-007-0240-2 Schär C., Vidale P.L., Lüthi D., Frei C., Häberli C., Liniger, M.A. & Appenzeller C. (2004).

Seneviratne S.I., Lüthi D., Litschi M. & Schär C. (2006). Land-atmosphere coupling and climate change in Europe. *Nature,* 443, 205–209, doi: 10.1038/nature05095 Seneviratne S.I. et al. (2010). Investigating soil moisture-climate interactions in a changing

Szépszó G. (2008). Regional change of climate extremes in Hungary based on different regional climate models of the PRUDENCE project. *Időjárás,* 112, 265-283 Teuling A.J., Seneviratne S.I., Stoeckli R., Reichstein M., Moors E., Ciais P., Luyssaert S., van

Vidale P.L., Lüthi D., Wegmann R., & Schär C. (2007). European summer climate variability

Wramneby A., Smith B. & Samuelsson P. (2010). Hot spots of vegetation-climate feedbacks

climate. *Report 218 Max-Planck-Institut für Meteorologie, Hamburg*

*Nature* 427, 332–336, doi: 10.1038/nature02300

climate: A review. *Earth. Sci. Rev.,* 99, 125-161

10.1007/s10584-006-9218-z

doi:10.1029/2010JD014307

exchange to heatwaves. *Nature Geoscience* 3(10), 722-727

Giorgetta M., Schlese U. & Schulzweida U. (1996). The atmospheric general circulation model ECHAM-4: Model description and simulation of the present day

Schlese U. & Schulzweida U. (2006). Sensitivity of simulated climate to horizontal and vertical resolution in the ECHAM5 atmosphere model. *J. Clim.* 19, 3771–3791 Sánchez E., Gaertner M.A., Gallardo C., Padorno E., Arribas A. & Castro M. (2007). Impacts

of a change in vegetation description on simulated European summer present-day

The role of increasing temperature variability in European summer heatwaves.

den Hurk B., Ammann C., Bernhofer C., Dellwik E., Gianelle D., Gielen B., Gruenwald T., Klumpp K., Montagnani L., Moureaux C., Sottocornola M., & Wohlfahrt G. (2010). Contrasting response of European forest and grassland energy

in a heterogeneous multi-model ensemble. *Climatic Change*, 81, 209–232. doi:

under future greenhouse forcing in Europe. *J. Geophys. Res.*, 115, D21119,

Studies of climate change traditionally focus on long-term changes taking place in the troposphere. This is understandable, as this is the part of the atmosphere that most directly affects life on the surface. However, long-term trends higher in the atmosphere can have important consequences as well. The middle and upper atmosphere consist of the stratosphere (~15-50 km), the mesosphere (50-90 km), the thermosphere (90-800 km), and the embedded ionosphere, the ionized part of the upper atmosphere (see figure 1).

Many satellites operate within the thermosphere. Since the drag exerted upon them is proportional to the ambient atmospheric density, long-term changes in the density in the thermosphere can affect their trajectories and orbital lifetimes. Also, long-term trends in the ionosphere can affect radio wave propagation, and therefore the performance of the Global Positioning System (GPS) and other space-based navigation and communication systems that rely on radio waves (Laštovička et al., 2006a). In addition to these in-situ effects, there is evidence that perturbations in the upper atmosphere may propagate downward (e.g. Haynes et al., 1991), and that the state of the middle and upper atmosphere has an influence on the troposphere (e.g. Baldwin & Dunkerton, 2001; Sassi et al., 2010). There is thus a possibility that long-term trends in the upper atmosphere could play a role in tropospheric climate too. For these reasons it is important to gain a good understanding of the processes that cause long-term trends in this part of the atmosphere.

As in the lower atmosphere, the increase in carbon-dioxide concentration is thought to be one of the main causes for climatic changes in the upper atmosphere (Laštovička et al., 2006a). However, it has become clear in recent years that several other mechanisms for longterm change must be considered as well (e.g. Qian et al. 2011). These include changes in ozone, methane and water vapour concentration, the secular variation of the Earth's magnetic field, and long-term changes in solar and geomagnetic activity. Also climate change in the lower atmosphere could have an effect.

The aim of this chapter is to review our current knowledge of the processes mentioned above and their role in causing climate change in the upper atmosphere. A brief overview of observed trends in several variables is given first (section 2), followed by a discussion of the possible responsible mechanisms (sections 3-6). Estimates of trends caused by each of the mechanisms as determined from model simulations are compared to observed trends where possible.

Climate Change in the Upper Atmosphere 317

Fig. 1. Schematic summary of observed global mean trends in the atmosphere, after Laštovička et al. (2006a). Changes in the temperature profile (orange) are indicated by red (warming), blue (cooling) and green (no clear trend) arrows. Changes in the electron density profile (purple) are indicated by black arrows. Vertical arrows indicate the movement of the height of the peak electron density of the layer, while horizontal arrows indicate changes in the peak electron density itself. Trends in the F2 layer vary strongly with location, season, and local time, so that no meaningful global mean can be defined. They are further

Beig et al. (2003) summarized temperature trends in the mesosphere. They reported that the lower and middle mesosphere has cooled on average by 2-3 K/decade and at tropical latitudes the cooling trend increases with height in the upper mesosphere. However, near the mesopause the general consensus is that there is no significant trend. Some dependency of trends on latitude has been noted, but due to a limited number of sites, no clear pattern

For the thermospheric temperature, only a few long-term trend studies are available. Semenov (1996) inferred a temperature trend of -30 K/decade based on changes in the atomic oxygen 630 nm emission layer, which is normally located at ~270 km. Holt & Zhang (2008) and Zhang et al. (2011) used incoherent scatter radar data from Millstone Hill (46.2°N, 288.5°E) to determine the long-term trend in ion temperature, which should be similar to the trend in neutral temperature, due to close thermal coupling between neutrals and ions. Zhang et al. (2011) showed that a warming trend is found between 110 and 200 km altitude, and a cooling trend above 200 km, which increases with height and is stronger for lower solar activity. At 375 km the trend is as strong as -47±11 K/decade (Holt & Zhang, 2008). Donaldson et al. (2010), in a similar analysis of incoherent scatter

discussed in section 2.4.

has yet emerged.

## **2. Observed long-term trends in the upper atmosphere**

## **2.1 Introduction**

When talking about climate change, we generally are referring to quasi-linear changes or "trends" that occur over a period of 30 years or more (by the definition of the World Meteorological Organization). In practice, climate change is not necessarily linear, and trends can (and eventually will) change over time. However, for the purposes of this chapter we will assume that trends that have been observed over the last 50-60 years, or subsets of that time frame, can be meaningfully described in terms of a linear change per decade.

To measure trends, we need high quality, long-term datasets, which are much sparser in the upper atmosphere than they are in the troposphere. In addition, the upper atmosphere is strongly influenced by the 11-year solar cycle, as well as fluctuations in geomagnetic activity (see section 5). This shorter-term variability must be corrected for to be able to detect a long-term trend. Various studies have shown that this is a non-trivial task, as different methodologies may yield different results (Laštovička et al., 2006b), and residual solar signals can remain, especially if the dataset is not sufficiently long (Clilverd et al., 2003).

Trends in the height of the peak electron density of the ionospheric F2 layer, hmF2 (see figure 1), suffer from an additional complication when ionosonde data are used. In that case hmF2 is not directly measured, but calculated from the M(3000)F2 parameter, which is a transmission factor related to the highest frequency that can be propagated between two sites 3000 km apart by ionospheric propagation alone. There are different empirical formulas to calculate hmF2 from M(3000)F2, and unfortunately these can result in different trends from the same dataset, sometimes even with a different sign (Ulich, 2000; Jarvis et al., 2002). When incoherent scatter data are used, hmF2 can be directly measured, but the data record is much shorter, and far fewer sites are available.

Despite these difficulties, considerable progress has been made in building a global picture of long-term trends in the upper atmosphere. In this chapter we focus on a few selected variables for which both observational and theoretical trend estimates are available, so that the two can be directly compared. These are temperature, density, hmF2, and the critical frequency of the F2 layer, foF2, which is directly related to the peak electron density NmF2 as:

$$\rm{N\_mF\_2=1.24} \times 10^{10} \text{(f}\_o\text{F}\_2\text{)}^2 \tag{1}$$

where NmF2 is in m-3 and foF2 is in MHz.

#### **2.2 Temperature**

Figure 1 gives a schematic overview of the observed trends in the temperature and the electron density profile, and also labels the various regions of the atmosphere. Trends in temperature have been determined from a large number of studies and a variety of measurement techniques (e.g. rocketsonde, lidar, radar, satellite data). Trends in the upper atmosphere are mostly negative, except in the mesopause/lower thermosphere region.

When talking about climate change, we generally are referring to quasi-linear changes or "trends" that occur over a period of 30 years or more (by the definition of the World Meteorological Organization). In practice, climate change is not necessarily linear, and trends can (and eventually will) change over time. However, for the purposes of this chapter we will assume that trends that have been observed over the last 50-60 years, or subsets of that time frame, can be meaningfully described in terms of a linear change per

To measure trends, we need high quality, long-term datasets, which are much sparser in the upper atmosphere than they are in the troposphere. In addition, the upper atmosphere is strongly influenced by the 11-year solar cycle, as well as fluctuations in geomagnetic activity (see section 5). This shorter-term variability must be corrected for to be able to detect a long-term trend. Various studies have shown that this is a non-trivial task, as different methodologies may yield different results (Laštovička et al., 2006b), and residual solar signals can remain, especially if the dataset is not sufficiently long (Clilverd et al.,

Trends in the height of the peak electron density of the ionospheric F2 layer, hmF2 (see figure 1), suffer from an additional complication when ionosonde data are used. In that case hmF2 is not directly measured, but calculated from the M(3000)F2 parameter, which is a transmission factor related to the highest frequency that can be propagated between two sites 3000 km apart by ionospheric propagation alone. There are different empirical formulas to calculate hmF2 from M(3000)F2, and unfortunately these can result in different trends from the same dataset, sometimes even with a different sign (Ulich, 2000; Jarvis et al., 2002). When incoherent scatter data are used, hmF2 can be directly measured, but the data

Despite these difficulties, considerable progress has been made in building a global picture of long-term trends in the upper atmosphere. In this chapter we focus on a few selected variables for which both observational and theoretical trend estimates are available, so that the two can be directly compared. These are temperature, density, hmF2, and the critical frequency of the F2 layer, foF2, which is directly related to the peak electron density NmF2 as:

NmF2 = 1.241010(foF2)2 (1)

Figure 1 gives a schematic overview of the observed trends in the temperature and the electron density profile, and also labels the various regions of the atmosphere. Trends in temperature have been determined from a large number of studies and a variety of measurement techniques (e.g. rocketsonde, lidar, radar, satellite data). Trends in the upper atmosphere are mostly negative, except in the mesopause/lower thermosphere

**2. Observed long-term trends in the upper atmosphere** 

record is much shorter, and far fewer sites are available.

where NmF2 is in m-3 and foF2 is in MHz.

**2.2 Temperature** 

region.

**2.1 Introduction** 

decade.

2003).

Fig. 1. Schematic summary of observed global mean trends in the atmosphere, after Laštovička et al. (2006a). Changes in the temperature profile (orange) are indicated by red (warming), blue (cooling) and green (no clear trend) arrows. Changes in the electron density profile (purple) are indicated by black arrows. Vertical arrows indicate the movement of the height of the peak electron density of the layer, while horizontal arrows indicate changes in the peak electron density itself. Trends in the F2 layer vary strongly with location, season, and local time, so that no meaningful global mean can be defined. They are further discussed in section 2.4.

Beig et al. (2003) summarized temperature trends in the mesosphere. They reported that the lower and middle mesosphere has cooled on average by 2-3 K/decade and at tropical latitudes the cooling trend increases with height in the upper mesosphere. However, near the mesopause the general consensus is that there is no significant trend. Some dependency of trends on latitude has been noted, but due to a limited number of sites, no clear pattern has yet emerged.

For the thermospheric temperature, only a few long-term trend studies are available. Semenov (1996) inferred a temperature trend of -30 K/decade based on changes in the atomic oxygen 630 nm emission layer, which is normally located at ~270 km. Holt & Zhang (2008) and Zhang et al. (2011) used incoherent scatter radar data from Millstone Hill (46.2°N, 288.5°E) to determine the long-term trend in ion temperature, which should be similar to the trend in neutral temperature, due to close thermal coupling between neutrals and ions. Zhang et al. (2011) showed that a warming trend is found between 110 and 200 km altitude, and a cooling trend above 200 km, which increases with height and is stronger for lower solar activity. At 375 km the trend is as strong as -47±11 K/decade (Holt & Zhang, 2008). Donaldson et al. (2010), in a similar analysis of incoherent scatter

Climate Change in the Upper Atmosphere 319

Fig. 2. Trends in hmF2 for May-June-July at noon hours (10-14 LT) colour-coded for strength.

km/decade, green: -0.2 to +0.2 km/decade, yellow: +0.2 to + 1 km/decade, orange: +1 to +5 km/decade, red: >5 km/decade. Filled circles represent trends that are larger than the standard deviation and were calculated based on at least 80 data points. They are therefore the most reliable. All other trends are represented by open circles. Trends were calculated

Some studies have found regional patterns in F2 layer trends. Danilov & Mikhailov (1999) found that the magnitude of trends in foF2 increases with magnetic latitude. However, Bremer (1998) argued against this, and instead found a dependence on longitude: trends in both hmF2 and foF2 were in general negative west of 30°E, but positive east of 30°E. Jarvis (2009) confirmed this finding. Bencze (2007, 2009) found that ionospheric stations showing negative trends in hmF2 are mostly located near seashores, while stations in continental areas

The CO2 concentration in the Earth's atmosphere has increased from its estimated level of 280 ppm before the industrial revolution to 317 ppm in 1960 and 390 ppm in 2010 as measured at ground level in Mauna Loa, Hawaii (P. Tans & R. Keeling, data obtained from www.esrl.noaa.gov/gmd/ccgg/trends, June 2011). The concentration of methane has also increased, from 1151 ppb in 1985 to 1355 ppb in 2008 as observed in the lower stratosphere (Rinsland et al., 2009). Partly as a result of the increase in methane, stratospheric water vapour has increased as well, by 2 ppmv from 1945 to 2000, with its present-day level at

CO2, methane and water vapour all act as greenhouse gases. In the troposphere, greenhouse gases absorb outgoing infrared radiation, and emit it back to the surface, which results in warming. In the upper atmosphere they have the opposite effect. Greenhouse gases cool the

Dark blue: <-5 km/decade, medium blue: -5 to -1 km/decade, light blue: -1 to -0.2

using the Shimazaki (1955) formula by Th. Ulich (pers. comm., 2007).

usually exhibit positive trends.

**3.1.1 Greenhouse gases** 

**3. Greenhouse gases and ozone 3.1 Background and mechanisms** 

4-6 ppmv (Rosenlof et al., 2001; Hurst et al., 2011).

data from Saint Santin (44.1°N, 2.3°E), also found a warming at 120-130 km of ~10 K/decade, and a cooling of -30 K/decade at 350 km, increasing with height. They also noted that negative trends were much larger at noon (up to -60 K/decade) than at midnight (nearly zero). For both locations trends were larger for the period from ~1980 onward.

## **2.3 Density**

Because the atmosphere is nearly in hydrostatic equilibrium, a decrease in temperature results in contraction of the upper atmosphere, and a downward displacement of constant pressure surfaces. This means that the thermospheric density at a fixed height is expected to decrease. The same principle also explains why a warming is found in the lower thermosphere: as the thermosphere contracts the temperature profile essentially moves down. This results in an apparent warming at fixed height in the lower thermosphere due to the strong positive vertical gradient in temperature there (Akmaev & Fomichev, 1998).

Information about thermospheric density can be obtained from the orbits of near-Earth objects. Keating et al. (2000), Emmert et al. (2004, 2008), Emmert & Picone (2011), Marcos et al. (2005) and Saunders et al. (2011) used this technique and confirmed that the thermospheric density at fixed height is indeed decreasing. Observed trend magnitudes range from about -2%/decade to about -6%/decade, and generally increase with altitude. Trends are again larger for solar minimum conditions than for solar maximum conditions (Emmert et al., 2004). Emmert et al. (2008) reported that there is also a dependence on season, with trends being strongest in October and weakest in January and February. No dependence on local time or latitude was found.

## **2.4 Ionospheric trends**

Another consequence of atmospheric contraction as a result of global cooling is that ionospheric layers are expected to move downward, as they tend to stay on the same pressure level. On the other hand, ion and electron densities should not be affected much as a result of global cooling, as both ion production rates and recombination coefficients are expected to be affected in a similar way (Rishbeth, 1990). Any increase in ion production should therefore mostly be offset by a similar increase in ion loss. Still, both long-term changes in the height of ionospheric layers and in their critical frequencies have been found.

Trends in the F2 layer vary strongly with location, season, and local time. Bremer et al. (2004) reported median trends in foF2 and hmF2, derived from over 50 ionospheric stations, of -0.01 MHz/decade and -0.09 km/decade, respectively. However, local trends tend to be much stronger and can be either positive or negative. Figure 2 shows trends in hmF2 at noon for May-June-July, colour-coded by strength, and demonstrates that trends of the order of ±1-5 km/decade are the most common of the trends that are considered reliable (filled circles). Both trends in hmF2 and foF2 depend on the season and time of day (even switching sign), but this dependence is different for different stations (e.g. Elias & Ortiz de Adler, 2006). Similarly, no consistent dependence on solar activity level has been found (based on calculations by Th. Ulich, pers. comm., 2007).

data from Saint Santin (44.1°N, 2.3°E), also found a warming at 120-130 km of ~10 K/decade, and a cooling of -30 K/decade at 350 km, increasing with height. They also noted that negative trends were much larger at noon (up to -60 K/decade) than at midnight (nearly zero). For both locations trends were larger for the period from ~1980

Because the atmosphere is nearly in hydrostatic equilibrium, a decrease in temperature results in contraction of the upper atmosphere, and a downward displacement of constant pressure surfaces. This means that the thermospheric density at a fixed height is expected to decrease. The same principle also explains why a warming is found in the lower thermosphere: as the thermosphere contracts the temperature profile essentially moves down. This results in an apparent warming at fixed height in the lower thermosphere due to the strong positive vertical gradient in temperature there (Akmaev & Fomichev, 1998).

Information about thermospheric density can be obtained from the orbits of near-Earth objects. Keating et al. (2000), Emmert et al. (2004, 2008), Emmert & Picone (2011), Marcos et al. (2005) and Saunders et al. (2011) used this technique and confirmed that the thermospheric density at fixed height is indeed decreasing. Observed trend magnitudes range from about -2%/decade to about -6%/decade, and generally increase with altitude. Trends are again larger for solar minimum conditions than for solar maximum conditions (Emmert et al., 2004). Emmert et al. (2008) reported that there is also a dependence on season, with trends being strongest in October and weakest in January and February. No

Another consequence of atmospheric contraction as a result of global cooling is that ionospheric layers are expected to move downward, as they tend to stay on the same pressure level. On the other hand, ion and electron densities should not be affected much as a result of global cooling, as both ion production rates and recombination coefficients are expected to be affected in a similar way (Rishbeth, 1990). Any increase in ion production should therefore mostly be offset by a similar increase in ion loss. Still, both long-term changes in the height of ionospheric layers and in their critical frequencies have

Trends in the F2 layer vary strongly with location, season, and local time. Bremer et al. (2004) reported median trends in foF2 and hmF2, derived from over 50 ionospheric stations, of -0.01 MHz/decade and -0.09 km/decade, respectively. However, local trends tend to be much stronger and can be either positive or negative. Figure 2 shows trends in hmF2 at noon for May-June-July, colour-coded by strength, and demonstrates that trends of the order of ±1-5 km/decade are the most common of the trends that are considered reliable (filled circles). Both trends in hmF2 and foF2 depend on the season and time of day (even switching sign), but this dependence is different for different stations (e.g. Elias & Ortiz de Adler, 2006). Similarly, no consistent dependence on solar activity level has been found (based on

dependence on local time or latitude was found.

calculations by Th. Ulich, pers. comm., 2007).

**2.4 Ionospheric trends** 

been found.

onward.

**2.3 Density** 

Fig. 2. Trends in hmF2 for May-June-July at noon hours (10-14 LT) colour-coded for strength. Dark blue: <-5 km/decade, medium blue: -5 to -1 km/decade, light blue: -1 to -0.2 km/decade, green: -0.2 to +0.2 km/decade, yellow: +0.2 to + 1 km/decade, orange: +1 to +5 km/decade, red: >5 km/decade. Filled circles represent trends that are larger than the standard deviation and were calculated based on at least 80 data points. They are therefore the most reliable. All other trends are represented by open circles. Trends were calculated using the Shimazaki (1955) formula by Th. Ulich (pers. comm., 2007).

Some studies have found regional patterns in F2 layer trends. Danilov & Mikhailov (1999) found that the magnitude of trends in foF2 increases with magnetic latitude. However, Bremer (1998) argued against this, and instead found a dependence on longitude: trends in both hmF2 and foF2 were in general negative west of 30°E, but positive east of 30°E. Jarvis (2009) confirmed this finding. Bencze (2007, 2009) found that ionospheric stations showing negative trends in hmF2 are mostly located near seashores, while stations in continental areas usually exhibit positive trends.

## **3. Greenhouse gases and ozone**

#### **3.1 Background and mechanisms**

#### **3.1.1 Greenhouse gases**

The CO2 concentration in the Earth's atmosphere has increased from its estimated level of 280 ppm before the industrial revolution to 317 ppm in 1960 and 390 ppm in 2010 as measured at ground level in Mauna Loa, Hawaii (P. Tans & R. Keeling, data obtained from www.esrl.noaa.gov/gmd/ccgg/trends, June 2011). The concentration of methane has also increased, from 1151 ppb in 1985 to 1355 ppb in 2008 as observed in the lower stratosphere (Rinsland et al., 2009). Partly as a result of the increase in methane, stratospheric water vapour has increased as well, by 2 ppmv from 1945 to 2000, with its present-day level at 4-6 ppmv (Rosenlof et al., 2001; Hurst et al., 2011).

CO2, methane and water vapour all act as greenhouse gases. In the troposphere, greenhouse gases absorb outgoing infrared radiation, and emit it back to the surface, which results in warming. In the upper atmosphere they have the opposite effect. Greenhouse gases cool the

Climate Change in the Upper Atmosphere 321

the meridional wind is dominant. Qian et al. (2009) have shown that this mechanism may be responsible for some of the spatial variation in trends in hmF2, as discussed in section 3.2.3.

Fig. 3. Schematic illustration of the ionospheric plasma motion (red arrows) along a magnetic field line (black line with arrow) induced by horizontal neutral winds (blue arrows) blowing equatorward. In this case the ionospheric plasma is forced up magnetic field lines, resulting in an increase in hmF2. The opposite is true in case of poleward neutral winds. Note that the vertical component of the plasma motion becomes smaller near the equator, where the magnetic field makes a smaller angle with the Earth's surface.

Roble & Dickinson (1989) were the first to quantify the effect of an increase in greenhouse gases on the upper atmosphere. They used a 1-D model of the upper atmosphere to show that a doubling of the CO2 and methane concentration would cause a cooling of the thermosphere of up to 50 K. Since this initial pioneering study, many more have followed,

Most modelling studies employed a doubling of the CO2 concentration, like Roble & Dickinson (1989). This can make it somewhat difficult to make direct quantitative comparisons between observed and modelled trends, as a doubling has not yet occurred. When variables depend more or less linearly on the CO2 concentration, such as the thermospheric temperature or hmF2 (Cnossen, 2009; Cnossen et al., 2009), modelling results can be linearly interpolated. However, the response in thermospheric density decreases with increasing CO2 concentration, so that linear interpolation results in an underestimate of

Akmaev & Fomichev (2000) avoided such problems by performing simulations with the Spectral Mesosphere/Lower Thermosphere Model (SMLTM; Akmaev et al., 1992), using the CO2 concentrations of 1955 and 1995 (313 and 360 ppm, respectively). The global mean vertical profile of the thermal response they found was in qualitative agreement with observations, showing cooling in the mesosphere, little change near the mesopause, a slight warming in the lower thermosphere, and a cooling from 120-125 km that increased with height. However, their mesospheric cooling was about -0.8 K/decade, ~3 times smaller than observed. Also the thermospheric cooling was smaller than observed, and the turning point

Akmaev et al. (2006) therefore performed additional model simulations with the SMLTM which included also changes in water vapour and ozone concentration. The inclusion of

from warming to cooling in the lower thermosphere occurred at too low altitude.

using increasingly sophisticated numerical models of the upper atmosphere.

**3.2 Effect estimates and comparison to observations** 

**3.2.1 Temperature** 

trends (Cnossen, 2009).

upper atmosphere, because they emit infrared radiation mostly to space upon relaxation from an excited state induced by collisions. This way they act to remove thermal energy from the upper atmosphere. This process overcomes the effects of extra absorption of radiation in the lower atmosphere, resulting in a net cooling effect. CO2 is by far the most important contributor to infrared cooling in the middle atmosphere, followed by water vapour, while methane makes only a very small contribution (e.g. Fomichev, 2009). In the thermosphere both cooling by CO2 and by nitric oxide (NO) is important.

## **3.1.2 Ozone**

Ozone is also an important gas for the radiative budget of the stratosphere and mesosphere. Ozone absorbs solar ultraviolet (UV) radiation in the Chappuis (450-750 nm), Huggins (310- 360 nm), and Hartley (450-750 nm) bands, which results in heating. The maximum heating occurs near the stratopause. Ozone also contributes to infrared cooling, but this is a much smaller effect (e.g. Fomichev, 2009).

The total ozone concentration decreased from 375 Dobson units (DU) in 1970 to 325 DU in 2000 as measured in Switzerland, with much stronger decreases occurring over the polar regions (Solomon, 1999). While the downward trend in ozone concentration may have slowed or even reversed in recent years (Stolarski & Frith, 2006; Salby et al., 2011), any data from the 1970s to the present-day will contain the effects that may have arisen from the original decrease in ozone concentration. A decrease in ozone results in less heating (i.e. cooling), and so acts in the same way as an increase in greenhouse gases. Note that while ozone is most important in the stratosphere and mesosphere, the cooling influence may still be felt in the thermosphere through the effect of atmospheric contraction.

## **3.1.3 Indirect effects**

Once changes in temperature occur, whether they are caused by changes in greenhouse gases, ozone, or another process, this induces further indirect effects. We already noted that global cooling leads to thermal contraction, which has consequences for the density at a fixed height and the vertical electron density distribution. Changes in horizontal temperature structure can lead to changes in dynamics through changes in pressure gradients. The changes in dynamics can then cause further changes in temperature structure, and so on.

Not only the neutral atmosphere can be affected this way; ionospheric plasma transport is affected by changes in the neutral winds via ion-neutral collisions. Because the ionospheric plasma tends to be "frozen-in" to the magnetic field lines, meaning that it tends to flow along magnetic field lines, horizontal neutral winds can induce a vertical component in the plasma motion due to the inclination of magnetic field lines, as shown schematically in figure 3 (see also Rishbeth, 1998; Cnossen & Richmond, 2008). Due to the configuration of the Earth's magnetic field, ionospheric plasma is forced up magnetic field lines when neutral winds blow towards the magnetic equator, which acts to increase hmF2, and vice versa when neutral winds blow poleward. Because the geographic equator does not exactly coincide with the magnetic equator, not only the meridional wind, but also the zonal wind can contribute to the plasma transport taking place this way, although the contribution from

upper atmosphere, because they emit infrared radiation mostly to space upon relaxation from an excited state induced by collisions. This way they act to remove thermal energy from the upper atmosphere. This process overcomes the effects of extra absorption of radiation in the lower atmosphere, resulting in a net cooling effect. CO2 is by far the most important contributor to infrared cooling in the middle atmosphere, followed by water vapour, while methane makes only a very small contribution (e.g. Fomichev, 2009). In the

Ozone is also an important gas for the radiative budget of the stratosphere and mesosphere. Ozone absorbs solar ultraviolet (UV) radiation in the Chappuis (450-750 nm), Huggins (310- 360 nm), and Hartley (450-750 nm) bands, which results in heating. The maximum heating occurs near the stratopause. Ozone also contributes to infrared cooling, but this is a much

The total ozone concentration decreased from 375 Dobson units (DU) in 1970 to 325 DU in 2000 as measured in Switzerland, with much stronger decreases occurring over the polar regions (Solomon, 1999). While the downward trend in ozone concentration may have slowed or even reversed in recent years (Stolarski & Frith, 2006; Salby et al., 2011), any data from the 1970s to the present-day will contain the effects that may have arisen from the original decrease in ozone concentration. A decrease in ozone results in less heating (i.e. cooling), and so acts in the same way as an increase in greenhouse gases. Note that while ozone is most important in the stratosphere and mesosphere, the cooling influence may still

Once changes in temperature occur, whether they are caused by changes in greenhouse gases, ozone, or another process, this induces further indirect effects. We already noted that global cooling leads to thermal contraction, which has consequences for the density at a fixed height and the vertical electron density distribution. Changes in horizontal temperature structure can lead to changes in dynamics through changes in pressure gradients. The changes in dynamics can then cause further changes in temperature

Not only the neutral atmosphere can be affected this way; ionospheric plasma transport is affected by changes in the neutral winds via ion-neutral collisions. Because the ionospheric plasma tends to be "frozen-in" to the magnetic field lines, meaning that it tends to flow along magnetic field lines, horizontal neutral winds can induce a vertical component in the plasma motion due to the inclination of magnetic field lines, as shown schematically in figure 3 (see also Rishbeth, 1998; Cnossen & Richmond, 2008). Due to the configuration of the Earth's magnetic field, ionospheric plasma is forced up magnetic field lines when neutral winds blow towards the magnetic equator, which acts to increase hmF2, and vice versa when neutral winds blow poleward. Because the geographic equator does not exactly coincide with the magnetic equator, not only the meridional wind, but also the zonal wind can contribute to the plasma transport taking place this way, although the contribution from

thermosphere both cooling by CO2 and by nitric oxide (NO) is important.

be felt in the thermosphere through the effect of atmospheric contraction.

**3.1.2 Ozone** 

smaller effect (e.g. Fomichev, 2009).

**3.1.3 Indirect effects** 

structure, and so on.

the meridional wind is dominant. Qian et al. (2009) have shown that this mechanism may be responsible for some of the spatial variation in trends in hmF2, as discussed in section 3.2.3.

Fig. 3. Schematic illustration of the ionospheric plasma motion (red arrows) along a magnetic field line (black line with arrow) induced by horizontal neutral winds (blue arrows) blowing equatorward. In this case the ionospheric plasma is forced up magnetic field lines, resulting in an increase in hmF2. The opposite is true in case of poleward neutral winds. Note that the vertical component of the plasma motion becomes smaller near the equator, where the magnetic field makes a smaller angle with the Earth's surface.

#### **3.2 Effect estimates and comparison to observations**

## **3.2.1 Temperature**

Roble & Dickinson (1989) were the first to quantify the effect of an increase in greenhouse gases on the upper atmosphere. They used a 1-D model of the upper atmosphere to show that a doubling of the CO2 and methane concentration would cause a cooling of the thermosphere of up to 50 K. Since this initial pioneering study, many more have followed, using increasingly sophisticated numerical models of the upper atmosphere.

Most modelling studies employed a doubling of the CO2 concentration, like Roble & Dickinson (1989). This can make it somewhat difficult to make direct quantitative comparisons between observed and modelled trends, as a doubling has not yet occurred. When variables depend more or less linearly on the CO2 concentration, such as the thermospheric temperature or hmF2 (Cnossen, 2009; Cnossen et al., 2009), modelling results can be linearly interpolated. However, the response in thermospheric density decreases with increasing CO2 concentration, so that linear interpolation results in an underestimate of trends (Cnossen, 2009).

Akmaev & Fomichev (2000) avoided such problems by performing simulations with the Spectral Mesosphere/Lower Thermosphere Model (SMLTM; Akmaev et al., 1992), using the CO2 concentrations of 1955 and 1995 (313 and 360 ppm, respectively). The global mean vertical profile of the thermal response they found was in qualitative agreement with observations, showing cooling in the mesosphere, little change near the mesopause, a slight warming in the lower thermosphere, and a cooling from 120-125 km that increased with height. However, their mesospheric cooling was about -0.8 K/decade, ~3 times smaller than observed. Also the thermospheric cooling was smaller than observed, and the turning point from warming to cooling in the lower thermosphere occurred at too low altitude.

Akmaev et al. (2006) therefore performed additional model simulations with the SMLTM which included also changes in water vapour and ozone concentration. The inclusion of

Climate Change in the Upper Atmosphere 323

between the Akmaev et al. (2006) and Cnossen (2009) model estimates and observed trends in the middle atmosphere and lower thermosphere (25-120 km). However, at 130- 180 km, most observations show a strong apparent warming, which is not reproduced by the models. Also, the strong cooling observed higher up in the thermosphere (>200 km) is underestimated by the modelling studies by Cnossen (2009) and Qian & Solomon (2011). Both studies predict a cooling that remains more or less constant with increasing altitude (>200 km), while observations indicate a strengthening of the trend with increasing

Akmaev et al. (2006) argued that the lack of a strong warming in the lower thermosphere in their results might be due to an overestimate of the amount of CO2 present in the upper mesosphere and lower thermosphere in the SMLTM. Still, this does not explain why too little cooling is predicted by modelling studies above 200 km. Although the observed thermospheric temperature trends are based on data from just two sites, it does appear that

The same modelling studies described in the previous section also provided trends in density, which are compared to each other and to observations in figure 5. The density trends modelled by Akmaev et al. (2006) are about twice as strong as those modelled by Cnossen (2009). However, there is also a large spread in observed trends. At 200 km, the Akmaev et al. (2006) trends seem to agree quite well with trends observed by Saunders et al. (2011), while the Cnossen (2009) trends agree better with Emmert & Picone (2011) and Emmert et al. (2004). The trends modelled by Qian & Solomon (2011) at higher altitude also tend to agree best with Emmert & Picone (2011) and Emmert et al. (2004), as well as Marcos

Depending on which observations and modelling results are chosen for comparison, it may be argued that trends in density show better agreement between models and observations than trends in temperature. However, this may be partly so because the density at a certain height responds to the temperature structure of the entire atmosphere below. That means that the effects of insufficient warming below 200 km and insufficient cooling above 200 km

The modelled seasonal dependence of trends in density does not match with observations. Emmert et al. (2008) found that density trends are weakest in January and February, while the modelled trends by Akmaev et al. (2006) are stronger for January than for March. Changes in CO2 and ozone concentration can therefore not explain the observed seasonal

There is however an explanation for the dependence of trends in density and temperature on the solar activity level. Qian et al. (2006) showed that stronger trends during solar minimum are due to CO2 cooling being a relatively more important cooling mechanism at solar minimum, while cooling by nitric oxide (NO) becomes more important at solar maximum. A change in CO2 concentration therefore has a larger impact on the radiative balance in the upper atmosphere during solar minimum than during solar maximum.

changes in CO2 and ozone concentration cannot fully explain those trends.

could to some extent cancel each other out at altitudes above 200 km.

altitude.

**3.2.2 Density** 

et al. (2005).

dependence.

ozone resulted in a much stronger modelled cooling in the mesosphere of nearly 2 K/decade, almost as strong as the observed trend. The addition of ozone, and to a lesser extent water vapour, also produced a more pronounced warming in the lower thermosphere of up to 3 K/decade at 115 km, which turned into a cooling again above 150 km.

Fig. 4. Observed and modelled temperature trends (dots) with error bars where available (lines). Note that modelling results are in black/grey while observational results are in colour. For Akmaev et al. (2006) both January and March results are shown (stronger trends are for January) and for Cnossen (2009) both March and June results are shown (stronger trends are for June, though there is only a small difference). All other results are seasonal averages.

Cnossen (2009) obtained results similar to those of Akmaev et al. (2006) using simulations with the Coupled Middle Atmosphere and Thermosphere Model 2 (CMAT2; Harris, 2006), studying the combined effects of changes in CO2 and ozone concentration between 1965 and 1995. Qian et al. (2006) and Qian & Solomon (2011) used a global mean model to investigate trends higher up in the thermosphere, but considered only changes in the CO2 concentration from 1970 to 2000.

The temperature trends calculated by Akmaev et al. (2006), Cnossen (2009) and Qian & Solomon (2011) are compared to observed trends in figure 4. There is good agreement between the Akmaev et al. (2006) and Cnossen (2009) model estimates and observed trends in the middle atmosphere and lower thermosphere (25-120 km). However, at 130- 180 km, most observations show a strong apparent warming, which is not reproduced by the models. Also, the strong cooling observed higher up in the thermosphere (>200 km) is underestimated by the modelling studies by Cnossen (2009) and Qian & Solomon (2011). Both studies predict a cooling that remains more or less constant with increasing altitude (>200 km), while observations indicate a strengthening of the trend with increasing altitude.

Akmaev et al. (2006) argued that the lack of a strong warming in the lower thermosphere in their results might be due to an overestimate of the amount of CO2 present in the upper mesosphere and lower thermosphere in the SMLTM. Still, this does not explain why too little cooling is predicted by modelling studies above 200 km. Although the observed thermospheric temperature trends are based on data from just two sites, it does appear that changes in CO2 and ozone concentration cannot fully explain those trends.

## **3.2.2 Density**

322 Greenhouse Gases – Emission, Measurement and Management

ozone resulted in a much stronger modelled cooling in the mesosphere of nearly 2 K/decade, almost as strong as the observed trend. The addition of ozone, and to a lesser extent water vapour, also produced a more pronounced warming in the lower thermosphere of up to 3 K/decade at 115 km, which turned into a cooling again above

Fig. 4. Observed and modelled temperature trends (dots) with error bars where available (lines). Note that modelling results are in black/grey while observational results are in colour. For Akmaev et al. (2006) both January and March results are shown (stronger trends are for January) and for Cnossen (2009) both March and June results are shown (stronger trends are for June, though there is only a small difference). All other results are seasonal

Cnossen (2009) obtained results similar to those of Akmaev et al. (2006) using simulations with the Coupled Middle Atmosphere and Thermosphere Model 2 (CMAT2; Harris, 2006), studying the combined effects of changes in CO2 and ozone concentration between 1965 and 1995. Qian et al. (2006) and Qian & Solomon (2011) used a global mean model to investigate trends higher up in the thermosphere, but considered only changes in the CO2 concentration

The temperature trends calculated by Akmaev et al. (2006), Cnossen (2009) and Qian & Solomon (2011) are compared to observed trends in figure 4. There is good agreement

150 km.

averages.

from 1970 to 2000.

The same modelling studies described in the previous section also provided trends in density, which are compared to each other and to observations in figure 5. The density trends modelled by Akmaev et al. (2006) are about twice as strong as those modelled by Cnossen (2009). However, there is also a large spread in observed trends. At 200 km, the Akmaev et al. (2006) trends seem to agree quite well with trends observed by Saunders et al. (2011), while the Cnossen (2009) trends agree better with Emmert & Picone (2011) and Emmert et al. (2004). The trends modelled by Qian & Solomon (2011) at higher altitude also tend to agree best with Emmert & Picone (2011) and Emmert et al. (2004), as well as Marcos et al. (2005).

Depending on which observations and modelling results are chosen for comparison, it may be argued that trends in density show better agreement between models and observations than trends in temperature. However, this may be partly so because the density at a certain height responds to the temperature structure of the entire atmosphere below. That means that the effects of insufficient warming below 200 km and insufficient cooling above 200 km could to some extent cancel each other out at altitudes above 200 km.

The modelled seasonal dependence of trends in density does not match with observations. Emmert et al. (2008) found that density trends are weakest in January and February, while the modelled trends by Akmaev et al. (2006) are stronger for January than for March. Changes in CO2 and ozone concentration can therefore not explain the observed seasonal dependence.

There is however an explanation for the dependence of trends in density and temperature on the solar activity level. Qian et al. (2006) showed that stronger trends during solar minimum are due to CO2 cooling being a relatively more important cooling mechanism at solar minimum, while cooling by nitric oxide (NO) becomes more important at solar maximum. A change in CO2 concentration therefore has a larger impact on the radiative balance in the upper atmosphere during solar minimum than during solar maximum.

Climate Change in the Upper Atmosphere 325

changes in dynamics, and Qian et al. (2009) found indeed that changes in neutral winds were smaller at solar maximum. The smaller changes found under solar maximum conditions could again be linked to the smaller contribution to the radiative budget of CO2

The changes in hmF2 modelled by Qian et al. (2009) were mostly negative, as expected, but they were occasionally positive for solar minimum, usually after midnight. To compare the magnitude of the changes they found to observed trends, figure 6 shows trends in hmF2 and foF2 in km/decade and MHz/decade, linearly interpolated from their results for 3 LT. The interpolation was done based on the actual change in CO2 concentration between 1960 and 2010 (73 ppm; 14.6 ppm/decade), so that their original numbers were divided by 25

Fig. 6. Trends in hmF2 in km/decade (left) and foF2 in MHz/decade (right) at 3 LT linearly

A typical interpolated trend in hmF2 is -1 km/decade, while maximum negative trends go up to about -2 km/decade (at 12 LT; not shown) and maximum positive trends up to 0.5 km/decade. These values are somewhat larger than the median change in hmF2 reported by Bremer et al. (2004), but 2-5 times smaller than typical changes at individual stations. Modelled trends for solar maximum conditions are even smaller. Estimated trends in foF2 vary between 0 and -0.04 MHz/decade. The strongest trend of -0.04 MHz/decade is also larger than the median change in foF2 of -0.01 MHz/decade reported by Bremer et al. (2004), but again underestimates changes observed at many individual stations. Positive trends that have been reported in some locations are not explained at all. From these results we can conclude that the change in CO2 concentration has contributed to long-term trends in the F2

The Earth's magnetic field varies slowly in strength and in orientation over time, and also the relative contributions of the main dipole component and higher order field components vary. We are currently at a time of relatively strong change in terms of field intensity, as the Earth's dipole moment has decreased by about 5% per century since 1840, while little

interpolated from the model simulations by Qian et al. (2009) at June solstice.

layer, but it is unlikely to be the sole cause.

**4.1 Background and mechanism** 

**4. Secular variation of the Earth's magnetic field** 

cooling relative to NO cooling at solar maximum (Qian et al., 2006).

(365/14.6).

Fig. 5. Observed and modelled density trends (dots) with error bars where available (lines) Note that modelling results are in black/grey while observational results are in colour. For Akmaev et al. (2006) both January and March results are shown (stronger trends are for January) and for Cnossen (2009) both March and June results are shown (stronger trends are for June). All other results are seasonal averages.

## **3.2.3 The ionospheric F2 layer**

Qian et al. (2009) performed simulations with the Thermosphere-Ionosphere-Electrodynamics General Circulation Model (TIE-GCM; Roble et al., 1998; Richmond et al., 1992), focusing more on the effects of CO2 cooling on the F2 layer ionosphere. They doubled the CO2 concentration from 365 ppmv to 730 ppmv and noted that certain features of the changes in NmF2 and hmF2 followed the magnetic equator, indicating the role of electrodynamics in generating these trends.

Qian et al. (2009) showed that changes in both the meridional and zonal neutral wind affected the transport of ionospheric plasma by neutral winds, through the mechanism described in section 3.1.3. This was an important contributor to changes in hmF2, especially under solar minimum conditions. At solar maximum, they found that the changes in NmF2 matched quite well with the pattern of changes in the O/N2 ratio, which is proportional to the balance of ion production and loss rates. This result indicated a less important role for

Fig. 5. Observed and modelled density trends (dots) with error bars where available (lines) Note that modelling results are in black/grey while observational results are in colour. For Akmaev et al. (2006) both January and March results are shown (stronger trends are for January) and for Cnossen (2009) both March and June results are shown (stronger trends are

Qian et al. (2009) performed simulations with the Thermosphere-Ionosphere-Electrodynamics General Circulation Model (TIE-GCM; Roble et al., 1998; Richmond et al., 1992), focusing more on the effects of CO2 cooling on the F2 layer ionosphere. They doubled the CO2 concentration from 365 ppmv to 730 ppmv and noted that certain features of the changes in NmF2 and hmF2 followed the magnetic equator, indicating the role of

Qian et al. (2009) showed that changes in both the meridional and zonal neutral wind affected the transport of ionospheric plasma by neutral winds, through the mechanism described in section 3.1.3. This was an important contributor to changes in hmF2, especially under solar minimum conditions. At solar maximum, they found that the changes in NmF2 matched quite well with the pattern of changes in the O/N2 ratio, which is proportional to the balance of ion production and loss rates. This result indicated a less important role for

for June). All other results are seasonal averages.

electrodynamics in generating these trends.

**3.2.3 The ionospheric F2 layer** 

changes in dynamics, and Qian et al. (2009) found indeed that changes in neutral winds were smaller at solar maximum. The smaller changes found under solar maximum conditions could again be linked to the smaller contribution to the radiative budget of CO2 cooling relative to NO cooling at solar maximum (Qian et al., 2006).

The changes in hmF2 modelled by Qian et al. (2009) were mostly negative, as expected, but they were occasionally positive for solar minimum, usually after midnight. To compare the magnitude of the changes they found to observed trends, figure 6 shows trends in hmF2 and foF2 in km/decade and MHz/decade, linearly interpolated from their results for 3 LT. The interpolation was done based on the actual change in CO2 concentration between 1960 and 2010 (73 ppm; 14.6 ppm/decade), so that their original numbers were divided by 25 (365/14.6).

Fig. 6. Trends in hmF2 in km/decade (left) and foF2 in MHz/decade (right) at 3 LT linearly interpolated from the model simulations by Qian et al. (2009) at June solstice.

A typical interpolated trend in hmF2 is -1 km/decade, while maximum negative trends go up to about -2 km/decade (at 12 LT; not shown) and maximum positive trends up to 0.5 km/decade. These values are somewhat larger than the median change in hmF2 reported by Bremer et al. (2004), but 2-5 times smaller than typical changes at individual stations. Modelled trends for solar maximum conditions are even smaller. Estimated trends in foF2 vary between 0 and -0.04 MHz/decade. The strongest trend of -0.04 MHz/decade is also larger than the median change in foF2 of -0.01 MHz/decade reported by Bremer et al. (2004), but again underestimates changes observed at many individual stations. Positive trends that have been reported in some locations are not explained at all. From these results we can conclude that the change in CO2 concentration has contributed to long-term trends in the F2 layer, but it is unlikely to be the sole cause.

## **4. Secular variation of the Earth's magnetic field**

#### **4.1 Background and mechanism**

The Earth's magnetic field varies slowly in strength and in orientation over time, and also the relative contributions of the main dipole component and higher order field components vary. We are currently at a time of relatively strong change in terms of field intensity, as the Earth's dipole moment has decreased by about 5% per century since 1840, while little

Climate Change in the Upper Atmosphere 327

the inclination of the magnetic field changed the most. The changes in inclination were the dominant cause of the trends in the F2 layer, by changing plasma transport up and down magnetic field lines driven by neutral winds. Neutral winds also changed somewhat, but no

Fig. 7. Trends in hmF2 in km/decade (left) and foF2 in MHz/decade (right) at 12 LT at June

Figure 7 shows the average trends in hmF2 and foF2 in km/decade and MHz/decade, respectively, linearly interpolated from the change between 1957 and 1997. Over South America and the southern Atlantic Ocean, modelled trends in hmF2 and foF2 are of the order of -5 to +2 km/decade and ±0.1 MHz/decade, respectively. Trends show a strong spatial

The magnitudes of the trends due to magnetic field changes in the regions that are most strongly affected are quite similar to the magnitudes of typical observed trends, and 2-5 times larger than the maximum changes predicted to be caused by changes in the CO2 concentration. Still, a direct comparison between modelled and observed trends at specific stations indicated that modelled trends were usually smaller. At Concepción (36.8°S, 73.0°W) and Argentine Islands (65.2°S, 64.3°W), both located in the region where modelled trends were strongest, changes in the magnetic field could account for 30-50% and 20% of the observed trends. The observed and modelled seasonal-diurnal patterns of trends for

The predicted changes outside the region of South America and the southern Atlantic Ocean were very small. However, the spatial pattern of modelled trends, showing a marked change at a longitude of 0-20°, with weak trends east of that line, and much stronger trends to the west, was somewhat similar to the findings of Bremer (1998) and Jarvis (2009), although they found positive trends west of 30° and negative trends east of 30°. An increase of foF2 trends with geomagnetic latitude, as found by Danilov & Mikhailov (1999), was not

Based on the study by Cnossen & Richmond (2008) we can conclude that long-term changes in the magnetic field are important in some regions, but do not explain observed trends elsewhere. However, Cnossen & Richmond (2008) had to neglect any effects of changes in

solstice due to changes in the Earth's magnetic field, interpolated from the model

dependence, and also vary (even in sign) depending on the season and time of day.

simulations by Cnossen & Richmond (2008).

these stations did not match.

found.

significant effects on the thermospheric temperature were found.

change in field strength occurred from 1590 to 1840 (Gubbins et al., 2006). The angle between the geomagnetic dipole and the Earth's rotation axis, the tilt angle, has changed also over the last half century, from 11.7° in 1960 to 10.5° in 2005, following more than a century where it remained nearly constant (Amit & Olson, 2008).

The Earth's magnetic field plays a role in the state of the upper atmosphere in various ways. First of all, it is what shapes the Earth's magnetosphere, and controls to some extent the interactions of the magnetosphere with the solar wind and the ionosphere. This determines the flux and energy of energetic particles precipitating into the upper atmosphere, as well as the high-latitude electric field and ionospheric convection pattern (see e.g. Kivelson & Russell (1995) for further details).

Secondly, the orientation of the Earth's magnetic field influences the transport of ionospheric plasma by neutral winds. This process has been described in section 3.1.3 in the context of indirect effects of changes in circulation. However, even if there were no changes in the neutral wind, changes in the inclination and declination of the Earth's magnetic field could still cause changes in the plasma motion. The inclination is the angle between the magnetic field and the Earth's surface and the declination is the angle with geographic North. A change in the inclination *I* directly affects the vertical component of the plasma motion driven by neutral winds, which is equal to the component of the horizontal neutral wind parallel to the magnetic field, vn,par, times the factor sin(*I*)cos(*I*). A change in declination changes the magnitude of vn,par, as it changes the projection of the neutral wind onto magnetic field lines. The changes in magnetic field orientation also cause changes to the ion drag acting on the neutral wind, so that neutral winds are likely to change as well.

The neutral wind is also responsible for the generation of the dynamo electric field **E**dyn through **E**dyn = **v**nx**B**, where **vn** is the neutral wind and **B** the magnetic field. Both changes in the neutral wind and changes in declination and inclination alter the component of **v**<sup>n</sup> perpendicular to **B**. In addition, a change in the magnitude of **B** changes the ionospheric Pedersen and Hall conductivities, and these combined effects induce a change in the electrostatic field **E**. The change in **E** and **B** does not only cause changes to the ion drag exerted on the neutral wind, further modifying **v**n, but also modifies the **E**x**B** drift of ions and electrons. The vertical component of this drift can be expected to change hmF2 and foF2 too.

While the ionosphere is expected to be most directly affected by changes in the magnetic field, changes to the neutral atmosphere can occur as well. Possible effects on the neutral wind via ion-neutral collisions have already been mentioned, and as noted before, any changes in circulation can have secondary effects on for instance the neutral temperature structure. The neutral temperature in the thermosphere may also be directly affected through changes in Joule heating.

#### **4.2 Effect estimates and comparison to observations**

Cnossen & Richmond (2008) quantified the global effects of magnetic field changes on the ionospheric F2 layer using simulations with the Thermosphere-Ionosphere-Electrodynamics general circulation model (TIE-GCM). They found that changes in the Earth's magnetic field from 1957 to 1997 had a substantial effect on hmF2 and foF2 in some parts of the world, primarily South America and the southern Atlantic Ocean (see figure 7). In these regions,

change in field strength occurred from 1590 to 1840 (Gubbins et al., 2006). The angle between the geomagnetic dipole and the Earth's rotation axis, the tilt angle, has changed also over the last half century, from 11.7° in 1960 to 10.5° in 2005, following more than a

The Earth's magnetic field plays a role in the state of the upper atmosphere in various ways. First of all, it is what shapes the Earth's magnetosphere, and controls to some extent the interactions of the magnetosphere with the solar wind and the ionosphere. This determines the flux and energy of energetic particles precipitating into the upper atmosphere, as well as the high-latitude electric field and ionospheric convection pattern (see e.g. Kivelson &

Secondly, the orientation of the Earth's magnetic field influences the transport of ionospheric plasma by neutral winds. This process has been described in section 3.1.3 in the context of indirect effects of changes in circulation. However, even if there were no changes in the neutral wind, changes in the inclination and declination of the Earth's magnetic field could still cause changes in the plasma motion. The inclination is the angle between the magnetic field and the Earth's surface and the declination is the angle with geographic North. A change in the inclination *I* directly affects the vertical component of the plasma motion driven by neutral winds, which is equal to the component of the horizontal neutral wind parallel to the magnetic field, vn,par, times the factor sin(*I*)cos(*I*). A change in declination changes the magnitude of vn,par, as it changes the projection of the neutral wind onto magnetic field lines. The changes in magnetic field orientation also cause changes to the ion drag acting on the neutral wind, so that neutral winds are likely to change as well. The neutral wind is also responsible for the generation of the dynamo electric field **E**dyn through **E**dyn = **v**nx**B**, where **vn** is the neutral wind and **B** the magnetic field. Both changes in the neutral wind and changes in declination and inclination alter the component of **v**<sup>n</sup> perpendicular to **B**. In addition, a change in the magnitude of **B** changes the ionospheric Pedersen and Hall conductivities, and these combined effects induce a change in the electrostatic field **E**. The change in **E** and **B** does not only cause changes to the ion drag exerted on the neutral wind, further modifying **v**n, but also modifies the **E**x**B** drift of ions and electrons. The vertical component of this drift can be expected to change hmF2 and foF2

While the ionosphere is expected to be most directly affected by changes in the magnetic field, changes to the neutral atmosphere can occur as well. Possible effects on the neutral wind via ion-neutral collisions have already been mentioned, and as noted before, any changes in circulation can have secondary effects on for instance the neutral temperature structure. The neutral temperature in the thermosphere may also be directly affected

Cnossen & Richmond (2008) quantified the global effects of magnetic field changes on the ionospheric F2 layer using simulations with the Thermosphere-Ionosphere-Electrodynamics general circulation model (TIE-GCM). They found that changes in the Earth's magnetic field from 1957 to 1997 had a substantial effect on hmF2 and foF2 in some parts of the world, primarily South America and the southern Atlantic Ocean (see figure 7). In these regions,

century where it remained nearly constant (Amit & Olson, 2008).

Russell (1995) for further details).

through changes in Joule heating.

**4.2 Effect estimates and comparison to observations** 

too.

the inclination of the magnetic field changed the most. The changes in inclination were the dominant cause of the trends in the F2 layer, by changing plasma transport up and down magnetic field lines driven by neutral winds. Neutral winds also changed somewhat, but no significant effects on the thermospheric temperature were found.

Fig. 7. Trends in hmF2 in km/decade (left) and foF2 in MHz/decade (right) at 12 LT at June solstice due to changes in the Earth's magnetic field, interpolated from the model simulations by Cnossen & Richmond (2008).

Figure 7 shows the average trends in hmF2 and foF2 in km/decade and MHz/decade, respectively, linearly interpolated from the change between 1957 and 1997. Over South America and the southern Atlantic Ocean, modelled trends in hmF2 and foF2 are of the order of -5 to +2 km/decade and ±0.1 MHz/decade, respectively. Trends show a strong spatial dependence, and also vary (even in sign) depending on the season and time of day.

The magnitudes of the trends due to magnetic field changes in the regions that are most strongly affected are quite similar to the magnitudes of typical observed trends, and 2-5 times larger than the maximum changes predicted to be caused by changes in the CO2 concentration. Still, a direct comparison between modelled and observed trends at specific stations indicated that modelled trends were usually smaller. At Concepción (36.8°S, 73.0°W) and Argentine Islands (65.2°S, 64.3°W), both located in the region where modelled trends were strongest, changes in the magnetic field could account for 30-50% and 20% of the observed trends. The observed and modelled seasonal-diurnal patterns of trends for these stations did not match.

The predicted changes outside the region of South America and the southern Atlantic Ocean were very small. However, the spatial pattern of modelled trends, showing a marked change at a longitude of 0-20°, with weak trends east of that line, and much stronger trends to the west, was somewhat similar to the findings of Bremer (1998) and Jarvis (2009), although they found positive trends west of 30° and negative trends east of 30°. An increase of foF2 trends with geomagnetic latitude, as found by Danilov & Mikhailov (1999), was not found.

Based on the study by Cnossen & Richmond (2008) we can conclude that long-term changes in the magnetic field are important in some regions, but do not explain observed trends elsewhere. However, Cnossen & Richmond (2008) had to neglect any effects of changes in

Climate Change in the Upper Atmosphere 329

activity is unlikely to have contributed much to those (Laštovička, 2005). However, for other

Geomagnetic activity is a manifestation and measure of "space weather", arising from the interaction between the solar wind and the Earth's magnetosphere. This interaction generates currents in the ionosphere and magnetosphere which cause small perturbations to the main magnetic field of the Earth that can be measured at the surface. These magnetic perturbations form the basis of indices of geomagnetic activity, such as the Ap, aa, and Kp indices. Each of these indices is derived from the K index, which is related to the maximum fluctuations of the horizontal components of the observed geomagnetic field, relative to a quiet day (a day with few disturbances due to the solar wind), during a three-hour interval. Various studies have indicated the presence of a long-term trend in geomagnetic activity over the past 50-150 years. Clilverd et al. (1998) and Stamper et al. (2002) found that the geomagnetic activity in terms of the aa-index was increasing throughout the 20th century, and Clilverd et al. (2002) found that the number of geomagnetic storms per solar cycle had also been increasing until it stabilized in the last few cycles. Long-term changes in the solar quiet-time day variation, Sq, have been found as well. Macmillan & Droujinina (2007) reported a weak upward trend of on average 1.3 nT per century (10%) in the Sq amplitude based on records from 14 magnetic observatories. Elias et al. (2010) analysed data from three

High geomagnetic activity indicates disturbed space weather conditions, usually resulting from disturbances in the solar wind. This tends to lead to stronger energetic particle precipitation and ionospheric convection at high latitudes, stronger ionospheric currents, and an enhancement of the Joule heating in the upper atmosphere, eventually leading to higher temperatures and stronger neutral winds. Variations in geomagnetic activity occur on short timescales from minutes to days, and are responsible for part of the large natural variability in the thermosphere-ionosphere system. A gradual, long-term change in the background level of geomagnetic activity could therefore cause a long-term trend in the

Because geomagnetic activity is a measure of the interaction between the solar wind and the Earth's magnetosphere-ionosphere-thermosphere system, the origin of long-term changes in geomagnetic activity could be associated either with the Sun (e.g. Stamper et al., 1999) or with the terrestrial system (or both). Clilverd et al. (2002) concluded from a simple calculation that involved various assumptions and approximations that changes in the Earth's magnetic field would have little effect on geomagnetic activity. However, a detailed analysis has not been done yet, and the simulations by Cnossen et al. (2011) indicate that the effect may be larger than previously thought, due to a stronger dependence of the ionospheric conductance on the magnetic field strength than was assumed by Clilverd et al. (2002). When considering the effects of changes in geomagnetic activity on the upper atmosphere, it is therefore possible that these may ultimately be due to changes in the Sun and/or changes in the Earth's magnetic field. Nevertheless, the effect of the observed change in geomagnetic activity on the upper atmosphere, regardless of which processes may

time intervals (e.g. 1900-1950) it could be a contributor.

stations, and found also positive trends in Sq of 5-10%/century.

**5.2 Geomagnetic activity** 

upper atmosphere.

have contributed to it, can still be investigated.

the high-latitude coupling between the magnetosphere and ionosphere-thermosphere system that might arise, as their model did not include a magnetosphere. This could have resulted in an underestimation of the effects of magnetic field changes, especially at high latitudes.

Recently the TIE-GCM has been successfully coupled to the Lyon-Fedder-Mobarry (LFM) MHD code, forming the Coupled Magnetosphere-Ionosphere-Thermosphere (CMIT) model (Wiltberger et al., 2004; Wang et al., 2004, 2008). With this model it is now possible to include the effects of changes in high-latitude magnetosphere-ionosphere coupling, and Cnossen et al. (2011) have started to use this to re-assess the effects of changes in the magnetic field on the ionosphere and thermosphere.

Cnossen et al. (2011) started with a simplified experiment, investigating the effect of changes in magnetic field strength only. They reduced the magnetic field strength to 75% of its present-day value, which caused changes in hmF2 of up to -40 and +60 km and changes in foF2 of up to -3 to +1 MHz. If a linear response is assumed, considering a 5% decrease in magnetic field strength per century, these would be equivalent to trends of -0.8 to +1.2 km/decade and -0.06 to +0.02 MHz/decade. While clearly smaller than the trends predicted from the full magnetic field changes between 1957 and 1997, as expected, this rough estimate indicates that changes in magnetic field strength alone could still make a significant contribution to long-term trends. The trends in hmF2 and foF2 predicted in this way appear to be of similar order of magnitude as those due to changes in CO2 concentration.

The spatial pattern of changes simulated by CMIT is different from that simulated with TIE-GCM for the full magnetic field changes between 1957 and 1997, which is perhaps not surprising, as they were caused (predominantly) by different mechanisms. The new experiments show strong changes in hmF2 over Australia, the Pacific Ocean, and the northern Indian Ocean, and smaller changes over the Atlantic Ocean. This opens up the possibility that effects of changes in the magnetic field are present over different and more widespread areas than previously thought. A detailed comparison between the changes modelled by CMIT and observed trends will be done once CMIT simulations with historic magnetic fields have been performed, which are planned for the near future.

## **5. Changes in solar and geomagnetic activity**

## **5.1 Solar activity**

The solar activity level varies over an approximately 11-year cycle, and there are indications that there are longer-term variations too (e.g. Pap & Fox, 2004). When the Sun is more active, it is brighter and emits more radiation, especially in the shorter wavelengths, such as the ultraviolet (UV) and extreme ultraviolet (EUV), which are important for the radiative budget of the middle and upper atmosphere. Absorption of EUV radiation also results in ionization of O, O2, and N2. Any long-term changes in solar activity may therefore be expected to cause long-term trends in the temperature and electron density distribution in the upper atmosphere.

Lean (2001) used EUV satellite measurements and proxies (proxies only before 1974) to study long-term trends in EUV irradiance, and found an increase over the first half of the 20th century, with no clear trend after 1950. Since most observed trends in the upper atmosphere have been derived from data collected after 1950, a long-term trend in solar activity is unlikely to have contributed much to those (Laštovička, 2005). However, for other time intervals (e.g. 1900-1950) it could be a contributor.

#### **5.2 Geomagnetic activity**

328 Greenhouse Gases – Emission, Measurement and Management

the high-latitude coupling between the magnetosphere and ionosphere-thermosphere system that might arise, as their model did not include a magnetosphere. This could have resulted in an underestimation of the effects of magnetic field changes, especially at high

Recently the TIE-GCM has been successfully coupled to the Lyon-Fedder-Mobarry (LFM) MHD code, forming the Coupled Magnetosphere-Ionosphere-Thermosphere (CMIT) model (Wiltberger et al., 2004; Wang et al., 2004, 2008). With this model it is now possible to include the effects of changes in high-latitude magnetosphere-ionosphere coupling, and Cnossen et al. (2011) have started to use this to re-assess the effects of changes in the

Cnossen et al. (2011) started with a simplified experiment, investigating the effect of changes in magnetic field strength only. They reduced the magnetic field strength to 75% of its present-day value, which caused changes in hmF2 of up to -40 and +60 km and changes in foF2 of up to -3 to +1 MHz. If a linear response is assumed, considering a 5% decrease in magnetic field strength per century, these would be equivalent to trends of -0.8 to +1.2 km/decade and -0.06 to +0.02 MHz/decade. While clearly smaller than the trends predicted from the full magnetic field changes between 1957 and 1997, as expected, this rough estimate indicates that changes in magnetic field strength alone could still make a significant contribution to long-term trends. The trends in hmF2 and foF2 predicted in this way appear to

The spatial pattern of changes simulated by CMIT is different from that simulated with TIE-GCM for the full magnetic field changes between 1957 and 1997, which is perhaps not surprising, as they were caused (predominantly) by different mechanisms. The new experiments show strong changes in hmF2 over Australia, the Pacific Ocean, and the northern Indian Ocean, and smaller changes over the Atlantic Ocean. This opens up the possibility that effects of changes in the magnetic field are present over different and more widespread areas than previously thought. A detailed comparison between the changes modelled by CMIT and observed trends will be done once CMIT simulations with historic

The solar activity level varies over an approximately 11-year cycle, and there are indications that there are longer-term variations too (e.g. Pap & Fox, 2004). When the Sun is more active, it is brighter and emits more radiation, especially in the shorter wavelengths, such as the ultraviolet (UV) and extreme ultraviolet (EUV), which are important for the radiative budget of the middle and upper atmosphere. Absorption of EUV radiation also results in ionization of O, O2, and N2. Any long-term changes in solar activity may therefore be expected to cause long-term trends in the temperature and electron density distribution in

Lean (2001) used EUV satellite measurements and proxies (proxies only before 1974) to study long-term trends in EUV irradiance, and found an increase over the first half of the 20th century, with no clear trend after 1950. Since most observed trends in the upper atmosphere have been derived from data collected after 1950, a long-term trend in solar

be of similar order of magnitude as those due to changes in CO2 concentration.

magnetic fields have been performed, which are planned for the near future.

**5. Changes in solar and geomagnetic activity** 

**5.1 Solar activity** 

the upper atmosphere.

magnetic field on the ionosphere and thermosphere.

latitudes.

Geomagnetic activity is a manifestation and measure of "space weather", arising from the interaction between the solar wind and the Earth's magnetosphere. This interaction generates currents in the ionosphere and magnetosphere which cause small perturbations to the main magnetic field of the Earth that can be measured at the surface. These magnetic perturbations form the basis of indices of geomagnetic activity, such as the Ap, aa, and Kp indices. Each of these indices is derived from the K index, which is related to the maximum fluctuations of the horizontal components of the observed geomagnetic field, relative to a quiet day (a day with few disturbances due to the solar wind), during a three-hour interval.

Various studies have indicated the presence of a long-term trend in geomagnetic activity over the past 50-150 years. Clilverd et al. (1998) and Stamper et al. (2002) found that the geomagnetic activity in terms of the aa-index was increasing throughout the 20th century, and Clilverd et al. (2002) found that the number of geomagnetic storms per solar cycle had also been increasing until it stabilized in the last few cycles. Long-term changes in the solar quiet-time day variation, Sq, have been found as well. Macmillan & Droujinina (2007) reported a weak upward trend of on average 1.3 nT per century (10%) in the Sq amplitude based on records from 14 magnetic observatories. Elias et al. (2010) analysed data from three stations, and found also positive trends in Sq of 5-10%/century.

High geomagnetic activity indicates disturbed space weather conditions, usually resulting from disturbances in the solar wind. This tends to lead to stronger energetic particle precipitation and ionospheric convection at high latitudes, stronger ionospheric currents, and an enhancement of the Joule heating in the upper atmosphere, eventually leading to higher temperatures and stronger neutral winds. Variations in geomagnetic activity occur on short timescales from minutes to days, and are responsible for part of the large natural variability in the thermosphere-ionosphere system. A gradual, long-term change in the background level of geomagnetic activity could therefore cause a long-term trend in the upper atmosphere.

Because geomagnetic activity is a measure of the interaction between the solar wind and the Earth's magnetosphere-ionosphere-thermosphere system, the origin of long-term changes in geomagnetic activity could be associated either with the Sun (e.g. Stamper et al., 1999) or with the terrestrial system (or both). Clilverd et al. (2002) concluded from a simple calculation that involved various assumptions and approximations that changes in the Earth's magnetic field would have little effect on geomagnetic activity. However, a detailed analysis has not been done yet, and the simulations by Cnossen et al. (2011) indicate that the effect may be larger than previously thought, due to a stronger dependence of the ionospheric conductance on the magnetic field strength than was assumed by Clilverd et al. (2002). When considering the effects of changes in geomagnetic activity on the upper atmosphere, it is therefore possible that these may ultimately be due to changes in the Sun and/or changes in the Earth's magnetic field. Nevertheless, the effect of the observed change in geomagnetic activity on the upper atmosphere, regardless of which processes may have contributed to it, can still be investigated.

Climate Change in the Upper Atmosphere 331

that are observed. A direct test of this hypothesis is therefore not yet possible, but it remains

Temperature trends in the mesosphere can be explained largely by changes in CO2 and ozone concentration. The effect of changes in water vapour concentration is much smaller. Observed changes in the thermospheric temperature have not been fully explained by model simulations of compositional changes. They mostly have the expected sign, but are not strong enough. Estimated trends in hmF2 and foF2 are also 2-5 times smaller than what is typically observed. There are several possible explanations for the discrepancies between modelled and observed trends in the thermosphere. There are errors associated with the data (see section 2.1 and figures 3 and 4), and also the models do not represent reality perfectly. There is for instance some uncertainty over the collisional excitation rate between CO2 and atomic oxygen, which determines the efficiency of CO2 cooling. The value chosen will affect the sensitivity in the model to changes in the CO2 concentration, and there are many other factors that can have an influence too. Still the difference in magnitude between modelled and observed trends in the thermosphere is sufficiently large to suggest that thermospheric

Indeed, simulations have shown that changes in the Earth's magnetic field also affect the F2 layer. An initial study by Cnossen & Richmond (2008) showed that this is primarily important over South America and the southern Atlantic Ocean, while more recent simulations indicate that other regions may be affected too, even if only changes in magnetic field strength are considered. Detailed comparisons between modelled and observed trends are needed to determine to what extent changes in the magnetic field can explain the observations in the F2 layer. Changes in the magnetic field do not explain trends in neutral

Further work is needed to get quantitative estimates of the effects of changes in geomagnetic activity level, and possible influences of long-term changes in the lower atmosphere. Also, in order to reduce the error bars on observed trends and detect any changes in trends that may occur over time, continued monitoring of the upper atmosphere, with as much global

I would like to thank Shunrong Zhang, Bill Oliver, John Emmert, Arrun Saunders and Rashid Akmaev for sending me their data used in creating figures 4 and 5, Liying Qian for creating figure 6 for me, and Thomas Ulich for calculating the trends in hmF2. I am also grateful to Bill Oliver, Art Richmond, John Emmert, and Liying Qian for helpful feedback on an earlier version of this manuscript. The National Center for Atmospheric Research is

Akmaev, R.A., Fomichev, V.I., Gavrilov, N.M., & Shved, G.M. (1992). Simulation of the

zonal mean climatology of the middle atmosphere with a 3-dimensional spectral

trends are not caused by changes in CO2 and ozone concentration alone.

an intriguing possibility.

temperature and density.

**8. Acknowledgment** 

**9. References** 

coverage as possible, is essential.

sponsored by the National Science Foundation.

**7. Conclusion** 

Laštovička (2005) claims that the role of geomagnetic activity in causing long-term trends was small for the second half of the 20th century, just like that of solar activity. However, there has been little effort so far to actually quantify the effect of reported long-term changes in geomagnetic activity on the upper atmosphere. No detailed modelling study exists, so this may be an area for future studies to explore.

## **6. Influences from the lower atmosphere**

The lower atmosphere influences the upper atmosphere through upwardly propagating planetary waves, gravity waves and tides, and acts in general as a boundary condition for the middle and upper atmosphere. Any long-term changes in the lower atmosphere therefore have the potential to induce long-term changes in the upper atmosphere.

Jarvis (2009) argued that the longitudinal variation in hmF2 trends over Eurasia found by himself and Bremer (1998) suggests the influence of a stationary wave-like feature between 3°W and 104°E, and that the fact that it is a geo-stationary feature implies a cause linked to the troposphere or solid Earth. On this basis, he proposed that the longitudinal variation in F2 region trends might be caused via long-term changes in non-migrating tides, originating in the troposphere. Bencze (2007, 2009) also proposed an influence of non-migrating tides to explain differences in hmF2 trends between continental regions and oceans/seashores. Note that "non-migrating" in this context means "not moving with the apparent motion of the Sun".

It has been found that in particular the eastward propagating zonal wavenumber-3 diurnal tide (DE-3) has a strong influence on the ionosphere and thermosphere, especially at low latitudes (Hagan & Forbes, 2002). The DE-3 tide originates in the tropical troposphere primarily through latent heat release associated with deep convective cloud systems. Forbes et al. (2006) explained that this process is intimately connected with the predominant wavenumber-4 longitudinal distribution of topography and land-sea differences at low latitudes. The interaction of the diurnal harmonic of solar radiation with the characteristics of the Earth's surface, roughly displaying a wavenumber-4 pattern, generates the DE-3 tide. From a Sun-synchronous perspective, this appears to the observer as a wavenumber-4 feature mimicking the longitudinal surface heating pattern (Forbes & Hagan, 2000).

Wavenumber-4 patterns in the thermosphere and ionosphere have indeed been observed (e.g. Immel et al., 2006; Häusler et al., 2007). Hagan et al. (2007) performed simulations with the TIME-GCM, using a lower boundary tidal forcing from the Global Scale Wave Model (GSWM; Hagan & Forbes, 2002), which includes only latent heat release as a source of nonmigrating tides. This confirmed that the DE-3 tide generated this way propagates all the way up into the thermosphere, where it excites a wavenumber-4 longitudinal structure, similar to observations. Jarvis (2009) noted that observed wavenumber-4 patterns in the thermosphere have a scale size similar to that of the longitudinal variations found in longterm trends in hmF2. It may therefore be that long-term changes in tropical convection could be ultimately responsible for at least part of the long-term change in the ionosphere.

Unfortunately, the ionospheric data currently available are not evenly spread over the globe, and therefore are insufficient to determine whether the longitudinal variation found between 3°W and 104°E is part of a global wavenumber-4 pattern. In addition, there is insufficient information on non-migrating tides to determine whether they exhibit long-term trends, let alone whether these would be large enough to produce the ionospheric trends that are observed. A direct test of this hypothesis is therefore not yet possible, but it remains an intriguing possibility.

## **7. Conclusion**

330 Greenhouse Gases – Emission, Measurement and Management

Laštovička (2005) claims that the role of geomagnetic activity in causing long-term trends was small for the second half of the 20th century, just like that of solar activity. However, there has been little effort so far to actually quantify the effect of reported long-term changes in geomagnetic activity on the upper atmosphere. No detailed modelling study exists, so

The lower atmosphere influences the upper atmosphere through upwardly propagating planetary waves, gravity waves and tides, and acts in general as a boundary condition for the middle and upper atmosphere. Any long-term changes in the lower atmosphere

Jarvis (2009) argued that the longitudinal variation in hmF2 trends over Eurasia found by himself and Bremer (1998) suggests the influence of a stationary wave-like feature between 3°W and 104°E, and that the fact that it is a geo-stationary feature implies a cause linked to the troposphere or solid Earth. On this basis, he proposed that the longitudinal variation in F2 region trends might be caused via long-term changes in non-migrating tides, originating in the troposphere. Bencze (2007, 2009) also proposed an influence of non-migrating tides to explain differences in hmF2 trends between continental regions and oceans/seashores. Note that "non-migrating" in this context means "not moving with the apparent motion of the

It has been found that in particular the eastward propagating zonal wavenumber-3 diurnal tide (DE-3) has a strong influence on the ionosphere and thermosphere, especially at low latitudes (Hagan & Forbes, 2002). The DE-3 tide originates in the tropical troposphere primarily through latent heat release associated with deep convective cloud systems. Forbes et al. (2006) explained that this process is intimately connected with the predominant wavenumber-4 longitudinal distribution of topography and land-sea differences at low latitudes. The interaction of the diurnal harmonic of solar radiation with the characteristics of the Earth's surface, roughly displaying a wavenumber-4 pattern, generates the DE-3 tide. From a Sun-synchronous perspective, this appears to the observer as a wavenumber-4

feature mimicking the longitudinal surface heating pattern (Forbes & Hagan, 2000).

be ultimately responsible for at least part of the long-term change in the ionosphere.

Unfortunately, the ionospheric data currently available are not evenly spread over the globe, and therefore are insufficient to determine whether the longitudinal variation found between 3°W and 104°E is part of a global wavenumber-4 pattern. In addition, there is insufficient information on non-migrating tides to determine whether they exhibit long-term trends, let alone whether these would be large enough to produce the ionospheric trends

Wavenumber-4 patterns in the thermosphere and ionosphere have indeed been observed (e.g. Immel et al., 2006; Häusler et al., 2007). Hagan et al. (2007) performed simulations with the TIME-GCM, using a lower boundary tidal forcing from the Global Scale Wave Model (GSWM; Hagan & Forbes, 2002), which includes only latent heat release as a source of nonmigrating tides. This confirmed that the DE-3 tide generated this way propagates all the way up into the thermosphere, where it excites a wavenumber-4 longitudinal structure, similar to observations. Jarvis (2009) noted that observed wavenumber-4 patterns in the thermosphere have a scale size similar to that of the longitudinal variations found in longterm trends in hmF2. It may therefore be that long-term changes in tropical convection could

therefore have the potential to induce long-term changes in the upper atmosphere.

this may be an area for future studies to explore.

**6. Influences from the lower atmosphere** 

Sun".

Temperature trends in the mesosphere can be explained largely by changes in CO2 and ozone concentration. The effect of changes in water vapour concentration is much smaller. Observed changes in the thermospheric temperature have not been fully explained by model simulations of compositional changes. They mostly have the expected sign, but are not strong enough. Estimated trends in hmF2 and foF2 are also 2-5 times smaller than what is typically observed.

There are several possible explanations for the discrepancies between modelled and observed trends in the thermosphere. There are errors associated with the data (see section 2.1 and figures 3 and 4), and also the models do not represent reality perfectly. There is for instance some uncertainty over the collisional excitation rate between CO2 and atomic oxygen, which determines the efficiency of CO2 cooling. The value chosen will affect the sensitivity in the model to changes in the CO2 concentration, and there are many other factors that can have an influence too. Still the difference in magnitude between modelled and observed trends in the thermosphere is sufficiently large to suggest that thermospheric trends are not caused by changes in CO2 and ozone concentration alone.

Indeed, simulations have shown that changes in the Earth's magnetic field also affect the F2 layer. An initial study by Cnossen & Richmond (2008) showed that this is primarily important over South America and the southern Atlantic Ocean, while more recent simulations indicate that other regions may be affected too, even if only changes in magnetic field strength are considered. Detailed comparisons between modelled and observed trends are needed to determine to what extent changes in the magnetic field can explain the observations in the F2 layer. Changes in the magnetic field do not explain trends in neutral temperature and density.

Further work is needed to get quantitative estimates of the effects of changes in geomagnetic activity level, and possible influences of long-term changes in the lower atmosphere. Also, in order to reduce the error bars on observed trends and detect any changes in trends that may occur over time, continued monitoring of the upper atmosphere, with as much global coverage as possible, is essential.

### **8. Acknowledgment**

I would like to thank Shunrong Zhang, Bill Oliver, John Emmert, Arrun Saunders and Rashid Akmaev for sending me their data used in creating figures 4 and 5, Liying Qian for creating figure 6 for me, and Thomas Ulich for calculating the trends in hmF2. I am also grateful to Bill Oliver, Art Richmond, John Emmert, and Liying Qian for helpful feedback on an earlier version of this manuscript. The National Center for Atmospheric Research is sponsored by the National Science Foundation.

## **9. References**

Akmaev, R.A., Fomichev, V.I., Gavrilov, N.M., & Shved, G.M. (1992). Simulation of the zonal mean climatology of the middle atmosphere with a 3-dimensional spectral

Climate Change in the Upper Atmosphere 333

Donaldson, J.K., Wellman, T.J., & Oliver, W.L. (2010), Long-term change in thermospheric

Elias, A.G., de Artigas, M.Z., & de Haro Barbas, B.F. (2010), Trends in the solar quiet

Elias, A.G., & Ortiz de Adler, N. (2006). Earth magnetic field and geomagnetic activity

Emmert, J.T., Picone, J.M., Lean, J.L., & Knowles, S.H. (2004). Global change in the

Emmert, J.T., Picone, J.M., & Meier, R.R. (2008). Thermospheric global average density

Emmert, J.T. & Picone, J.M. (2011). Statistical uncertainty of 1967-2005 thermospheric density

Fomichev, V.I. (2009). The radiative energy budget of the middle atmosphere and its

Forbes, J.M, & Hagan, M.E. (2000). Diurnal Kelvin wave in the atmosphere of Mars:

Forbes, J.M., Russell, J., Miyahara, S., Zhang, X., Palo, S., Mlynczak, M., Mertens, C.J., &

Gubbins, D., Jones, A.L., & Finlay, C.C. (2006). Fall in Earth's magnetic field is erratic.

Hagan, M.E., & Forbes, J.M. (2002). Migrating and nonmigrating diurnal tides in the middle

Hagan, M.E., Maute, A., Roble, R.G., Richmond, A.D., Immel, T.J., & England, S.L. (2007).

Harris, M.J. (2001). *A new coupled middle atmosphere and thermosphere general circulation model:* 

Häusler, K., Lühr, H., Rentz, S., & Köhler (2007). A statistical analysis of longitudinal

from CHAMP. *J. Atmos. Solar-Terr. Phys.*, Vol. 69, No. 12, pp. 1419-1430. Haynes, P.H., Marks, C.J., McIntyre, M.E., Shepherd, T.G., & Shine, K.P. (1991). On the

*atmosphere*, PhD thesis, University College London, London, UK.

zonal forces. *J. Atmos. Sci.,* Vol. 48, No. 4, pp. 651-679.

accelerometer. *Geophys. Res. Lett.*, Vol. 27, No. 21, pp. 3563-3566.

geomagnetic field variation linked to the Earth's magnetic field secular variation and increasing concentrations of greenhouse gases, *J. Geophys. Res.*, Vol. 115, No. 8,

effects on long-term trends in the F2 layer at mid-high latitudes. *J. Atmos. Solar-Terr.* 

thermosphere: compelling evidence of a secular decrease in density. *J. Geophys. Res.*,

trends, 1967-2007, derived from orbits of 5000 near-Earth objects. *Geophys. Res. Lett.*,

trends derived from orbital drag. *J. Geophys. Res.*, in press, doi.:

parameterization in general circulation models. *J. Atmos. Solar-Terr. Phys.*, Vol. 71,

Towards an understanding of 'stationary' density structures observed by the MGS

Hagan, M.E. (2006). Troposphere-thermosphere tidal coupling as measured by the SABER instrument on TIMED during July-September 2002. *J. Geophys. Res.*, Vol.

and upper atmosphere excited by tropospheric latent heat release. *J. Geophys. Res.*,

Connections between deep tropical clouds and the Earth's ionosphere. *Geophys. Res.* 

*studies of dynamic, energetic and photochemical coupling in the middle and upper* 

dependences of upper thermospheric zonal winds at dip equator latitudes derived

downward control of extratropical diabatic circulations by eddy-induced mean

temperature above Saint Santin, *J. Geophys. Res.*, Vol. 115, A11305.

A08316, doi: 10.1029/2009JA015136.

*Phys.*, Vol. 68, No. 17, 1871-1878.

Vol. 109, No. A2, A02301.

Vol. 35, No. 5, L05101.

10.1029/2010JAO16382.

pp. 1577-1585.

111, A10S06.

*Science*, Vol. 312, pp. 900-902.

Vol. 107, No. D24, 4754.

*Lett.*, Vol. 34, L20109.

model for solstice and equinox conditions. *J. Atmos. Solar-Terr. Phys.*, Vol. 54, No. 2, pp. 119-128.


Akmaev, R.A., & Fomichev, V.I. (1998). Cooling of the mesosphere and lower thermosphere

Akmaev, R.A., & Fomichev, V.I. (2000). A model estimate of cooling in the mesosphere and

Akmaev, R.A., Fomichev, V.I., & Zhu, X. (2006). Impact of middle-atmospheric composition

Amit, H. & Olson, P. (2008). Geomagnetic dipole tilt changes induced by core flow. *Phys.* 

Baldwin, M.P., & Dunkerton, T.J. (2001). Stratospheric harbingers of anomalous weather

Beig, G., Keckhut, P., Lowe, R.P., et al. (2003). Review of mesospheric temperature trends.

Bencze, P. (2007). What do we know of the long-term change of the Earth's ionosphere? *Adv.* 

Bencze, P. (2009). Geographical distribution of long-term changes in the height of the

Bremer, J. (1998). Trends in the ionospheric E and F regions over Europe. *Ann. Geophys.*, Vol.

Bremer, J., Alfonsi, L., Bencze, P., Laštovička, J., Mikhailov, A.V., & Rogers, N. (2004). Long-

Clilverd, M.A., Clark, T.D.G., Clarke, E., & Rishbeth, H. (1998), Increased magnetic storm activity from 1868 to 1995, *J. Atmos. Solar-Terr. Phys.*, Vol. 60, pp. 1047-1056. Clilverd, M.A., Clark, T.D.G., Clarke, E., Rishbeth, H., & Ulich, T. (2002). The causes of longterm changes in the aa index. *J. Geophys. Res.*, Vol. 107, No. A12, 1441. Clilverd, M.A., Ulich, T., & Jarvis, M.J. (2003). Residual solar cycle influence on trends in ionospheric F2-layer peak height. *J. Geophys. Res.*, Vol. 108, No. A12, 1450. Cnossen, I., & Richmond, A.D. (2008). Modelling the effects of changes in the Earth's

Cnossen (2009). *Modelling of long-term trends in the middle and upper atmosphere.* PhD thesis,

Cnossen. I., Harris, M.J., Arnold, N.F., & Yiğit, E. (2009). Modelled effect of changes in the

wave parameterization, *J. Atmos. Solar-Terr. Phys.*, Vol. 71, pp. 1484-1496. Cnossen, I., Richmond, A.D., Wiltberger, M., Wang, W., & Schmitt, P. (2011). The response

Danilov, A.D. & Mikhailov, A.V. (1999). Spatial and seasonal variations of the foF2 long-

*Geophysics*, Vol. 47 (supplement), No. 2-3, pp. 1009-1029.

*Atmos. Solar-Terr. Phys.*, Vol. 70, No. 11-12, pp. 1512-1524.

term trends. *Ann. Geophys.*, Vol. 17, No. 9, pp. 1239-1243.

University of Leicester, Leicester, UK.

maximum electron density of the F region: A nonmigrating tide effect? *J. Geophys.* 

term trends in the ionosphere and upper atmosphere parameters, *Annals of* 

magnetic field from 1957 to 1997 on the ionospheric hmF2 and foF2 parameters*. J.* 

CO2 concentration on the middle and upper atmosphere: Sensitivity to gravity

of the coupled magnetosphere-ionosphere-thermosphere system to a 25% reduction in the dipole moment of the Earth's magnetic field. *J. Geophys. Res.,* in

due to doubling of CO2. *Ann. Geophys.*, Vol. 16, pp. 1501-1512.

pp. 119-128.

*Lett.*, Vol. 27, No. 14, pp. 2113-2116.

*Earth Plan. Interiors*, Vol. 166, No. 3-4, pp. 226-238.

regimes. *Science,* Vol. 294, No. 5542, pp. 581-584.

Vol. 68, No. 17, pp. 1879-1889.

*Rev. Geophys.*, Vol. 41, No. 4, 1015.

*Space Res.*, Vol. 40, pp. 1121-1125.

*Res.*, Vol. 114, A06304.

16, pp. 986-996.

press.

model for solstice and equinox conditions. *J. Atmos. Solar-Terr. Phys.*, Vol. 54, No. 2,

lower thermosphere due to the CO2 increase over the last 3-4 decades*. Geophys. Res.* 

changes on greenhouse cooling in the upper atmosphere. *J. Atmos. Solar-Terr. Phys.*,


Climate Change in the Upper Atmosphere 335

Qian, L., Laštovička, J., Roble, R.G., & Solomon, S.C. (2011). Progress in observations and

Richmond, A.D., Ridley, E.C., & Roble, R.G. (1992). A thermosphere-ionosphere general

Rinsland, C.P., Chiou, L., Boone, C., Bernath, P., Mahieu, E., & Zander, R. (2009). Trend of

Rishbeth, H. (1990). A greenhouse effect in the ionosphere? *Plan. Space Sci.*, Vol. 38, No. 7,

Rishbeth, H. (1998). How the thermospheric circulation affects the ionospheric F2-layer. *J.* 

Roble, R.G., Ridley, E.C., Richmond, A.D., Dickinson, R.E. (1988). A coupled thermosphere

Roble, R.G., & Dickinson, R.E. (1989). How will changes in carbon-dioxide and methane

Rosenlof, K.H., Oltmans, S.J., Kley, D., Russell III, J.M., Chiou, E.W., Chu, W.P., Johnson,

Sassi, F., Garcia, R.R., Marsh, D., & Koppel, K.W. (2010). The role of the middle atmosphere

Saunders, A., Lewis, H., & Swinerd, G. (2011). Further evidence of long-term thermospheric

Semenov, A.I. (1996). Temperature regime of the lower thermosphere from emission measurements during the last decades. *Geomagn. Aeron.*, Vol. 36, No. 5, pp. 90-97. She, C.-Y., Krueger, D.A., Akmaev, R.A., Schmidt, H., Talaat, E., & Yee, S. (2009). Long-term

Shimazaki, T. (1955). World-wide variations in the height of the maximum electron density of the ionospheric F2 layer. *J. Radio Res. Labs. Japan*, Vol. 2, pp. 85-97. Solomon, S. (1999). Stratospheric ozone depletion: a review of concepts and history. *Rev.* 

Stamper, R., Lockwood, M., Wild, M.N., & Clark, T.D.G. (1999). Solar causes of the longterm increase in geomagnetic activity. *J. Geophys. Res*., Vol. 104*,* pp. 28325-28342. Stolarski, R. & Frith, S.M. (2006). Search for evidence of trend slow-down in the long-term

century. *Geophys. Res. Lett.*, Vol. 28, No. 7, pp. 1195-1198.

*Geophys. Res*., in press, doi.: 10.1029/2010JAO16358.

105°W) based on lidar observations from 1990 through 2007.

uncertainty. *Atmos. Chem. Phys.*, Vol. 6, pp. 4057-4065.

ionosphere general circulation model. *Geophys. Res. Lett.*, Vol. 15, No. 12, pp. 1325-

modify the mean structure of the mesosphere and thermosphere? *Geophys. Res.* 

D.G., Kelly, K.K., Michelsen, H.A., Nedoluha, G.E., Remsberg, E.E., Toon, G.C., & McCormick, M.P. (2001). Stratospheric water vapour increases over the past half-

in simulations of the troposphere during northern hemisphere winter: differences between high- and low-top models. *J. Atmos. Sci.,* Vol. 67, No. 9, pp. 3048-3064. Salby, M., Titova, E., & Deschamps, L. (2011). Rebound of Antarctic ozone. *Geophys. Res.* 

density change using a new method of satellite ballistic coefficient estimation. *J.* 

variability in mesopause region temperatures over Fort Collins, Colorado (41°N,

TOMS/SBUV total ozone data record: the importance of instrument drift

*Spectr. Rad. Transfer*, Vol. 110, pp. 1066-1071.

*Atmos. Solar-Terr. Phys.*, Vol. 60, 1385-1402.

*Lett.*, Vol. 16, No. 12, pp. 1441-1444.

*Geophys.*, Vol. 37, No. 3, pp. 275-316.

*Lett.*, Vol. 38, L09702.

A00H03.

pp. 601-604.

pp. 945-948.

1328.

simulations of global change in the upper atmosphere. *J. Geophys. Res.*, Vol. 116,

circulation model with coupled electrodynamics. *Geophys. Res. Lett.*, Vol. 19, No. 6,

lower stratospheric methane (CH4) from atmospheric chemistry experiment (ACE) and atmospheric trace molecule spectroscopy (ATMOS) measurements, *J. Quant.* 


Holt, J.M., & Zhang, S.R. (2008). Long-term temperature trends in the ionosphere above

Hurst, D.F., Oltmans, S.J., Vömel, H., Rosenlof, K.H., Davis, S.M., Ray, E.A., Hall, E.G., &

Analysis of the 30 year Boulder record. *J. Geophys. Res.*, Vol. 116, D02306. Immel, T.J., Sagawa, E., England, S.L., Henderson, S.B., Hagan, M.E., Mende, S.B., Frey,

morphology by atmospheric tides. *Geophys. Res. Lett.*, Vol. 33, L15108. Jarvis, M.J., Clilverd, M.A., & Ulich, Th. (2002). Methodological influences on F-region peak

Jarvis, M.J. (2009). Longitudinal variation in E- and F-region ionospheric trends. *J. Atmos.* 

Keating, G.M., Tolson, R.H., & Bradford, M.S. (2000). Evidence of long term global decline in

Kivelson, M.G. & Russell, C.T. (Eds.) (1995). *Introduction to space physics*. Cambridge

Laštovička, J. (2005). On the role of solar and geomagnetic activity in long-term trends in the

Laštovička, J., Akmaev, R.A., Beig, G., Bremer, J., & Emmert, J.T. (2006a). Global change in

Laštovička, J., Mikhailov, A.V., Ulich, T., Bremer, J., Elias, A.G., Ortiz de Adler, N., Jara, V.,

Lean, J.L., White, O.R., Livingstone, C., & Picone, J.M. (2001). Variability of a composite

Macmillan, S., & Droujinina, A. (2007). Long-term trends in geomagnetic daily variation.

Marcos, F.A., Wise, J.O., Kendra, M.J., Grossbard, N.J., & Bowman, B.R. (2005). Detection of

Pap, J.M., & Fox, P. (Eds.). (2004). *Solar variability and its effects on climate*, Geophysical Monograph 141, American Geophysical Union, ISBN 0-87590-406-8, USA. Qian, L., Roble, R.G., Solomon, S.C., & Kane, T.J. (2006). Calculated and observed climate

Qian, L., Burns, A.G., Solomon, S.C., & Roble, R.G. (2009). The effect of carbon dioxide

Qian, L., & Solomon, S.C. (2011). Thermospheric density: an overview of temporal and spatial variations, *Space Sci. Rev.*, in press, doi.: 10.1007/s11214-011-9810-z.

the upper atmosphere. *Science,* Vol. 314, No. 5803, pp. 1253-1254.

time periods. *J. Geophys. Res.,* Vol. 106, No. A6, pp. 10645-10658.

height trend analyses, *Phys. Chem. Earth*, Vol. 27, pp. 589-594.

*Solar-Terr. Phys.*, Vol. 71, 1415-1429.

*Earth Planets Space*, Vol. 59, pp. 391-195.

92.

pp. 1854-1870.

L04103.

Vol. 33, L23705.

1592-1601.

*Geophys. Res. Lett.*, Vol. 27, No. 10, pp. 1523-1526.

University Press, ISBN 0-521-45714-9, Cambridge, UK.

Jordan, A.F. (2011). Stratospheric water vapor trends over Boulder, Colorado:

H.U., Swenson, C.M., & Paxton, L.J. (2006). Control of equatorial ionospheric

the Earth's thermospheric densities apparently related to anthropogenic effects.

atmosphere-ionosphere system. *J. Atmos. Solar Terr. Phys.*, Vol. 67, No. 1-2, pp. 83-

Abaraca del Rio, R., Foppiano, A.J., Ovalle, E., & Danilov, A.D. (2006b). Long-term trends in foF2: A comparison of various methods. *J. Atmos. Solar-Terr. Phys.*, Vol. 68,

chromospheric irradiance index during the 11-year activity cycle and over longer

a long-term decrease in thermospheric neutral density. *Geophys. Res. Lett.*, Vol. 32,

change in the thermosphere, and a prediction for solar cycle 24. *Geophys. Res. Lett.*,

cooling on trends in the F2-layer ionosphere. *J. Atmos. Solar-Terr. Phys.,* Vol. 71, pp.

Millstone Hill. *Geophys. Res. Lett.*, Vol. 35, No. 5, L05813.


**16**

Shuang-Ye Wu *University of Dayton,* 

*USA* 

**Projecting Changes in Extreme Precipitation**

Based on the physics of global circulation, many expect an enhanced greenhouse effect to lead to a more active hydrological cycle with more precipitation on average (Hennessy et al. 1997). This expected increase has been found in observations (Zhang et al. 2007) and has also been suggested by climate models, although these models are not consistent with respect to the spatial and temporal variability about this change. An increase in mean precipitation depth, assuming no change in the shape of the frequency distribution, would imply an increased frequency of heavy-precipitation events. However, some studies (Hennessy et al. 1997, Allen and Soden, 2008) also suggest the increase in these extreme events could be disproportionate to the change in the mean, with a greater fraction of the total precipitation being delivered by such heavy precipitation events. Such a shift towards heavy events is a common conclusion of climate models (Cubasch et al. 2001, Meehl et al. 2007) as well as analyses of observed rainfall data at the continental scale (Easterling 2000, Kunkel 2003, Groisman 2005, Min et al. 2011). However, there is great spatial variation of this average pattern. This study aims to establish likely future projections for how extreme precipitation frequency and magnitude could change in the Midwestern region of the United States, and

Present global climate models (GCMs) typically produce results at the spatial resolution of 150-300 km. This level of spatial resolution of GCMs is insufficient for establishing localized future climate projections and examining their spatial variations at the scale of a state. For increased spatial resolution, we used a set of Regional Climate Models (RCMs) run by National Center for Atmospheric Research (NCAR) under the North American Regional Climate Change Assessment Program (NARCCAP). RCMs involve nesting a higher resolution climate models within a coarser resolution GCM. The GCM output is used to define boundary conditions around a limited domain, within which RCM further models the physical dynamics of the climate system. These RCMs are designed to produce high resolution climate change simulations in order to investigate uncertainties in regional scale

investigate the spatial variation of such changes within the area.

**1. Introduction** 

**in the Midwestern United States Using**

**Assessment Program (NARCCAP)**

**Regional Climate Models** 

**North American Regional Climate Change** 


## **Projecting Changes in Extreme Precipitation in the Midwestern United States Using North American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models**

Shuang-Ye Wu *University of Dayton, USA* 

## **1. Introduction**

336 Greenhouse Gases – Emission, Measurement and Management

Ulich, Th. (2000). *Solar variability and long-term trends in the ionosphere.* PhD thesis, Sodankylä

Wang, W., Wiltberger, M., Burns, A.G., Solomon, S.C., Killeen, T.L., Maruyama, N., & Lyon,

Wang, W.B., Lei, J.H., Burns, A.G., Wiltberger, M., Richmond, A.D., Solomon, S.C., Killeen,

thermosphere (CMIT) model, *Geophys. Res. Lett.*, Vol. 35, No. 18, L18105. Wiltberger, M., Wang, W., Burns, A.G., Solomon, S.C., Lyon, J.G., & Goodrich, C.C. (2004).

Zhang, S.-R., Holt, J.M., & Kurdzo, J. (2011), Millstone Hill ISR observations of upper

J.G. (2004). Initial results from the coupled magnetosphere-ionospherethermosphere model: thermosphere-ionosphere responses. *J. Atmos. Solar-Terr.* 

T.L., Talaat, E.R., & Anderson, D.N. (2008). Ionospheric electric field variations during a geomagnetic storm simulated by a coupled magnetosphere ionosphere

Initial results from the coupled magnetosphere ionosphere thermosphere model: magnetospheric and ionospheric responses. *J. Atmos. Solar-Terr. Phys*., Vol. 66, No.

atmospheric long-term changes: Height dependency, *J. Geophys. Res.*, Vol. 116,

Geophysical Observatory Publications, No. 87, Sodankylä, Finland.

*Phys.*, Vol. 66, No. 15-16, pp. 1425-1441.

15-16, pp. 1411-1423.

A00H05.

Based on the physics of global circulation, many expect an enhanced greenhouse effect to lead to a more active hydrological cycle with more precipitation on average (Hennessy et al. 1997). This expected increase has been found in observations (Zhang et al. 2007) and has also been suggested by climate models, although these models are not consistent with respect to the spatial and temporal variability about this change. An increase in mean precipitation depth, assuming no change in the shape of the frequency distribution, would imply an increased frequency of heavy-precipitation events. However, some studies (Hennessy et al. 1997, Allen and Soden, 2008) also suggest the increase in these extreme events could be disproportionate to the change in the mean, with a greater fraction of the total precipitation being delivered by such heavy precipitation events. Such a shift towards heavy events is a common conclusion of climate models (Cubasch et al. 2001, Meehl et al. 2007) as well as analyses of observed rainfall data at the continental scale (Easterling 2000, Kunkel 2003, Groisman 2005, Min et al. 2011). However, there is great spatial variation of this average pattern. This study aims to establish likely future projections for how extreme precipitation frequency and magnitude could change in the Midwestern region of the United States, and investigate the spatial variation of such changes within the area.

Present global climate models (GCMs) typically produce results at the spatial resolution of 150-300 km. This level of spatial resolution of GCMs is insufficient for establishing localized future climate projections and examining their spatial variations at the scale of a state. For increased spatial resolution, we used a set of Regional Climate Models (RCMs) run by National Center for Atmospheric Research (NCAR) under the North American Regional Climate Change Assessment Program (NARCCAP). RCMs involve nesting a higher resolution climate models within a coarser resolution GCM. The GCM output is used to define boundary conditions around a limited domain, within which RCM further models the physical dynamics of the climate system. These RCMs are designed to produce high resolution climate change simulations in order to investigate uncertainties in regional scale

Projecting Changes in Extreme Precipitation in the Midwestern United States Using North

Fig. 1. Study area: annual mean precipitation in the Midwest United States

For model evaluation, observed daily precipitation data 1971-2000 were obtained from US Historical Climatology Network (USHCN) for climate stations in the Midwest Region. The time period was selected because it was the temporal domain of retrospective runs carried out by NARCCAP RCMs. We excluded years with less than 350 days of records to preserve frequency distribution of the data series, and any station with less than 25 complete years of records. We finally had 204 stations, with an average of 28.5 complete years of record per

NARCCAP includes 6 RCMs driven by 4 GCMs over the domain of North America, and they typically have the spatial resolution of 50 km. Not all simulations are completed. At the time of this study, 6 of the 24 RCM-GCM combinations are finished and data distributed. Table 1 shows the RCM-GCM combinations used in this study. For model evaluation, we used RCM retrospective runs (1971-2000), which produce 3-hourly precipitation data series. This data was summarized into daily precipitation data series 1971-2000. For projecting future precipitation patterns, we summarized RCM output for 2040-2070 into future daily precipitation data series. All model runs are driven with A2 emission scenario defined by the Intergovernmental Panel on Climate Change (IPCC) Special Report on Emission Scenarios (SRES) (IPCC, 2000). A2 assumes a medium high level of CO2 emission based on a

**3. Data and methodology** 

development pattern close to business-as-usual.

**3.1 Data** 

station.

American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models 339

projections of future climate and generate climate change scenarios for use in regional and local impacts research (Mearns et al, 2009).

This study aims to achieve two main objectives. First, we evaluate the performance of NARCCAP models in terms of whether they capture the frequency distribution of daily precipitation data. This evaluation is based on a comparison of retrospective model runs with observed station-based daily precipitation data. Second, based on the evaluation, we correct the bias in mean precipitation and frequency distribution of precipitation output from RCMs. After the model biases have been corrected, we then project future changes of mean and extreme precipitation patterns in the Midwest Region.

## **2. The Midwestern Region**

The definition of the Midwestern region of the USA differs in literature. The Midwestern U.S. defined for the National Climate Change Assessment program (Easterling and Karl, 2000) includes the upper eight states south of the Great Lakes (Figure 1). We expanded the region to include Kentucky because of its similarities in both physical and socio-economic characters to the rest of the region. Based on 2010 census, the nine states contain about 21% of the nation's population. Their cumulative gross state product is approximately 20% of national gross domestic product of 2010. Therefore it is an area of great economic importance. Situated at the heart of the Corn Belt, the region's economy is dominated by farming, with 89% of the land area being used for agricultural purposes. As a result, climate variability and extreme weather play a key role in the economic productivity of the region. In recent decades, the Midwest region has observed a noticeable increase in average temperatures despite the strong year-to-year variations (Union of Concerned Scientists, 2009). The largest increase has occurred in winter. Both summer and winter precipitation has been above average for the last three decades, making it the wettest period in a century. Heavy rainfall events are also significantly more frequent than a century ago. The Midwest has experienced two record-breaking floods in the past 15 years (Union of Concerned Scientists, 2009). It can be seen from these already observed changes that climate change will have a profound impact on the region. As a result it is very important to investigate how precipitation pattern will change for the region in the near future.

The Midwestern region has a typical continental climate, although the Great Lakes have a great influence on nearby areas for both temperature and precipitation. The total annual precipitation of the region averages at 950 mm (1895-2005). In fall and winter, the precipitation is largely produced by mid-latitude wave cyclones. In spring and summer, it is dominated by varying amount of convective thunderstorm rainfall. In most part of the region, precipitation is highly seasonal, with most of the rainfall concentrated in spring and summer. However in the Ohio River Valley in the southeastern part of the region, the amount of precipitation is fairly constant throughout the year. Spatially, annual precipitation ranges from 600 mm in northwest to 1200 mm in southeast, because of the increasing proximity to the Gulf of Mexico source of moisture (Figure 1). The slight deviation to this precipitation gradient is leeway shores of the Great Lakes, where lake effect precipitation, particularly from August to December, generates relatively wet conditions that lead to an annual precipitation depth around 900 - 1000 mm (Huff and Angel, 1992). Most of extreme precipitation events occur in spring and summer. Heavy precipitation in spring, when evaporation is relatively low due to lower temperature, can lead to severe flooding.

Fig. 1. Study area: annual mean precipitation in the Midwest United States

## **3. Data and methodology**

## **3.1 Data**

338 Greenhouse Gases – Emission, Measurement and Management

projections of future climate and generate climate change scenarios for use in regional and

This study aims to achieve two main objectives. First, we evaluate the performance of NARCCAP models in terms of whether they capture the frequency distribution of daily precipitation data. This evaluation is based on a comparison of retrospective model runs with observed station-based daily precipitation data. Second, based on the evaluation, we correct the bias in mean precipitation and frequency distribution of precipitation output from RCMs. After the model biases have been corrected, we then project future changes of

The definition of the Midwestern region of the USA differs in literature. The Midwestern U.S. defined for the National Climate Change Assessment program (Easterling and Karl, 2000) includes the upper eight states south of the Great Lakes (Figure 1). We expanded the region to include Kentucky because of its similarities in both physical and socio-economic characters to the rest of the region. Based on 2010 census, the nine states contain about 21% of the nation's population. Their cumulative gross state product is approximately 20% of national gross domestic product of 2010. Therefore it is an area of great economic importance. Situated at the heart of the Corn Belt, the region's economy is dominated by farming, with 89% of the land area being used for agricultural purposes. As a result, climate variability and extreme weather play a key role in the economic productivity of the region. In recent decades, the Midwest region has observed a noticeable increase in average temperatures despite the strong year-to-year variations (Union of Concerned Scientists, 2009). The largest increase has occurred in winter. Both summer and winter precipitation has been above average for the last three decades, making it the wettest period in a century. Heavy rainfall events are also significantly more frequent than a century ago. The Midwest has experienced two record-breaking floods in the past 15 years (Union of Concerned Scientists, 2009). It can be seen from these already observed changes that climate change will have a profound impact on the region. As a result it is very important to investigate how

The Midwestern region has a typical continental climate, although the Great Lakes have a great influence on nearby areas for both temperature and precipitation. The total annual precipitation of the region averages at 950 mm (1895-2005). In fall and winter, the precipitation is largely produced by mid-latitude wave cyclones. In spring and summer, it is dominated by varying amount of convective thunderstorm rainfall. In most part of the region, precipitation is highly seasonal, with most of the rainfall concentrated in spring and summer. However in the Ohio River Valley in the southeastern part of the region, the amount of precipitation is fairly constant throughout the year. Spatially, annual precipitation ranges from 600 mm in northwest to 1200 mm in southeast, because of the increasing proximity to the Gulf of Mexico source of moisture (Figure 1). The slight deviation to this precipitation gradient is leeway shores of the Great Lakes, where lake effect precipitation, particularly from August to December, generates relatively wet conditions that lead to an annual precipitation depth around 900 - 1000 mm (Huff and Angel, 1992). Most of extreme precipitation events occur in spring and summer. Heavy precipitation in spring, when evaporation is relatively low due to lower temperature,

local impacts research (Mearns et al, 2009).

**2. The Midwestern Region** 

can lead to severe flooding.

mean and extreme precipitation patterns in the Midwest Region.

precipitation pattern will change for the region in the near future.

For model evaluation, observed daily precipitation data 1971-2000 were obtained from US Historical Climatology Network (USHCN) for climate stations in the Midwest Region. The time period was selected because it was the temporal domain of retrospective runs carried out by NARCCAP RCMs. We excluded years with less than 350 days of records to preserve frequency distribution of the data series, and any station with less than 25 complete years of records. We finally had 204 stations, with an average of 28.5 complete years of record per station.

NARCCAP includes 6 RCMs driven by 4 GCMs over the domain of North America, and they typically have the spatial resolution of 50 km. Not all simulations are completed. At the time of this study, 6 of the 24 RCM-GCM combinations are finished and data distributed. Table 1 shows the RCM-GCM combinations used in this study. For model evaluation, we used RCM retrospective runs (1971-2000), which produce 3-hourly precipitation data series. This data was summarized into daily precipitation data series 1971-2000. For projecting future precipitation patterns, we summarized RCM output for 2040-2070 into future daily precipitation data series. All model runs are driven with A2 emission scenario defined by the Intergovernmental Panel on Climate Change (IPCC) Special Report on Emission Scenarios (SRES) (IPCC, 2000). A2 assumes a medium high level of CO2 emission based on a development pattern close to business-as-usual.

Projecting Changes in Extreme Precipitation in the Midwestern United States Using North

between *Pj* and *Pj*

available.

*m*

American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models 341

Calculate the model error (*E*) for each station as the root mean squared difference

1

*j*

*E*

then calculated the mean error for all stations in the state for each model.

values. We propose a quantile mapping method to achieve this correction.

**3.3.1 Bias correction through quantile mapping** 

*i x*

the sorted observed precipitation data series where 1

This involves the following steps:

precipitation data series *mf*

series: *obs Pj* and *mc Pj*

**3.3 Method for projecting future mean and extreme precipitation** 

2

(4)

( ) *<sup>n</sup> <sup>m</sup> j j*

*P P*

*n*

Based this method, we evaluated *E* for complete precipitation data series, as well as the error for observed dry days (*Ed*) and wet days (*Ew*) separately at each climate station. We

A Common approach to studying extreme events is to establish their probability distributions, upon which the magnitude and return intervals (frequency) of extreme events can be established. This can be achieved either by fitting a theoretical distribution functions, or based on the empirical distribution of the data. The advantage of using theoretical distributions is that they can smooth out outliers, particularly when data is only available for short time frame, and they can project beyond the time period when data is

Before we can apply this approach, however, we need to correct biases of RCMs in the frequency distribution of their output daily precipitation data series (as evaluated based on the methods outlined in 3.2). It is important to correct the entire distribution, not just mean

Establish an empirical cumulative distribution function (*Fmf*)for modeled future daily

1 <sup>1</sup> ( ) 1{ } *<sup>n</sup> mf*

 For each modeled future precipitation value *mf Pj* , find the corresponding quantile values of same probability in the observed and modeled current daily precipitation data

( ) ( ( )) *mf*

( ) ( ( )) *mf*

Where *nobs* is the number of data points in the observed precipitation data series; *xobs* is

*obs mf j*

*mc mf j*

(6)

(7)

*obs <sup>x</sup> <sup>≤</sup>* <sup>2</sup>

(5)

*obs x … ≤ obs*

*<sup>i</sup> x* .

*mf i i Ft x t n*

*obs obs mf obs <sup>j</sup> mf <sup>j</sup> nF P P Q FP x*

*mc mc mf mc <sup>j</sup> mf <sup>j</sup> nF P P QFP x*


Table 1. GCM-RCM combinations used in this study

#### **3.2 Model evaluation methodology**

In order to examine the frequency and magnitude of extreme events, it is important for a model to capture an accurate distribution of the data series. We first computed and compared the statistics that describe the shape of distribution for observed and modeled daily precipitation data. We chose to use L-moments to characterize the frequency distributions based on methods developed in Hosking (1990). L-moments are similar to conventional moments in that they are calculated to summarize the shape of a probability distribution. L-mean is identical to the conventional mean, whereas L-scale, L-skewness and L-kurtosis are analogous to conventional standard deviation, skewness and kurtosis. Lmoments however are often considered superior to conventional moments in characterizing distribution shapes, because they are less sensitive to outliers (Ulrych et. al, 2000, Elamir et.al, 2003, and Hosking 2006).

We then attempted to quantify the deviation of the distributions of daily data output from 6 NARCCAP RCMs from observed daily precipitation distribution based on the following steps:

 Establish the empirical cumulative distribution function (*Fobs*) for observed precipitation series for each station: *xi* , *i=1,2…n*

$$F\_{obs}(t) = \frac{1}{n} \sum\_{i=1}^{n} \mathbf{1}\{\mathbf{x}\_i \le t\} \tag{1}$$

Where 1{A} is the indicator function of event A.

Based on *Fobs*, calculate the probability (*p*) for each observed precipitation value *Pj*

$$p(P\_j) = F\_{\text{obs}}(P\_j) = \frac{1}{n} \sum\_{i=1}^{n} \mathbf{1}\{\mathbf{x}\_i \le P\_j\} \tag{2}$$

 Find the corresponding value in the modeled daily precipitation data for the grid point closest to that station (*Pj <sup>m</sup>*) that has the same probability. That is, calculate the quantile (*Qm*)value at *p* in the modeled precipitation data series

$$P\_j^{m} = Q^{m}(p(P\_j)) = \boldsymbol{\pi}^{m}\_{\left[ \boldsymbol{\pi}\_{m} p(P\_j) \right]} \tag{3}$$

Where *nm* is the number of data points in the modeled precipitation data series; *xm* is the sorted modeled precipitation data series where 1 *mx <sup>≤</sup> <sup>m</sup> x*<sup>2</sup> *… <sup>≤</sup> <sup>m</sup> <sup>n</sup> x* ;⎾⏋is the ceiling function.

 Calculate the model error (*E*) for each station as the root mean squared difference between *Pj* and *Pj m*

$$E = \sqrt{\frac{\sum\_{j=1}^{n} (P\_j - P\_j^{\prime m})^2}{n}} \tag{4}$$

Based this method, we evaluated *E* for complete precipitation data series, as well as the error for observed dry days (*Ed*) and wet days (*Ew*) separately at each climate station. We then calculated the mean error for all stations in the state for each model.

#### **3.3 Method for projecting future mean and extreme precipitation**

A Common approach to studying extreme events is to establish their probability distributions, upon which the magnitude and return intervals (frequency) of extreme events can be established. This can be achieved either by fitting a theoretical distribution functions, or based on the empirical distribution of the data. The advantage of using theoretical distributions is that they can smooth out outliers, particularly when data is only available for short time frame, and they can project beyond the time period when data is available.

Before we can apply this approach, however, we need to correct biases of RCMs in the frequency distribution of their output daily precipitation data series (as evaluated based on the methods outlined in 3.2). It is important to correct the entire distribution, not just mean values. We propose a quantile mapping method to achieve this correction.

#### **3.3.1 Bias correction through quantile mapping**

This involves the following steps:

340 Greenhouse Gases – Emission, Measurement and Management

HRM3 X

In order to examine the frequency and magnitude of extreme events, it is important for a model to capture an accurate distribution of the data series. We first computed and compared the statistics that describe the shape of distribution for observed and modeled daily precipitation data. We chose to use L-moments to characterize the frequency distributions based on methods developed in Hosking (1990). L-moments are similar to conventional moments in that they are calculated to summarize the shape of a probability distribution. L-mean is identical to the conventional mean, whereas L-scale, L-skewness and L-kurtosis are analogous to conventional standard deviation, skewness and kurtosis. Lmoments however are often considered superior to conventional moments in characterizing distribution shapes, because they are less sensitive to outliers (Ulrych et. al, 2000, Elamir

We then attempted to quantify the deviation of the distributions of daily data output from 6 NARCCAP RCMs from observed daily precipitation distribution based on the following steps: Establish the empirical cumulative distribution function (*Fobs*) for observed precipitation

> 1 <sup>1</sup> ( ) 1{ } *n obs i i Ft xt n*

<sup>1</sup> ( ) ( ) 1{ } *n j obs j i j i*

Find the corresponding value in the modeled daily precipitation data for the grid point

( ( ))

*mm m P Q pP x j j*

Where *nm* is the number of data points in the modeled precipitation data series; *xm* is the

*p P FP xP n*

1

Based on *Fobs*, calculate the probability (*p*) for each observed precipitation value *Pj*

(1)

(2)

(3)

*mx <sup>≤</sup> <sup>m</sup> x*<sup>2</sup> *… <sup>≤</sup> <sup>m</sup>*

*<sup>n</sup> x* ;⎾⏋is the ceiling

*<sup>m</sup>*) that has the same probability. That is, calculate the quantile

( )

*n pP m j*

CCSM CGCM3 GFDL HADCM3

RCMs GCMs

RCM3 X

Table 1. GCM-RCM combinations used in this study

MM5I X

WRFG X

**3.2 Model evaluation methodology** 

et.al, 2003, and Hosking 2006).

closest to that station (*Pj*

function.

series for each station: *xi* , *i=1,2…n*

Where 1{A} is the indicator function of event A.

(*Qm*)value at *p* in the modeled precipitation data series

sorted modeled precipitation data series where 1

CRCM X X

 Establish an empirical cumulative distribution function (*Fmf*)for modeled future daily precipitation data series *mf i x*

$$F\_{mf}(t) = \frac{1}{n} \sum\_{i=1}^{n} \mathbf{1} \{ \mathbf{x}\_i^{mf} \le t \} \tag{5}$$

 For each modeled future precipitation value *mf Pj* , find the corresponding quantile values of same probability in the observed and modeled current daily precipitation data series: *obs Pj* and *mc Pj*

$$P\_j^{obs} = \mathbb{Q}^{obs} \left( F\_{mf} \left( P\_j^{mf} \right) \right) = \boldsymbol{\mathfrak{x}}\_{\left[ \boldsymbol{n}\_{\mathrm{obs}}, F\_{mf} \left( P\_j^{mf} \right) \right]}^{obs} \tag{6}$$

Where *nobs* is the number of data points in the observed precipitation data series; *xobs* is the sorted observed precipitation data series where 1 *obs <sup>x</sup> <sup>≤</sup>* <sup>2</sup> *obs x … ≤ obs <sup>i</sup> x* .

$$P\_j^{mc} = \mathbb{Q}^{mc} \{ F\_{mf} \{ P\_j^{mf} \} \} = \mathfrak{x}\_{\left\lceil \begin{smallmatrix} n\_{mc} \ F\_{mf} \end{smallmatrix} \{ P\_j^{mf} \} \right\rceil}^{mc} \tag{7}$$

Projecting Changes in Extreme Precipitation in the Midwestern United States Using North

function of the cumulative probability function *F(x)*.

as:

0.30

 0.35

 0.40

 0.45

**GL**

L-Kurtosis

 0.50

 0.55

 0.60

 0.65

American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models 343

daily precipitation data series. After the cumulative probably density function *F(x)* is established we can derive the magnitude of precipitation events of various return intervals

Where *Py* is the precipitation value with the return interval of *y* years, *F-1(x)* is the inverse

0.55 0.60 0.65 0.70 0.75 0.80

Fig. 2. L-moment ratio diagram for observed and modeled daily precipitation data in

comparison to the L-moment ratio curves of theoretical distributions

**Gamma Overall**

**LN**

L-Skewness

**lower bound**

<sup>1</sup> <sup>1</sup> (1 ) <sup>365</sup> *P F <sup>y</sup> <sup>y</sup>*

**GEV**

**Wakeby**

(11)

**GP**

Observed CCSM\_MM5I CCSM\_WRFG CGCM3\_CRCM CGCM3\_RCM3 GFDL\_RCM3 HADCM3\_HRM3

Where *nmc* is the size of the modeled current precipitation data series; *xmc* is the sorted modeled current precipitation data series where 1 *mc <sup>x</sup> <sup>≤</sup>* <sup>2</sup> *mc x … ≤ mc <sup>i</sup> x* .

 We then use the ratio of *obs Pj* and *mc Pj* to correct the bias in future precipitation data series.

$$\hat{P}\_{j}^{mf} = P\_{j}^{mf} \frac{P\_{j}^{obs}}{P\_{j}^{mc}} = P\_{j}^{mf} \frac{\mathcal{X}\_{\left\lceil \begin{smallmatrix} n\_{ch}, F\_{mf} \left( P\_{j}^{mf} \end{smallmatrix} \right) \right\rceil}}{\mathcal{X}\_{\left\lceil \begin{smallmatrix} n\_{mc}, F\_{mf} \left( P\_{j}^{mf} \end{smallmatrix} \right) \end{smallmatrix}}} \tag{8}$$

Based on this bias-correction, we derived a future daily precipitation data series from RCMs for each of the climate station.

#### **3.3.2 Frequency analysis**

After the bias in the distribution is corrected, we then fit a theoretical distribution function to both the observed and future modeled data, based on which the magnitude of precipitation events of certain return intervals can be established for comparison. A variety of distribution functions have been used to study extreme precipitation events, such as generalized extreme value (GEV), Weibull and Pearson III distributions for annual maxima data, generalized Pareto (GP) distribution for excesses over a high threshold, and generalized Logistic (GL), lognormal (LN) and Gamma distributions for full precipitation records. We used the L-moment ratio (L-skewness vs. L-kurtosis) diagram (Rao and Hamed, 1999) to help select the best distribution for our data. Figure 2 plot the L-skewness and Lkurtosis ratios for both observed and modeled data for all stations. Results show that Gamma distribution best approximate the distribution of daily precipitation totals. This selection is also supported by many previous studies (Crutcher et al. 1977, Buishand 1978, Guttman et al. 1993, Groisman et al. 1999). Gamma distribution has the following density function:

$$f(\mathbf{x}) = \frac{1}{a^{\beta} \Gamma(\beta)} \mathbf{x}^{\beta - 1} e^{-(\frac{\mathbf{x}}{a})} \tag{9}$$

The variable x has lower bound of zero, 0< x < ∞. For this family of distributions, the αparameter defines the shape of the distribution, while β-parameter characterizes the scale. Since it does not rain every day, a mixed distribution model is sometimes considered for daily precipitation totals (Groisman et al, 1999). Under this model, it is assumed that the occurrence of daily precipitation events has a binary distribution with the probability of a single event *ppr*, and the precipitation amount during this event is considered to have a gamma-distribution. The cumulative distribution function of precipitation totals *F(x)* is expressed as:

$$F(\mathbf{x}) = P(X \le \mathbf{x}) = (1 - p\_{pr}) + p\_{pr} \int\_0^\mathbf{x} f(\alpha, \beta, t) dt \tag{10}$$

with *f(α,β,t)dt* being the probability density function of gamma-distribution. *ppr* can be estimated as the percentage of wet days in the data series. The maximum likelihood method is used to estimate the parameters α and β for each of the observed and future modeled

We then use the ratio of *obs Pj* and *mc Pj* to correct the bias in future precipitation data

<sup>ˆ</sup> *mf*

*PP P*

*j nF P*

*obs obs nF P mf mf <sup>j</sup> mf jj j mc mc*

*<sup>x</sup> <sup>P</sup>*

*P x*

Based on this bias-correction, we derived a future daily precipitation data series from RCMs

After the bias in the distribution is corrected, we then fit a theoretical distribution function to both the observed and future modeled data, based on which the magnitude of precipitation events of certain return intervals can be established for comparison. A variety of distribution functions have been used to study extreme precipitation events, such as generalized extreme value (GEV), Weibull and Pearson III distributions for annual maxima data, generalized Pareto (GP) distribution for excesses over a high threshold, and generalized Logistic (GL), lognormal (LN) and Gamma distributions for full precipitation records. We used the L-moment ratio (L-skewness vs. L-kurtosis) diagram (Rao and Hamed, 1999) to help select the best distribution for our data. Figure 2 plot the L-skewness and Lkurtosis ratios for both observed and modeled data for all stations. Results show that Gamma distribution best approximate the distribution of daily precipitation totals. This selection is also supported by many previous studies (Crutcher et al. 1977, Buishand 1978, Guttman et al. 1993, Groisman et al. 1999). Gamma distribution has the following density

( ) <sup>1</sup> <sup>1</sup> ( ) ( )

*<sup>a</sup> fx x e*

The variable x has lower bound of zero, 0< x < ∞. For this family of distributions, the αparameter defines the shape of the distribution, while β-parameter characterizes the scale. Since it does not rain every day, a mixed distribution model is sometimes considered for daily precipitation totals (Groisman et al, 1999). Under this model, it is assumed that the occurrence of daily precipitation events has a binary distribution with the probability of a single event *ppr*, and the precipitation amount during this event is considered to have a gamma-distribution. The cumulative distribution function of precipitation totals *F(x)* is

> <sup>0</sup> ( ) ( ) (1 ) ( , , ) *<sup>x</sup> Fx PX x pr pr p pf*

with *f(α,β,t)dt* being the probability density function of gamma-distribution. *ppr* can be estimated as the percentage of wet days in the data series. The maximum likelihood method is used to estimate the parameters α and β for each of the observed and future modeled

  *x*

 

*t dt* (10)

(9)

modeled current precipitation data series where 1

series.

function:

expressed as:

for each of the climate station.

**3.3.2 Frequency analysis** 

Where *nmc* is the size of the modeled current precipitation data series; *xmc* is the sorted

*mc <sup>x</sup> <sup>≤</sup>* <sup>2</sup>

( )

*obs mf j*

( )

*mf mc mf j*

(8)

*mc x … ≤ mc*

*<sup>i</sup> x* .

daily precipitation data series. After the cumulative probably density function *F(x)* is established we can derive the magnitude of precipitation events of various return intervals as:

$$P\_y = F^{-1}(1 - \frac{1}{365y})\tag{11}$$

Where *Py* is the precipitation value with the return interval of *y* years, *F-1(x)* is the inverse function of the cumulative probability function *F(x)*.

Fig. 2. L-moment ratio diagram for observed and modeled daily precipitation data in comparison to the L-moment ratio curves of theoretical distributions

Projecting Changes in Extreme Precipitation in the Midwestern United States Using North

Observed 2.62 0.88 0.77 0.53

**Model Mean** 

precipitation data series

and wet days (*Ew*)

**4.2.1 Mean daily precipitation** 

increase in all Midwest region.

American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models 345

CCSM\_MM5I 2.40 -8.38 0.84 -3.58 0.72 -6.18 0.46 -12.05 CCSM\_WRFG 1.96 -25.39 0.87 -0.73 0.77 0.14 0.55 4.17 CGCM3\_CRCM 2.65 0.92 0.76 -13.28 0.62 -19.55 0.36 -32.03 CGCM3\_RCM3 3.09 18.01 0.80 -8.28 0.67 -12.36 0.42 -19.97 GFDL\_RCM3 2.90 10.52 0.79 -9.56 0.66 -14.52 0.40 -24.14 HADCM3\_HRM3 2.95 12.49 0.83 -4.94 0.71 -7.71 0.46 -13.37 Model mean 2.66 1.36 0.82 -6.73 0.69 -10.03 0.44 -16.23

Table 2. Summary statistics for frequency distributions of observed and modeled daily

**CCSM\_MM5I** 0.93 36 0.07 2.95 34 **CCSM\_WRFG** 1.11 42 0.05 3.63 42 **CGCM3\_CRCM** 1.65 63 0.36 4.65 54 **CGCM3\_RCM3** 0.85 32 0.27 2.16 25 **GFDL\_RCM3** 1.02 39 0.29 2.69 31 **HADCM3\_HRM3** 0.78 30 0.15 2.18 25 **Model mean** 1.06 40 0.20 3.04 35

Table 3. Root mean squared error of modeled quantile values for all data (*E*), dry days (*Ed*)

After applying bias correction based on quantile mapping method, we calculated future mean daily precipitation for the Midwest region for the time period 2040-2070. Table 4 summarizes how daily mean precipitation might change for the study area based on climate models. With the exception of CCSM\_WRFG (which projects very slight decrease), all other models project increase in mean precipitation in the region. The average change for all models is 7.7%, ranging from 0-12%. Figure 5 shows the spatial pattern of precipitation change based on mean projection of all models. It seems that mean precipitation is likely to increase more in north and less in the south. This is likely due to increased evaporation from the Great Lakes under a warmer temperature. Spatial pattern of mean precipitation change varies among different models (Figure 6), but most models project increase in the northern part of the region. CCSM\_WRFG is the only model that project decrease of precipitation in the south at the similar magnitude as the increase in the north. All other models project

**4.2 Projecting future precipitation patterns in the Midwest Region** 

**daily precipitation** *Ed Ew*

*Ew* **as % of mean wet day precipitation** 

**Model** *E E* **as % of mean** 

**(mm/day) L-scale L-skewness L-kurtosis**  Value % diff. Value % diff. Value % diff. Value % diff.

#### **4. Results and discussions**

#### **4.1 Model evaluation: Frequency distributions of observed and modeled daily precipitation data**

Summary L-moment statistics for frequency distributions of observed and modeled daily precipitation data series are presented in Table 2. It can be seen that CCSM\_WRFG daily precipitation series deviate most from the observed distribution of daily precipitation. The remaining models have comparable biases. They all have greater mean (except for CCSM\_MM5I), less variance, skewness and kurtosis. This seems to indicate that the modeled data is less spread out, have higher values in small precipitation events and lower values in large events (hence less skewed than observed).

Figure 3 illustrates the correlation between observed and modeled mean daily precipitation at all stations. The correlation coefficient (Pearson's r) values are also presented in the figure. It should be noted that to evaluate a model's performance at estimating mean precipitation, we should not only look at the correlation coefficient, but also how closely the points follow the x=y line. It can be seen that CCSM\_WRFG model consistently under-estimate the mean daily precipitation. Most models slightly over estimate the mean daily precipitation (such as CGCM\_RCM3, GFDL\_RCM3 and HADCM3\_HRM3). The model that performs best in terms of mean precipitation is CGCM3\_CRCM, which has the highest correlation coefficient (0.81), and follows x=y line most closely. However, close estimate in mean precipitation does not mean the model also simulates well the whole distribution of daily precipitation data series. Figure 4 presents the quantile-quantile (Q-Q) plots between observed and modeled data for all stations for each of the 6 RCMs in order to show the deviation from observed distribution as well as the spread of different models. It can be seen from figure 4 that despite its close estimate of mean precipitation, CGCM3\_CRCM performs the worst in estimating the complete distribution. It consistently underestimates most of the wet day quantile values. The close estimate of the mean is a result of over-estimation of small precipitation events offsetting this underestimation of larger quantile values. Figure 4 shows that majority of models underestimate large quantile values. All models overestimate days with no or small amount of precipitation. This can also be illustrated by the dry day quintile errors (*Ed*) presented in table 3. To some extent, this could be attributable to the fact that RCMs output average precipitation of a whole grid, which covers an area of 50 by 50 km, whereas station data are point-based.

Table 3 summarizes root mean squared error of modeled quantile values for all data (*E*), dry days (*Ed*) and wet days (*Ew*) respectively. Based on RMSE, it seems that HADCM3\_HRM3 performs best in estimating the observed frequency distribution of daily precipitation data, as it has lowest *E* and *Ew*, and fairly low *Ed*. The calculated model errors again confirm that CCSM\_WRFG and CGCM3\_CRCM have the poorest performance in simulating the frequency distribution, and its model errors are above the other models. The rest of the models have comparable performance. The average RMSE for full record quantiles for all models is about 1 mm, or about 40% of mean precipitation amount. This ranges from 0.78 to 1.65 mm, or 30 to 63% of mean daily precipitation. The average RMSE for wet day quantile values for all models is 3.04 mm, or 35% of the mean wet day precipitation. It ranges between 25-54% of mean wet day precipitation among different models.


Table 2. Summary statistics for frequency distributions of observed and modeled daily precipitation data series


Table 3. Root mean squared error of modeled quantile values for all data (*E*), dry days (*Ed*) and wet days (*Ew*)

### **4.2 Projecting future precipitation patterns in the Midwest Region**

## **4.2.1 Mean daily precipitation**

344 Greenhouse Gases – Emission, Measurement and Management

Summary L-moment statistics for frequency distributions of observed and modeled daily precipitation data series are presented in Table 2. It can be seen that CCSM\_WRFG daily precipitation series deviate most from the observed distribution of daily precipitation. The remaining models have comparable biases. They all have greater mean (except for CCSM\_MM5I), less variance, skewness and kurtosis. This seems to indicate that the modeled data is less spread out, have higher values in small precipitation events and lower

Figure 3 illustrates the correlation between observed and modeled mean daily precipitation at all stations. The correlation coefficient (Pearson's r) values are also presented in the figure. It should be noted that to evaluate a model's performance at estimating mean precipitation, we should not only look at the correlation coefficient, but also how closely the points follow the x=y line. It can be seen that CCSM\_WRFG model consistently under-estimate the mean daily precipitation. Most models slightly over estimate the mean daily precipitation (such as CGCM\_RCM3, GFDL\_RCM3 and HADCM3\_HRM3). The model that performs best in terms of mean precipitation is CGCM3\_CRCM, which has the highest correlation coefficient (0.81), and follows x=y line most closely. However, close estimate in mean precipitation does not mean the model also simulates well the whole distribution of daily precipitation data series. Figure 4 presents the quantile-quantile (Q-Q) plots between observed and modeled data for all stations for each of the 6 RCMs in order to show the deviation from observed distribution as well as the spread of different models. It can be seen from figure 4 that despite its close estimate of mean precipitation, CGCM3\_CRCM performs the worst in estimating the complete distribution. It consistently underestimates most of the wet day quantile values. The close estimate of the mean is a result of over-estimation of small precipitation events offsetting this underestimation of larger quantile values. Figure 4 shows that majority of models underestimate large quantile values. All models overestimate days with no or small amount of precipitation. This can also be illustrated by the dry day quintile errors (*Ed*) presented in table 3. To some extent, this could be attributable to the fact that RCMs output average precipitation of a whole grid, which covers an area of 50 by 50 km, whereas station

Table 3 summarizes root mean squared error of modeled quantile values for all data (*E*), dry days (*Ed*) and wet days (*Ew*) respectively. Based on RMSE, it seems that HADCM3\_HRM3 performs best in estimating the observed frequency distribution of daily precipitation data, as it has lowest *E* and *Ew*, and fairly low *Ed*. The calculated model errors again confirm that CCSM\_WRFG and CGCM3\_CRCM have the poorest performance in simulating the frequency distribution, and its model errors are above the other models. The rest of the models have comparable performance. The average RMSE for full record quantiles for all models is about 1 mm, or about 40% of mean precipitation amount. This ranges from 0.78 to 1.65 mm, or 30 to 63% of mean daily precipitation. The average RMSE for wet day quantile values for all models is 3.04 mm, or 35% of the mean wet day precipitation. It ranges

between 25-54% of mean wet day precipitation among different models.

**4.1 Model evaluation: Frequency distributions of observed and modeled daily** 

values in large events (hence less skewed than observed).

**4. Results and discussions** 

**precipitation data** 

data are point-based.

After applying bias correction based on quantile mapping method, we calculated future mean daily precipitation for the Midwest region for the time period 2040-2070. Table 4 summarizes how daily mean precipitation might change for the study area based on climate models. With the exception of CCSM\_WRFG (which projects very slight decrease), all other models project increase in mean precipitation in the region. The average change for all models is 7.7%, ranging from 0-12%. Figure 5 shows the spatial pattern of precipitation change based on mean projection of all models. It seems that mean precipitation is likely to increase more in north and less in the south. This is likely due to increased evaporation from the Great Lakes under a warmer temperature. Spatial pattern of mean precipitation change varies among different models (Figure 6), but most models project increase in the northern part of the region. CCSM\_WRFG is the only model that project decrease of precipitation in the south at the similar magnitude as the increase in the north. All other models project increase in all Midwest region.

Projecting Changes in Extreme Precipitation in the Midwestern United States Using North

0 50 100 150 200

Observed quantile value (mm)

**CGCM3\_CRCM**

0 50 100 150 200

Observed quantile value (mm)

**GFDL\_RCM3**

0 50 100 150 200

Observed quantile value (mm)

Fig. 4. Q-Q plots for all stations for each of the RCMs

**CCSM\_MM5I**

0

0

0

 50

 100

Modeled quantile value

 150

 200

 50

 100

Modeled quantile value

 150

 200

 50

 100

Modeled quantile value

 150

 200

American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models 347

0

0

0

 50

 100

Modeled quantile value

 150

 200

 50

 100

Modeled quantile value

 150

 200

 50

 100

Modeled quantile value

 150

 200

0 50 100 150 200

Observed quantile value (mm)

**CGCM3\_RCM3**

0 50 100 150 200

Observed quantile value (mm)

**HADCM3\_HRM3**

0 50 100 150 200

Observed quantile value (mm)

**CCSM\_WRFG**

Fig. 3. Correlation between observed and modeled mean daily precipitation at all stations.

1.0 1.5 2.0 2.5 3.0 3.5 4.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0

Modeled mean daily precipitation (mm)

Fig. 3. Correlation between observed and modeled mean daily precipitation at all stations.

Modeled mean daily precipitation (mm)

Modeled mean daily precipitation (mm)

1.0 1.5 2.0 2.5 3.0 3.5 4.0

r = 0.75

r = 0.81

r = 0.76

Observed mean daily precipitation (mm)

**CGCM3\_RCM3**

1.0 1.5 2.0 2.5 3.0 3.5 4.0

Observed mean daily precipitation (mm)

**HADCM3\_HRM3**

1.0 1.5 2.0 2.5 3.0 3.5 4.0

Observed mean daily precipitation (mm)

**CCSM\_WRFG**

1.0 1.5 2.0 2.5 3.0 3.5 4.0

r = 0.71

r = 0.81

r = 0.77

Observed mean daily precipitation (mm)

**CGCM3\_CRCM**

1.0 1.5 2.0 2.5 3.0 3.5 4.0

Observed mean daily precipitation (mm)

**GFDL\_RCM3**

1.0 1.5 2.0 2.5 3.0 3.5 4.0

Observed mean daily precipitation (mm)

**CCSM\_MM5I**

1.0 1.5 2.0 2.5 3.0 3.5 4.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0

Modeled mean daily precipitation (mm)

Modeled mean daily precipitation (mm)

Modeled mean daily precipitation (mm)

Fig. 4. Q-Q plots for all stations for each of the RCMs

Projecting Changes in Extreme Precipitation in the Midwestern United States Using North

**4.2.2 Extreme precipitation events** 

**Future (2040-2070)** 

American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models 349

We used the mixed probability model (eq. 10) to project extreme precipitation events. We used the wet day frequency as an estimate for precipitation probability (*ppr*), and fit a Gamma distribution to the wet day precipitation data using maximum likelihood method. Applying the same methodology on the observed and bias-corrected future daily precipitation data, we derived the magnitudes of extreme events of return interval of 1, 5, 10, 15, 20 and 25 years for both present (1970-2000) and future (2040-2070). Table 5 summarized the results. All models show increase in precipitation depths of extreme events. In general, extreme precipitation events increase at greater magnitudes than mean precipitation. The increase is greater for more extreme events. For average results of all models, compared with 8% increase in mean precipitation, precipitation event of 1 year return interval is likely to increase by 13%, and that of 25 year could increase by 19%. There are variations among models in how they project spatial distribution of such changes. Our model evaluation suggests that mean of all models seems to capture the spatial variation of precipitation best. Based on this observation, we used mean values of all models to interpolate the spatial distribution of how extreme precipitation could change in the Midwest (Figure 7). It can be seen that in general the intensity for extreme precipitation increase disproportionately

**Return interval 1 year 5 year 10 year 15 year 20 year 25 year Observed (1970-2000)** mm/day 55 82 96 103 108 112

**CCSM\_MM5I** mm/day 62 93 108 117 124 128

**CCSM\_WRFG** mm/day 62 93 108 116 124 129

**CGCM3\_CRCM** mm/day 61 91 106 114 121 125

**CGCM3\_RCM3** mm/day 64 100 117 127 135 141

**GFDL\_RCM3** mm/day 61 94 111 122 132 137

**HADCM3\_HRM3** mm/day 64 98 114 126 134 139

**Model mean** mm/day 62 95 111 120 128 133

Table 5. Present and future magnitudes of extreme precipitation events.

% change 12.59 13.26 13.19 13.72 14.51 14.70

% change 11.80 12.70 12.90 12.87 14.31 14.86

% change 9.90 10.96 10.94 10.86 11.82 12.11

% change 16.11 20.86 22.45 23.21 24.98 25.69

% change 9.72 13.61 16.01 18.56 21.43 22.76

% change 15.94 19.44 19.48 22.15 23.75 24.37

% change 12.68 15.14 15.83 16.90 18.47 19.08


Table 4. Changes in the Midwest mean daily precipitation in 2040-2070

Fig. 5. Changes in mean daily precipitation in 2040-2070 (model mean).

#### **4.2.2 Extreme precipitation events**

348 Greenhouse Gases – Emission, Measurement and Management

**Models Mean daily precipitation (mm/d) % change** 

**CCSM\_MM5I** 2.87 9.51 **CCSM\_WRFG** 2.61 -0.44 **CGCM3\_CRCM** 2.79 6.59 **CGCM3\_RCM3** 2.89 10.11 **GFDL\_RCM3** 2.86 8.95 **HADCM3\_HRM3** 2.93 11.67 **Model mean** 2.82 7.73

Table 4. Changes in the Midwest mean daily precipitation in 2040-2070

Fig. 5. Changes in mean daily precipitation in 2040-2070 (model mean).

**Observed (1971-2000)** 2.62

**Future (2040-2070)** 

We used the mixed probability model (eq. 10) to project extreme precipitation events. We used the wet day frequency as an estimate for precipitation probability (*ppr*), and fit a Gamma distribution to the wet day precipitation data using maximum likelihood method. Applying the same methodology on the observed and bias-corrected future daily precipitation data, we derived the magnitudes of extreme events of return interval of 1, 5, 10, 15, 20 and 25 years for both present (1970-2000) and future (2040-2070). Table 5 summarized the results. All models show increase in precipitation depths of extreme events. In general, extreme precipitation events increase at greater magnitudes than mean precipitation. The increase is greater for more extreme events. For average results of all models, compared with 8% increase in mean precipitation, precipitation event of 1 year return interval is likely to increase by 13%, and that of 25 year could increase by 19%. There are variations among models in how they project spatial distribution of such changes. Our model evaluation suggests that mean of all models seems to capture the spatial variation of precipitation best. Based on this observation, we used mean values of all models to interpolate the spatial distribution of how extreme precipitation could change in the Midwest (Figure 7). It can be seen that in general the intensity for extreme precipitation increase disproportionately


Table 5. Present and future magnitudes of extreme precipitation events.

Projecting Changes in Extreme Precipitation in the Midwestern United States Using North

Fig. 7. Changes in extreme precipitation (2040-2070).

American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models 351

Fig. 6. Changes in mean daily precipitation in 2040-2070 from 6 RCMs

Fig. 6. Changes in mean daily precipitation in 2040-2070 from 6 RCMs

Fig. 7. Changes in extreme precipitation (2040-2070).

Projecting Changes in Extreme Precipitation in the Midwestern United States Using North

*Analysis* 43 (3): 299–314. doi:10.1016/S0167-9473(02)00250-5.

climate change. *Climatic Change* 42:243-283.

*Statistical Planning and Inference* 136: 193–198.

UN Environment Program, Nairobi, Kenya.

Cambridge University Press, New York.

change-midwest.html. Accessed Nov. 1, 2011.

2340.

124.

305.

311-312.

Florida.

Bulletin 71.

American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models 353

Elamir, EAH, Seheult, AH. 2003. "Trimmed L-moments". *Computational Statistics & Data* 

Groisman PY, Karl TR, Easterling DR, Knight RW, Jamason, PF, Hennessy, KJ, Suppiah, R.

Groisman PY, Knight RW, Easterling DR, Karl TR, Hegerl GC, Razuvaev VN. 2005. Trends in intense precipitation in the climate record. *Journal of Climate* 18:1326-1350. Guttman, NB, Hosking, JRM, Wallis, JR. 1993. Regional precipitation quantile values for the

Hennessy KJ, Gregory JM, Mitchel JFB. 1997. et al. 1995. Changes in daily precipitation under enhanced greenhouse conditions. *Climate Dynamics* 13: 667-680. Hosking JR. 1990. L-moments: analysis and estimation of distributions using

Hosking, JRM. 2006. "On the characterization of distributions by their L-moments". *Journal of* 

Huff, FA, Angel JR. 1992. *Rainfall Frequency Atlas of the Midwest*. Midwestern Climate Center

Intergovernmental Panel on Climate Change. 2000. Special Report on Emission Scenarios.

Kunkel KE. 2003. North American trends in extreme precipitation. *Natural Hazards* 29: 291–

Mearns LO, Gutowski WJ, Jones R, Leung L-Y, McGinnis S, Nunes AMB, and Qian, Y. 2009.

Meehl GA, Stocker TF, Collins WD, Friedlingstein P, Gaye AT, Gregory JM, Kitoh A. Knutti

Min SK, Zhang X, Zwiers FW, Hegerl GC. 2011 Human contribution to more-intense

Rao, AR, Hamed KH. 2000 *Flood Frequency Analysis.* CRC Press, Boca Raton,

Ulrych, T J, Velis, DR, Woodbury, AD, Sacchi, MD. 2000. "L-moments and C-moments".

*Stochastic Environmental Research and Risk Assessment*, 14 (1), 50–68. Union of Concerned Scientists. 2009. Confronting Climate Change in the U.S. Midwest. http://www.ucsusa.org/global\_warming/science\_and\_impacts/impacts/climate-

precipitation. *Nature* 470:378-381. doi:10.1038/nature09763.

A regional climate change assessment program for North America. *EOS* 90(36):

R, Murphy JM, Noda A, Raper SCB, Waterson IG, Weaver AJ, Zhao Z-C. 2007. Global climate projections. In: *Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change.* Edited by S. Solomon et al. pp. 749-845.

Page, CM, Wibig, J, Fortuniak, K, Razuvaev VN, Douglas A. Forland, E, Zhai PM. 1999. Changes in the probability of heavy precipitation: Important indicators of

Continental United States computed from L-moments. *Journal of Climate*, 6:2326-

linear combinations of order statistics. *Journal of Royal Statistical Society B*, 52: 105-

more for more extreme events. Similar to the pattern of mean precipitation change, there seems to be more increase in extreme events in the northern part of the region. Since a lot of extreme precipitation events occur in spring, this increase could be attributed to decreased ice cover during the winter, early melt of lake ice, and consequent increased evaporation of the great lakes under a warmer climate.

### **5. Conclusion**

In this study, we first evaluated the performance of NARCCAP regional climate models in terms of its ability to capture the frequency distribution of daily precipitation data in the Midwest region of the United States. We found all models have biases. In general, they tend to overestimate small events, and underestimate large events. Based on such evaluation, we corrected model biases based on quantile-mapping method. We then used the bias-corrected future daily precipitation output from NARCCAP RCMs to project future changes in precipitation using a mixed probability model based on Gamma distribution. Based on the models, on average, the mean precipitation could increase by 7.7% by 2040-2070 for the Midwestern Region. Magnitudes of extreme precipitation are likely to increase more than average. The increase could be disproportionately larger for more extreme precipitation events. Spatially, northern part of the study area could see more increase in both mean and extreme precipitation, likely due to increased evaporation of the Great Lakes from a combination of higher temperature and less ice cover both spatially and temporally under a warmer climate.

#### **6. References**


more for more extreme events. Similar to the pattern of mean precipitation change, there seems to be more increase in extreme events in the northern part of the region. Since a lot of extreme precipitation events occur in spring, this increase could be attributed to decreased ice cover during the winter, early melt of lake ice, and consequent increased evaporation of

In this study, we first evaluated the performance of NARCCAP regional climate models in terms of its ability to capture the frequency distribution of daily precipitation data in the Midwest region of the United States. We found all models have biases. In general, they tend to overestimate small events, and underestimate large events. Based on such evaluation, we corrected model biases based on quantile-mapping method. We then used the bias-corrected future daily precipitation output from NARCCAP RCMs to project future changes in precipitation using a mixed probability model based on Gamma distribution. Based on the models, on average, the mean precipitation could increase by 7.7% by 2040-2070 for the Midwestern Region. Magnitudes of extreme precipitation are likely to increase more than average. The increase could be disproportionately larger for more extreme precipitation events. Spatially, northern part of the study area could see more increase in both mean and extreme precipitation, likely due to increased evaporation of the Great Lakes from a combination of higher temperature and less ice cover both spatially and temporally under a

Allan RP, Soden B J. 2008. Atmospheric warming and the amplification of precipitation

Buishand, TA. 1978. Some remarks on the use of daily rainfall models. *Journal of Hydrology*,

Crutcher, HL, McKay, GF, Fullbright, DC. 1977. A note on a gamma distribution

Cubasch U, Boer GJ, Stouffer RJ, Dix M, Noda A, Senior CA, Raper S, Yap KS. 2001.

Easterlin DR and Karl TR. 2000 Potential consequences of climate variability and change for

Easterling DR, Evans JL, Ya P, Groisman PY, Karl TR, Kunkel KE, Ambenje P. 2000.

computer program and computer produced graphs. NOAA Technical Report EDS24, Environmental Data Service, NOAA, Department of Commerce,

Projections of future climate change, in *Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change.* Edited by J.T. Houghton et al., pp. 525-582. Cambridge

the Midwestern United States. In *Climate Change Impacts on the U.S.: The Potential Consequences of Climate Variability and Change*, ed. National Climate Assessment

Observed variability and trends in extreme climate events: a brief review. *Bulletin of* 

the great lakes under a warmer climate.

extremes. *Science* 321:1481–1484.

University Press, New York.

*the American Meteorological Society* 81(3): 417-425.

**5. Conclusion** 

warmer climate.

**6. References** 

36:295-308.

Washington DC.

Team, p 167-188.


**17**

*Japan* 

**Future Changes in the Quasi-Biennial** 

**by a Chemistry-Climate Model**

Kiyotaka Shibata and Makoto Deushi *Meteorological Research Institute, Tsukuba* 

**Oscillation Under a Greenhouse Gas Increase and Ozone Recovery in Transient Simulations** 

In the equatorial stratosphere, the zonal wind direction changes from westward to eastward and vice versa at intervals of from less than two to about three years and centered at about 28 months (Reed et al., 1961; Veryard & Ebdon, 1961). This phenomenon is called the quasibiennial oscillation (QBO) from the length of its period (Angell & Korshover, 1964). The QBO initiates in the upper stratosphere, then descends gradually at a rate of about 1 km per month with an amplitude of about 20 ms-1, and diminishes near the tropopause (~16 km). The QBO structure for zonal wind and temperature is highly symmetric in longitude and latitude with a half-width of about 12° latitude, and the QBO dominates the annual and semiannual cycles in the equatorial stratosphere (e.g., Baldwin et al., 2001; Pascoe et al., 2005). The QBO can also be seen in changes in the abundances of ozone (e.g., Angell & Korshover, 1964; Randel & Wu, 1996) and other trace gases (e.g., Luo et al., 1997; Dunkerton, 2001) because their transport and chemistry are strongly dependent on the wind and

The QBO plays a very important role in the stratosphere not only because the QBO is the largest variation in the equatorial stratosphere but also because its effects significantly extend to the northern extratropical stratosphere (e.g. Holton & Tan, 1980, 1982), with the result that the stratospheric polar vortex tends to be weaker (stronger) during the QBO easterly (westerly) phase in the lower stratosphere. In addition, the QBO effect can also be seen in the extratropical troposphere at or near the surface not only in the Northern Hemisphere (NH) (e.g., Coughlin & Tung, 2001; Marshall & Scaife, 2009) but also in the Southern Hemisphere (SH) (Kuroda & Yamazaki, 2010), indicating that it has dynamical impacts on the tropospheric circulation even though the physical mechanism is still

The QBO effect on the polar vortex, which has been confirmed by many statistical analyses, is referred to as the Holton-Tan relation (e.g., Labitzke, 1982; Naito & Hirota, 1997; Lu et al., 2008). The Holton-Tan relation is ascribed to modulation by the QBO of the strength of the effective waveguide for the mid-latitude planetary waves propagating through the winter

**1. Introduction** 

temperature fields, respectively.

unknown.

Zhang X, Zwiers FW, Hegerl BC, Lambert FG, Gillett NP, Solomon S, Stott PA, Nozawa T. 2007. Detection of human influence on twentieth-century precipitation trends. *Nature* 448: 461-465. doi:10.1038/nature06025

## **Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model**

Kiyotaka Shibata and Makoto Deushi *Meteorological Research Institute, Tsukuba Japan* 

## **1. Introduction**

354 Greenhouse Gases – Emission, Measurement and Management

Zhang X, Zwiers FW, Hegerl BC, Lambert FG, Gillett NP, Solomon S, Stott PA, Nozawa T.

*Nature* 448: 461-465. doi:10.1038/nature06025

2007. Detection of human influence on twentieth-century precipitation trends.

In the equatorial stratosphere, the zonal wind direction changes from westward to eastward and vice versa at intervals of from less than two to about three years and centered at about 28 months (Reed et al., 1961; Veryard & Ebdon, 1961). This phenomenon is called the quasibiennial oscillation (QBO) from the length of its period (Angell & Korshover, 1964). The QBO initiates in the upper stratosphere, then descends gradually at a rate of about 1 km per month with an amplitude of about 20 ms-1, and diminishes near the tropopause (~16 km). The QBO structure for zonal wind and temperature is highly symmetric in longitude and latitude with a half-width of about 12° latitude, and the QBO dominates the annual and semiannual cycles in the equatorial stratosphere (e.g., Baldwin et al., 2001; Pascoe et al., 2005). The QBO can also be seen in changes in the abundances of ozone (e.g., Angell & Korshover, 1964; Randel & Wu, 1996) and other trace gases (e.g., Luo et al., 1997; Dunkerton, 2001) because their transport and chemistry are strongly dependent on the wind and temperature fields, respectively.

The QBO plays a very important role in the stratosphere not only because the QBO is the largest variation in the equatorial stratosphere but also because its effects significantly extend to the northern extratropical stratosphere (e.g. Holton & Tan, 1980, 1982), with the result that the stratospheric polar vortex tends to be weaker (stronger) during the QBO easterly (westerly) phase in the lower stratosphere. In addition, the QBO effect can also be seen in the extratropical troposphere at or near the surface not only in the Northern Hemisphere (NH) (e.g., Coughlin & Tung, 2001; Marshall & Scaife, 2009) but also in the Southern Hemisphere (SH) (Kuroda & Yamazaki, 2010), indicating that it has dynamical impacts on the tropospheric circulation even though the physical mechanism is still unknown.

The QBO effect on the polar vortex, which has been confirmed by many statistical analyses, is referred to as the Holton-Tan relation (e.g., Labitzke, 1982; Naito & Hirota, 1997; Lu et al., 2008). The Holton-Tan relation is ascribed to modulation by the QBO of the strength of the effective waveguide for the mid-latitude planetary waves propagating through the winter

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

N2O, and CO2 is included in CCMs.

as frequently as in the southern polar region.

otherwise reduce the confidence level of each detected signal.

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 357

period response may reside partly in the change in the mean tropical upwelling (i.e., the Brewer-Dobson circulation (BDC)), which is weakening in the simulation by Giorgetta & Doege (2005) and strengthening in that by Kawatani et al. (2011). Either way, it indicates the importance of the vertical momentum transport to the future QBO momentum budget.

Climate change is driven not only by increases in GHGs but also by changes in the ozone layer, because the stratospheric thermal structure is substantially affected by the stratospheric ozone through solar heating. The quasi-global (60S-60N) stratospheric ozone abundances decreased from the 1970s to the mid-1990s (e.g., WMO, 2011) because of increases in emissions of anthropogenic ozone-depleting substances (ODSs) such as Chlorofluorocarbons (CFCs) and halons. Since the mid-1990s, stratospheric ozone has been stable or slightly increasing (e.g., WMO, 2011) because of international regulation of ODS production. ODS-originated halogens cause severe ozone depletion through the gas-phase reactions in the upper stratosphere, which in turn induces stratospheric cooling (e.g., WMO, 2007). The CO2 increase also causes cooling in the stratosphere through emission of longwave radiation to space; in particular, it causes strong cooling in the upper stratosphere, which in turn increases ozone abundances through chemical reactions. Hence, radiation and chemistry interaction between ozone and CO2 (e.g., Shepherd & Jonsson, 2008) should be included in a model to obtain realistic simulations of climate change, at least, in the stratosphere. Accordingly, CCMs have advantage over GCMs because the interaction between dynamics, radiation, and chemistry for radiatively active gases such as O3, CH4,

In the polar regions, ODS increases led to severe ozone depletion in spring through the heterogeneous reactions on the polar stratospheric clouds (PSCs), which manifested as the Antarctic ozone hole (Chubachi, 1984; Farman et al., 1985), defined as an area with less than 220 Dobson Units (DU) of total ozone. The ozone hole developed rapidly from about 1980 to the mid-1990s; since then, the area with severely depleted ozone has remained nearly constant, except during a few years, when wave activity was unusually active (e.g., WMO, 2011). Severe ozone depletion also occurs in the northern polar region if the winter stratospheric temperature is cold enough for PSCs to form (e.g. Rex et al., 2006), though not

Here, to address future QBO behavior under climate changes due to increasing GHGs and decreasing ODSs, Meteorological Research Institute (MRI) CCM transient simulation data from 1960 to 2100 are analyzed focusing on the QBO trend and its forcings. The future QBO effect on the polar vortex, that is, the future Holton-Tan relation, in MRI-CCM simulations is described in another paper (Naoe & Shibata, 2012). Three simulations, using the three scenarios of CCMVal phase 2 (CCMVal-2, Eyring et al., 2008; SPARC, 2010) for the future projection of the stratospheric ozone and related species, are used. The scenarios used in this study specify only anthropogenic forcings of GHGs/SST and/or ODSs, and include no natural forcings, that is, neither the 11-year solar cycle nor volcanic aerosols. The simulation data are thereby free from the complicated overlapping effects among the QBO, solar cycle, and volcanic aerosols (e.g., Lee & Smith, 2003; Yamashita et al., 2011), which would

The rest of this paper is organized as follows. Section 2 describes the model and simulation conditions, and section 3 presents past and future climatologies of zonal wind and temperature. Section 4 describes the QBO and its forcings in the past, and section 5 treats the

stratosphere (e.g., Holton & Tan, 1980, 1982). Although the Holton-Tan relation has been reproduced by general circulation models (GCM) (e.g., Niwano & Takahashi, 1998; Hamilton 1998, Anstey et al., 2010) and chemistry-climate models (CCM) (Naoe & Shibata, 2010), the details of its mechanism are not yet well understood because the Eliassen-Palm (EP) flux diagnostics do not show a more poleward propagation in the mid-latitude stratosphere during the easterly QBO phase. Rather, planetary waves propagate more equatorward as well as more upward during the easterly QBO phase in both 25-year (1980- 2004) data of the reanalysis and a CCM ensemble simulation (Naoe & Shibata, 2010) under the REF-1 scenario of CCM validation phase 1 (CCMVal-1) activity for stratospheric processes and their role in climate (SPARC) (Eyring et al., 2005). Furthermore, the polar vortex intensity depends not only on the QBO but also on the 11-year solar cycle (e.g., Labitzke & van Loon, 1999); as a result, the Holton-Tan relation varies irregularly or nonstationarily with the solar activity (e.g., Gray et al., 2001; Labitzke, et al., 2006; Lu et al., 2009), making the circumstances of the Holton-Tan relation more complicated than what Holton & Tan (1980, 1982) first suggested.

The mechanism of the QBO is generally explained with a wave-mean flow interaction, wherein gravity waves and/or equatorial waves excited in the troposphere deposit their westward or eastward momentum, depending on the mean flow direction during their vertical propagation in the stratosphere, as has been demonstrated with numerical models (e.g., Lindzen & Holton 1968; Plumb 1977). This mechanism, which has also been reproduced in a laboratory experiment of Plumb & McEwan (1978), is one reason why the oscillation does not have a narrow annual or semiannual period but a peculiar, broad 28 month period. The QBO is hence strongly dependent on the wave source and the propagation conditions, that is, the wave generation in the troposphere and the background wind field in the troposphere and stratosphere.

Under the changing climate resulting from increases in greenhouse gases (GHG), the stratosphere is projected to continue to cool because of strong CO2 longwave emission, in a sharp contrast to the tropospheric warming caused by the sea surface temperature (SST) rise driven by the enhancement of downward longwave radiation by CO2 and H2O. This raises the question as to how the QBO might behave in the future under global warming caused by increased GHGs. To date, two equilibrium GCM simulations under doubled CO2 levels have projected the future QBO. Giorgetta & Doege (2005) investigated the effect of variations in the source strength of parameterized gravity wave forcing (GWF) with a middle atmosphere GCM (Roeckner et al., 2003) at the T42L90 resolution and showed that the QBO period is shortened when the source strength of GWF increases. The future QBO periods are 26, 22, and 17 months for a 0%, 10%, and 20% increase in source strength, respectively, against a 29 month period under the current CO2 conditions. Kawatani et al. (2011) carried out a similar experiment with a GCM (Hasumi & Emori, 2004) at the T106L72 resolution. They did not employ a parameterized GWF, so that the wave forcing of the QBO was by resolved waves alone. In their simulation, the future QBO has a period that is 1-5 months longer than that of the control QBO, which has a period of 24 months. Moreover, the future QBO amplitude is smaller than the control; in particular, the amplitude is much smaller below 50 hPa. Thus, the two GCM results show an opposite change in the QBO period, although both Giorgetta & Doege (2005) and Kawatani et al. (2011) attributed the future QBO change mainly to changes in the residual circulation (mean tropical upwelling) and to the enhanced wave forcings. Aside from the use of a parameterized or resolved GWF, the difference in the QBO

stratosphere (e.g., Holton & Tan, 1980, 1982). Although the Holton-Tan relation has been reproduced by general circulation models (GCM) (e.g., Niwano & Takahashi, 1998; Hamilton 1998, Anstey et al., 2010) and chemistry-climate models (CCM) (Naoe & Shibata, 2010), the details of its mechanism are not yet well understood because the Eliassen-Palm (EP) flux diagnostics do not show a more poleward propagation in the mid-latitude stratosphere during the easterly QBO phase. Rather, planetary waves propagate more equatorward as well as more upward during the easterly QBO phase in both 25-year (1980- 2004) data of the reanalysis and a CCM ensemble simulation (Naoe & Shibata, 2010) under the REF-1 scenario of CCM validation phase 1 (CCMVal-1) activity for stratospheric processes and their role in climate (SPARC) (Eyring et al., 2005). Furthermore, the polar vortex intensity depends not only on the QBO but also on the 11-year solar cycle (e.g., Labitzke & van Loon, 1999); as a result, the Holton-Tan relation varies irregularly or nonstationarily with the solar activity (e.g., Gray et al., 2001; Labitzke, et al., 2006; Lu et al., 2009), making the circumstances of the Holton-Tan relation more complicated than what

The mechanism of the QBO is generally explained with a wave-mean flow interaction, wherein gravity waves and/or equatorial waves excited in the troposphere deposit their westward or eastward momentum, depending on the mean flow direction during their vertical propagation in the stratosphere, as has been demonstrated with numerical models (e.g., Lindzen & Holton 1968; Plumb 1977). This mechanism, which has also been reproduced in a laboratory experiment of Plumb & McEwan (1978), is one reason why the oscillation does not have a narrow annual or semiannual period but a peculiar, broad 28 month period. The QBO is hence strongly dependent on the wave source and the propagation conditions, that is, the wave generation in the troposphere and the background

Under the changing climate resulting from increases in greenhouse gases (GHG), the stratosphere is projected to continue to cool because of strong CO2 longwave emission, in a sharp contrast to the tropospheric warming caused by the sea surface temperature (SST) rise driven by the enhancement of downward longwave radiation by CO2 and H2O. This raises the question as to how the QBO might behave in the future under global warming caused by increased GHGs. To date, two equilibrium GCM simulations under doubled CO2 levels have projected the future QBO. Giorgetta & Doege (2005) investigated the effect of variations in the source strength of parameterized gravity wave forcing (GWF) with a middle atmosphere GCM (Roeckner et al., 2003) at the T42L90 resolution and showed that the QBO period is shortened when the source strength of GWF increases. The future QBO periods are 26, 22, and 17 months for a 0%, 10%, and 20% increase in source strength, respectively, against a 29 month period under the current CO2 conditions. Kawatani et al. (2011) carried out a similar experiment with a GCM (Hasumi & Emori, 2004) at the T106L72 resolution. They did not employ a parameterized GWF, so that the wave forcing of the QBO was by resolved waves alone. In their simulation, the future QBO has a period that is 1-5 months longer than that of the control QBO, which has a period of 24 months. Moreover, the future QBO amplitude is smaller than the control; in particular, the amplitude is much smaller below 50 hPa. Thus, the two GCM results show an opposite change in the QBO period, although both Giorgetta & Doege (2005) and Kawatani et al. (2011) attributed the future QBO change mainly to changes in the residual circulation (mean tropical upwelling) and to the enhanced wave forcings. Aside from the use of a parameterized or resolved GWF, the difference in the QBO

Holton & Tan (1980, 1982) first suggested.

wind field in the troposphere and stratosphere.

period response may reside partly in the change in the mean tropical upwelling (i.e., the Brewer-Dobson circulation (BDC)), which is weakening in the simulation by Giorgetta & Doege (2005) and strengthening in that by Kawatani et al. (2011). Either way, it indicates the importance of the vertical momentum transport to the future QBO momentum budget.

Climate change is driven not only by increases in GHGs but also by changes in the ozone layer, because the stratospheric thermal structure is substantially affected by the stratospheric ozone through solar heating. The quasi-global (60S-60N) stratospheric ozone abundances decreased from the 1970s to the mid-1990s (e.g., WMO, 2011) because of increases in emissions of anthropogenic ozone-depleting substances (ODSs) such as Chlorofluorocarbons (CFCs) and halons. Since the mid-1990s, stratospheric ozone has been stable or slightly increasing (e.g., WMO, 2011) because of international regulation of ODS production. ODS-originated halogens cause severe ozone depletion through the gas-phase reactions in the upper stratosphere, which in turn induces stratospheric cooling (e.g., WMO, 2007). The CO2 increase also causes cooling in the stratosphere through emission of longwave radiation to space; in particular, it causes strong cooling in the upper stratosphere, which in turn increases ozone abundances through chemical reactions. Hence, radiation and chemistry interaction between ozone and CO2 (e.g., Shepherd & Jonsson, 2008) should be included in a model to obtain realistic simulations of climate change, at least, in the stratosphere. Accordingly, CCMs have advantage over GCMs because the interaction between dynamics, radiation, and chemistry for radiatively active gases such as O3, CH4, N2O, and CO2 is included in CCMs.

In the polar regions, ODS increases led to severe ozone depletion in spring through the heterogeneous reactions on the polar stratospheric clouds (PSCs), which manifested as the Antarctic ozone hole (Chubachi, 1984; Farman et al., 1985), defined as an area with less than 220 Dobson Units (DU) of total ozone. The ozone hole developed rapidly from about 1980 to the mid-1990s; since then, the area with severely depleted ozone has remained nearly constant, except during a few years, when wave activity was unusually active (e.g., WMO, 2011). Severe ozone depletion also occurs in the northern polar region if the winter stratospheric temperature is cold enough for PSCs to form (e.g. Rex et al., 2006), though not as frequently as in the southern polar region.

Here, to address future QBO behavior under climate changes due to increasing GHGs and decreasing ODSs, Meteorological Research Institute (MRI) CCM transient simulation data from 1960 to 2100 are analyzed focusing on the QBO trend and its forcings. The future QBO effect on the polar vortex, that is, the future Holton-Tan relation, in MRI-CCM simulations is described in another paper (Naoe & Shibata, 2012). Three simulations, using the three scenarios of CCMVal phase 2 (CCMVal-2, Eyring et al., 2008; SPARC, 2010) for the future projection of the stratospheric ozone and related species, are used. The scenarios used in this study specify only anthropogenic forcings of GHGs/SST and/or ODSs, and include no natural forcings, that is, neither the 11-year solar cycle nor volcanic aerosols. The simulation data are thereby free from the complicated overlapping effects among the QBO, solar cycle, and volcanic aerosols (e.g., Lee & Smith, 2003; Yamashita et al., 2011), which would otherwise reduce the confidence level of each detected signal.

The rest of this paper is organized as follows. Section 2 describes the model and simulation conditions, and section 3 presents past and future climatologies of zonal wind and temperature. Section 4 describes the QBO and its forcings in the past, and section 5 treats the

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

present condition.

for 2015.

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 359

evolving forcings of GHGs, ODSs, and SST and with fixed solar minimum and background aerosol conditions. Fixed conditions are used because plausible specification of future solar variability and volcanic aerosols is very difficult. SST/sea-ice data for the entire period from the spin-up (covering the 1950s) to 2100 are taken from a transient simulation (1%/year CO2 increase) by the MRI coupled ocean-atmosphere model MRI-CGCM2.3.2 (Yukimoto et al., 2006), in which the atmospheric GCM is nearly the same as in MRI-CCM. Note that even the past simulation from the 1950s uses not observed but simulated SST/sea-ice data in order to avoid gap in SST/sea-ice data, stemming from model SST biases, in the vicinity of the

GHG and ODS abundances are assumed to be geographically uniform and specified at the surface (Eyring et al., 2008). The future scenarios for GHGs and ODSs are taken from the Special Report on Emissions Scenarios (SRES) A1B GHG scenario (IPCC, 2000) and the adjusted A1 halogen scenario. The A1B GHG scenario represents a future world of very rapid economic growth, a global population that peaks in the mid-twenty-first century and declines thereafter, and the rapid introduction of new and more efficient technologies, balanced across all GHG sources. On the other hand, the adjusted A1 halogen scenario includes the earlier phase-out of hydrochlorofluorocarbons (HCFCs) that was agreed to by the Parties to the Montreal Protocol in 2007, with other species such as CFCs, halons, and other non-HCFCs remaining identical to the original A1 scenario (WMO, 2007). In the original A1 scenario, future production of ODSs is determined from the Montreal Protocol limitations or the most recent annual estimates, whichever is less; future emissions are determined in order to yield banks consistent with those of IPCC/TEAP (2005); and the future annual fractional bank release is adjusted so as to attain the IPCC/TEAP (2005) bank

Beside the two reference simulations, two other simulations were performed by fixing either ODSs or GHGs at those values at 1960 levels; these simulations correspond to two of the eight sensitivity simulations of CCMVal-2 (Eyring et al., 2008). The first simulation (SCN-B2b, i.e., fixed halogens) uses fixed ODSs and the same time-evolving GHGs and SST/seaice data as REF-B2, whereas the second (SCN-B2c, i.e., no-climate-change) uses fixed GHGs and annually repeating SST/sea-ice data and the same time-evolving ODSs as REF-B2. Note

GHGs: A1B ODSs: A1

Fig. 1. Evolutions of GHGs and ODSs. Time evolutions of the (a) SRES A1B GHGs scenario for CO2 (ppmv), CH4 (ppmv), and N2O (10 pptv) from 1950 to 2100, and (b) total halogens as CCly (ppbv) and total bromine as CBry (10 pptv) of the SRES A1 adjusted ODSs scenario.

CCly (ppbv)

CBry (pptv)

that the SST/sea ice at 1960 levels is the 10-year average centered on 1960.

CO2

CH4

N2O

evolutions of the QBO and of its forcings. The results are discussed in section 6, and section 7 is a conclusion.

## **2. Model simulation and data**

### **2.1 MRI-CCM**

MRI-CCM is an upgraded version of the MRI chemistry transport model (Shibata et al., 2005), and its details are described by Shibata & Deushi (2008a, b) and references therein. The dynamics module of MRI-CCM is a spectral global model with triangular truncation, a maximum total wavenumber 42 (about 2.8° by 2.8° in longitude and latitude), and 68 layers in the eta-coordinate extending from the surface to 0.01 hPa (about 80 km). The vertical spacing is about 500 m in the stratosphere between 100 and 10 hPa, with tapering below and above. Non-orographic gravity wave (GW) forcing by Hines (1997) is incorporated, and GW source strength is enhanced symmetrically with respect to latitudes between 30°S and 30°N by superimposing a Gaussian function (15 e-folding latitude) source (0.7 ms-1) on an isotropic source (2.3 ms-1). The gravity wave spectrum is isotropically launched toward eight equally spaced azimuths (north, northwest, west etc) at the lowest level with horizontal wavenumber k\* = 5.010-6 (m-1).

Biharmonic (∆2) horizontal diffusion is weakened only in the middle atmosphere from that in the troposphere in order to spontaneously reproduce the QBO in zonal wind while minimizing the changes in the troposphere (Shibata & Deushi, 2005a, b). The e-folding time at the maximum total wavenumber (n = 42) is 18 h below 100 hPa, 100 h above 150 hPa, and intermediate values in between. This e-folding time of 100 h in the middle atmosphere results in an e-folding time of about 3.6 years at n = 10; almost all of the power (>99%) is included within this total wavenumber for a representative equatorial wave of Gaussian meridional structure centered at the equator with a 15° e-folding latitude (Shibata & Deushi, 2005a). In addition, vertical diffusion is not applied in the middle atmosphere in order to keep the sharp vertical shear in the QBO.

The chemistry-transport module employs a hybrid semi-Lagrangian transport scheme compatible with the continuity equation to satisfy the mass conservation. The horizontal form is an ordinary semi-Lagrangian scheme, and the vertical form is equivalent to a massconserving flux form in the transformed pressure coordinate specified by the vertical velocity. The vertical procedure employs the piecewise rational method (Xiao & Peng, 2004), and the horizontal procedure uses quintic interpolation. The chemistry scheme treats 36 long-lived species including 7 families, and 15 short-lived species with 80 gas-phase reactions, 35 photochemical reactions and 9 heterogeneous reactions on PSCs and sulfate aerosols. Since MRI-CCM does not include all of the chlorine species, the chlorine species (CCl4, CFCl3, CF2Cl2, and CH3Cl) are lumped, except for BrCl, such that the modeled total chlorine corresponds to the input total chlorine at the surface.

#### **2.2 Simulations with the CCMVal-2 scenarios**

Two scenarios in CCMVal-2 are used for reference simulations (Eyring et al., 2008): REF-B1 is used for simulations of the past, from 1960 to 2006, with all observed forcings of GHGs, ODSs, SST, solar spectral irradiance variability, and (volcanic) aerosols in the stratosphere, and REF-B2 is used for both past and future simulations, from 1960 to 2100, with time-

evolutions of the QBO and of its forcings. The results are discussed in section 6, and section

MRI-CCM is an upgraded version of the MRI chemistry transport model (Shibata et al., 2005), and its details are described by Shibata & Deushi (2008a, b) and references therein. The dynamics module of MRI-CCM is a spectral global model with triangular truncation, a maximum total wavenumber 42 (about 2.8° by 2.8° in longitude and latitude), and 68 layers in the eta-coordinate extending from the surface to 0.01 hPa (about 80 km). The vertical spacing is about 500 m in the stratosphere between 100 and 10 hPa, with tapering below and above. Non-orographic gravity wave (GW) forcing by Hines (1997) is incorporated, and GW source strength is enhanced symmetrically with respect to latitudes between 30°S and 30°N by superimposing a Gaussian function (15 e-folding latitude) source (0.7 ms-1) on an isotropic source (2.3 ms-1). The gravity wave spectrum is isotropically launched toward eight equally spaced azimuths (north, northwest, west etc) at the lowest level with horizontal

Biharmonic (∆2) horizontal diffusion is weakened only in the middle atmosphere from that in the troposphere in order to spontaneously reproduce the QBO in zonal wind while minimizing the changes in the troposphere (Shibata & Deushi, 2005a, b). The e-folding time at the maximum total wavenumber (n = 42) is 18 h below 100 hPa, 100 h above 150 hPa, and intermediate values in between. This e-folding time of 100 h in the middle atmosphere results in an e-folding time of about 3.6 years at n = 10; almost all of the power (>99%) is included within this total wavenumber for a representative equatorial wave of Gaussian meridional structure centered at the equator with a 15° e-folding latitude (Shibata & Deushi, 2005a). In addition, vertical diffusion is not applied in the middle atmosphere in order to

The chemistry-transport module employs a hybrid semi-Lagrangian transport scheme compatible with the continuity equation to satisfy the mass conservation. The horizontal form is an ordinary semi-Lagrangian scheme, and the vertical form is equivalent to a massconserving flux form in the transformed pressure coordinate specified by the vertical velocity. The vertical procedure employs the piecewise rational method (Xiao & Peng, 2004), and the horizontal procedure uses quintic interpolation. The chemistry scheme treats 36 long-lived species including 7 families, and 15 short-lived species with 80 gas-phase reactions, 35 photochemical reactions and 9 heterogeneous reactions on PSCs and sulfate aerosols. Since MRI-CCM does not include all of the chlorine species, the chlorine species (CCl4, CFCl3, CF2Cl2, and CH3Cl) are lumped, except for BrCl, such that the modeled total

Two scenarios in CCMVal-2 are used for reference simulations (Eyring et al., 2008): REF-B1 is used for simulations of the past, from 1960 to 2006, with all observed forcings of GHGs, ODSs, SST, solar spectral irradiance variability, and (volcanic) aerosols in the stratosphere, and REF-B2 is used for both past and future simulations, from 1960 to 2100, with time-

7 is a conclusion.

**2.1 MRI-CCM** 

**2. Model simulation and data** 

wavenumber k\* = 5.010-6 (m-1).

keep the sharp vertical shear in the QBO.

chlorine corresponds to the input total chlorine at the surface.

**2.2 Simulations with the CCMVal-2 scenarios** 

evolving forcings of GHGs, ODSs, and SST and with fixed solar minimum and background aerosol conditions. Fixed conditions are used because plausible specification of future solar variability and volcanic aerosols is very difficult. SST/sea-ice data for the entire period from the spin-up (covering the 1950s) to 2100 are taken from a transient simulation (1%/year CO2 increase) by the MRI coupled ocean-atmosphere model MRI-CGCM2.3.2 (Yukimoto et al., 2006), in which the atmospheric GCM is nearly the same as in MRI-CCM. Note that even the past simulation from the 1950s uses not observed but simulated SST/sea-ice data in order to avoid gap in SST/sea-ice data, stemming from model SST biases, in the vicinity of the present condition.

GHG and ODS abundances are assumed to be geographically uniform and specified at the surface (Eyring et al., 2008). The future scenarios for GHGs and ODSs are taken from the Special Report on Emissions Scenarios (SRES) A1B GHG scenario (IPCC, 2000) and the adjusted A1 halogen scenario. The A1B GHG scenario represents a future world of very rapid economic growth, a global population that peaks in the mid-twenty-first century and declines thereafter, and the rapid introduction of new and more efficient technologies, balanced across all GHG sources. On the other hand, the adjusted A1 halogen scenario includes the earlier phase-out of hydrochlorofluorocarbons (HCFCs) that was agreed to by the Parties to the Montreal Protocol in 2007, with other species such as CFCs, halons, and other non-HCFCs remaining identical to the original A1 scenario (WMO, 2007). In the original A1 scenario, future production of ODSs is determined from the Montreal Protocol limitations or the most recent annual estimates, whichever is less; future emissions are determined in order to yield banks consistent with those of IPCC/TEAP (2005); and the future annual fractional bank release is adjusted so as to attain the IPCC/TEAP (2005) bank for 2015.

Beside the two reference simulations, two other simulations were performed by fixing either ODSs or GHGs at those values at 1960 levels; these simulations correspond to two of the eight sensitivity simulations of CCMVal-2 (Eyring et al., 2008). The first simulation (SCN-B2b, i.e., fixed halogens) uses fixed ODSs and the same time-evolving GHGs and SST/seaice data as REF-B2, whereas the second (SCN-B2c, i.e., no-climate-change) uses fixed GHGs and annually repeating SST/sea-ice data and the same time-evolving ODSs as REF-B2. Note that the SST/sea ice at 1960 levels is the 10-year average centered on 1960.

Fig. 1. Evolutions of GHGs and ODSs. Time evolutions of the (a) SRES A1B GHGs scenario for CO2 (ppmv), CH4 (ppmv), and N2O (10 pptv) from 1950 to 2100, and (b) total halogens as CCly (ppbv) and total bromine as CBry (10 pptv) of the SRES A1 adjusted ODSs scenario.

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

stratosphere and a warm bias in the upper stratosphere at high latitudes.

(a) (b)

(c) (d)

(a) (b)

(c) (d)

temperatures and red colors higher temperatures.

negative (easterly) winds.

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 361

the stratopause. On the other hand, observed temperatures exhibit a local warm maximum at around 200 hPa and a local cold minimum at around 30 hPa, in the polar lower stratosphere, indicating that MRI-CCM simulations show a cold bias in the lower

Fig. 2. Annual- and zonal-mean temperature. Latitude-pressure cross sections for annualand zonal-mean temperature (K) in the past 30-year (1970-1999) are for (a) ERA-Interim, (b)

Fig. 3. Annual- and zonal-mean zonal wind. Same as Fig.2 except for zonal wind (ms-1). The contour interval is 5 ms-1, and blue colors represent positive (westerly) winds and red colors

B21, (c) Sb, and (d) Sc. The contour interval is 10 K, and blue colors represent lower

Figure 1 exhibits time evolutions of GHGs (CO2, CH4, and N2O) in the A1B scenario and ODSs (as total chlorine, CCly, and total bromine, CBry) in the A1 adjusted scenario for 1960 to 2100. CO2 and N2O abundances increase monotonically throughout the entire period, reaching maxima of about 700 ppmv and 370 ppbv, respectively, in 2100, whereas CH4 abundances attains a maximum of about 2.4 ppmv around the 2050s and then decreases to about 2.0 ppmv in 2100. We henceforth refer to this GHGs evolution simply as the GHG increase, neglecting the non-monotonic behavior of CH4. CCly increases rapidly from 0.8 ppbv around 1960 to a maximum of about 3.7 ppbv in the first half of the 1990s and then gradually decreases to about 1.3 ppmv in 2100. CBry also behaves qualitatively like CCly but it starts to increase much later (from about 1980) and attains a maximum of about 17 pptv around 2000.

To reduce noise stemming from internal variability among simulations, REF-B2 simulations were performed for three members, in which different initial conditions were taken from the spin-up runs of the REF-B1 simulations. The three members are henceforth referred to as B21, B22, and B23. On the other hand, single-member simulations were made for SCN-B2b and SCN-B2c, referred to as Sb and Sc, respectively. These simulations can be combined into different ensembles, depending on the key variables; for example, a realistic forcing (B21, B22, B23), or idealized forcing (Sb, Sc) ensemble, or GHG increase (B21, B22, B23, Sb) or time-varying ODS (B21, B22, B23, Sc) ensembles.

#### **2.3 Data processing**

As observed data, for comparison with past simulations, we used reanalysis data sets compiled by the European Centre for Medium-Range Weather Forecasts (ECMWF): namely, ERA40 (Uppala et al., 2005) from 1958 to 1988, and ERA-Interim (Dee et al., 2011) from 1989 to 2009. The two reanalysis data sets are smoothly merged such that the data for 1989 overlap, without any gap in any month, thus producing a long, continuous time series of data from 1958 to 2009, which is henceforth referred to simply as ERA-Interim. Monthlymean data are used in this study for both the observation and the simulations.

A Lanczos filter (e.g., Duchon, 1979) was applied to the monthly-mean time series to get band-passed data within the QBO period (20-40 months) or low-passed data in lowfrequency ranges, with periods much longer than the QBO period. The cut-off for the low frequency data period was set at 120 months. The running-mean was used as an alternative way of deriving low-frequency data. Student's *t*-test was used to evaluate the linear trend in the QBO and related quantities over the entire integration period, and results were considered statistically significant at the 95% or 99% confidence level.

## **3. Simulated global atmosphere**

#### **3.1 Zonal mean climatology**

Simulated annual- and zonal-mean temperature and zonal wind in the past 30 years (1970- 1999) are displayed for B21, Sb, and Sc in Figs. 2 and 3, together with observed values. Although the overall features of temperature are very similar between the observation and the simulations, the simulated temperatures show common biases in the mid- and highlatitude stratosphere in both the NH and SH, which are substantially due to biases during the winter. In the NH, the simulated temperatures show local minima at around 200 hPa in the polar lower stratosphere, above which temperatures increase monotonically upward to

Figure 1 exhibits time evolutions of GHGs (CO2, CH4, and N2O) in the A1B scenario and ODSs (as total chlorine, CCly, and total bromine, CBry) in the A1 adjusted scenario for 1960 to 2100. CO2 and N2O abundances increase monotonically throughout the entire period, reaching maxima of about 700 ppmv and 370 ppbv, respectively, in 2100, whereas CH4 abundances attains a maximum of about 2.4 ppmv around the 2050s and then decreases to about 2.0 ppmv in 2100. We henceforth refer to this GHGs evolution simply as the GHG increase, neglecting the non-monotonic behavior of CH4. CCly increases rapidly from 0.8 ppbv around 1960 to a maximum of about 3.7 ppbv in the first half of the 1990s and then gradually decreases to about 1.3 ppmv in 2100. CBry also behaves qualitatively like CCly but it starts to increase much later

To reduce noise stemming from internal variability among simulations, REF-B2 simulations were performed for three members, in which different initial conditions were taken from the spin-up runs of the REF-B1 simulations. The three members are henceforth referred to as B21, B22, and B23. On the other hand, single-member simulations were made for SCN-B2b and SCN-B2c, referred to as Sb and Sc, respectively. These simulations can be combined into different ensembles, depending on the key variables; for example, a realistic forcing (B21, B22, B23), or idealized forcing (Sb, Sc) ensemble, or GHG increase (B21, B22, B23, Sb) or

As observed data, for comparison with past simulations, we used reanalysis data sets compiled by the European Centre for Medium-Range Weather Forecasts (ECMWF): namely, ERA40 (Uppala et al., 2005) from 1958 to 1988, and ERA-Interim (Dee et al., 2011) from 1989 to 2009. The two reanalysis data sets are smoothly merged such that the data for 1989 overlap, without any gap in any month, thus producing a long, continuous time series of data from 1958 to 2009, which is henceforth referred to simply as ERA-Interim. Monthly-

A Lanczos filter (e.g., Duchon, 1979) was applied to the monthly-mean time series to get band-passed data within the QBO period (20-40 months) or low-passed data in lowfrequency ranges, with periods much longer than the QBO period. The cut-off for the low frequency data period was set at 120 months. The running-mean was used as an alternative way of deriving low-frequency data. Student's *t*-test was used to evaluate the linear trend in the QBO and related quantities over the entire integration period, and results were

Simulated annual- and zonal-mean temperature and zonal wind in the past 30 years (1970- 1999) are displayed for B21, Sb, and Sc in Figs. 2 and 3, together with observed values. Although the overall features of temperature are very similar between the observation and the simulations, the simulated temperatures show common biases in the mid- and highlatitude stratosphere in both the NH and SH, which are substantially due to biases during the winter. In the NH, the simulated temperatures show local minima at around 200 hPa in the polar lower stratosphere, above which temperatures increase monotonically upward to

mean data are used in this study for both the observation and the simulations.

considered statistically significant at the 95% or 99% confidence level.

(from about 1980) and attains a maximum of about 17 pptv around 2000.

time-varying ODS (B21, B22, B23, Sc) ensembles.

**3. Simulated global atmosphere** 

**3.1 Zonal mean climatology** 

**2.3 Data processing** 

the stratopause. On the other hand, observed temperatures exhibit a local warm maximum at around 200 hPa and a local cold minimum at around 30 hPa, in the polar lower stratosphere, indicating that MRI-CCM simulations show a cold bias in the lower stratosphere and a warm bias in the upper stratosphere at high latitudes.

Fig. 2. Annual- and zonal-mean temperature. Latitude-pressure cross sections for annualand zonal-mean temperature (K) in the past 30-year (1970-1999) are for (a) ERA-Interim, (b) B21, (c) Sb, and (d) Sc. The contour interval is 10 K, and blue colors represent lower temperatures and red colors higher temperatures.

Fig. 3. Annual- and zonal-mean zonal wind. Same as Fig.2 except for zonal wind (ms-1). The contour interval is 5 ms-1, and blue colors represent positive (westerly) winds and red colors negative (easterly) winds.

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

(c) (a) (b)

Fig. 5. Future changes in zonal wind. Same as Fig.4 except for zonal wind (ms-1). The contour interval is 1 ms-1, and blue colors represent positive (westerly/eastward)

change in the troposphere and stratosphere, consistent with the temperature results.

The GHG increase affects the mass exchange between the troposphere and the stratosphere, which can be evaluated as the upward mass flux across the tropopause. Most GCMs and

**3.3 Evolution of the Brewer-Dobson circulation and the ozone hole area** 

Zonal wind changes in the three B2 members and Sb display a very similar pattern in the polar night jet regions in both hemispheres and in the equatorial lower stratosphere. Westerly wind intensification extends from the upper region of the subtropical jet upward along the equatorward flank of the polar night jet axis, reaching a maximum at around 70 hPa and 40° in both hemispheres. In the SH, the westerly wind intensification also extends down to the surface at around 55°S, resulting in poleward spreading of the subtropical jet. In the tropics, westerly wind strengthening (eastward acceleration) occurs above 30 hPa and weakening (westward acceleration) at 50 hPa, below which very weak westerly wind strengthening occurs down to the surface. In the no-climate-change simulation Sc, on the other hand, the zonal wind change pattern is completely different. In the SH, weakening of the polar night jet and both weakening of the poleward and strengthening of the equatorward flank of the subtropical jet down to the surface occur, changes which are almost opposite to those in B2 and Sb in the troposphere above southern high latitudes. This change in the subtropical jet in Sc means that the equatorward shift of the jet is associated with ozone recovery (e.g., Son et al., 2008). In the NH, the poleward flank of the polar night jet is slightly intensified above the middle stratosphere. In the tropics, there is almost no

acceleration and red colors negative (easterly/westward) acceleration.

from those in the B2 members (not shown).

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 363

ozone abundances were projected to recover to or exceed pre-1980 levels globally in the future, except for the tropical lower stratosphere due to the increased inflow of ozone-poor tropospheric air mass (e.g., SPARC, 210). In other words, the upper stratospheric cooling in the past 1970-1999 was due to an ozone decrease caused by gas-phase reactions with reactive chlorine and bromine atoms photodissociated from ODSs under strong ultraviolet solar radiation, whereas the pattern in the southern high latitudes was associated with severe ozone depletion (i.e., the ozone hole) caused by heterogeneous reactions on PSCs (e.g., SPARC, 2010). This ODS-decrease effect is also included in the B2 simulations; thus, temperature changes similar to those in Sc can be obtained by subtracting the changes in Sb

Consistent with these temperature biases, the simulated zonal winds also show common biases in the subtropical jet and the polar night jet as a result of the thermal wind balance. The core velocity of the subtropical jet is too strong, and its separation from the polar night jet is too weak; that is, the zonal wind at around 50 hPa and 60°N is too strong, and the polar night jet in the upper stratosphere is too weak. In the SH, there are also cold biases in the mid- and high-latitude lower stratosphere at around 200 hPa. Similar to the NH, in the SH the simulated subtropical jet is stronger than the observed one, but the simulated polar night jet is weaker.

#### **3.2 Future climate change in temperature and zonal wind**

Figures 4 and 5 display the changes in the future 30-year period (2070-2999, 100 years later than the past 30-year period (1970-1999)) in annual- and zonal-mean temperature and zonal wind. During this future period, the ODS abundances are almost restored to pre-1970s levels (Fig. 1), so that the forcing is almost entirely due to GHGs alone. The temperature changes in the three B2 members (B21, B22, and B23) and in the fixed-halogen simulation Sb exhibit very similar CO2- (GHG-) induced patterns (e.g., Yukimoto et al., 2006; IPCC, 2007): warming in the troposphere through SST warming (indirect CO2 effect) and cooling in the stratosphere (direct CO2 effect) (Kodama et al., 2007). In addition, weak but substantial cooling is seen in the tropical lower stratosphere. The tropospheric warming reaches a maximum of about 4 K in the tropical upper troposphere, reflecting the enhanced latent heat release resulting from intensified convective activity caused by the SST rise. In the B2 simulations, the maximum stratospheric cooling, to about -6 K, in the upper stratosphere is caused both by enhanced CO2 terrestrial radiation and by recovered ozone solar radiation as described below.

Fig. 4. Future changes in temperature. Latitude-pressure cross sections of future changes (difference between 2070-2099 and 1970-1999) in annual- and zonal-mean temperature (K) for (a) B21, (b) Sb, and (b) Sc. The contour interval is 1 K for B21 and Sb and 0.5 K for Sc, and blue colors represent cooling and red colors warming.

In the no-climate-change simulation Sc, on the other hand, strong warming occurs in the upper stratosphere above about 5 hPa in both hemispheres, except in the southern high latitudes, where prominent warming occurs in the lower stratosphere at 300-30 hPa, centered at 100 hPa, and the cooling above 30 hPa. This temperature change pattern is completely different from the CO2-induced pattern and is due solely to the ODS decrease, or equivalently, the ozone recovery, because the average ODS concentration was higher in the past 30 years (1970-1999) than in the future 30 years (2070-2099) (Fig. 1). In fact, stratospheric

Consistent with these temperature biases, the simulated zonal winds also show common biases in the subtropical jet and the polar night jet as a result of the thermal wind balance. The core velocity of the subtropical jet is too strong, and its separation from the polar night jet is too weak; that is, the zonal wind at around 50 hPa and 60°N is too strong, and the polar night jet in the upper stratosphere is too weak. In the SH, there are also cold biases in the mid- and high-latitude lower stratosphere at around 200 hPa. Similar to the NH, in the SH the simulated subtropical jet is stronger than the observed one, but the simulated polar

Figures 4 and 5 display the changes in the future 30-year period (2070-2999, 100 years later than the past 30-year period (1970-1999)) in annual- and zonal-mean temperature and zonal wind. During this future period, the ODS abundances are almost restored to pre-1970s levels (Fig. 1), so that the forcing is almost entirely due to GHGs alone. The temperature changes in the three B2 members (B21, B22, and B23) and in the fixed-halogen simulation Sb exhibit very similar CO2- (GHG-) induced patterns (e.g., Yukimoto et al., 2006; IPCC, 2007): warming in the troposphere through SST warming (indirect CO2 effect) and cooling in the stratosphere (direct CO2 effect) (Kodama et al., 2007). In addition, weak but substantial cooling is seen in the tropical lower stratosphere. The tropospheric warming reaches a maximum of about 4 K in the tropical upper troposphere, reflecting the enhanced latent heat release resulting from intensified convective activity caused by the SST rise. In the B2 simulations, the maximum stratospheric cooling, to about -6 K, in the upper stratosphere is caused both by enhanced CO2 terrestrial radiation and by recovered ozone solar radiation as

(c) (a) (b)

Fig. 4. Future changes in temperature. Latitude-pressure cross sections of future changes (difference between 2070-2099 and 1970-1999) in annual- and zonal-mean temperature (K) for (a) B21, (b) Sb, and (b) Sc. The contour interval is 1 K for B21 and Sb and 0.5 K for Sc, and

In the no-climate-change simulation Sc, on the other hand, strong warming occurs in the upper stratosphere above about 5 hPa in both hemispheres, except in the southern high latitudes, where prominent warming occurs in the lower stratosphere at 300-30 hPa, centered at 100 hPa, and the cooling above 30 hPa. This temperature change pattern is completely different from the CO2-induced pattern and is due solely to the ODS decrease, or equivalently, the ozone recovery, because the average ODS concentration was higher in the past 30 years (1970-1999) than in the future 30 years (2070-2099) (Fig. 1). In fact, stratospheric

blue colors represent cooling and red colors warming.

**3.2 Future climate change in temperature and zonal wind** 

night jet is weaker.

described below.

ozone abundances were projected to recover to or exceed pre-1980 levels globally in the future, except for the tropical lower stratosphere due to the increased inflow of ozone-poor tropospheric air mass (e.g., SPARC, 210). In other words, the upper stratospheric cooling in the past 1970-1999 was due to an ozone decrease caused by gas-phase reactions with reactive chlorine and bromine atoms photodissociated from ODSs under strong ultraviolet solar radiation, whereas the pattern in the southern high latitudes was associated with severe ozone depletion (i.e., the ozone hole) caused by heterogeneous reactions on PSCs (e.g., SPARC, 2010). This ODS-decrease effect is also included in the B2 simulations; thus, temperature changes similar to those in Sc can be obtained by subtracting the changes in Sb from those in the B2 members (not shown).

Fig. 5. Future changes in zonal wind. Same as Fig.4 except for zonal wind (ms-1). The contour interval is 1 ms-1, and blue colors represent positive (westerly/eastward) acceleration and red colors negative (easterly/westward) acceleration.

Zonal wind changes in the three B2 members and Sb display a very similar pattern in the polar night jet regions in both hemispheres and in the equatorial lower stratosphere. Westerly wind intensification extends from the upper region of the subtropical jet upward along the equatorward flank of the polar night jet axis, reaching a maximum at around 70 hPa and 40° in both hemispheres. In the SH, the westerly wind intensification also extends down to the surface at around 55°S, resulting in poleward spreading of the subtropical jet. In the tropics, westerly wind strengthening (eastward acceleration) occurs above 30 hPa and weakening (westward acceleration) at 50 hPa, below which very weak westerly wind strengthening occurs down to the surface. In the no-climate-change simulation Sc, on the other hand, the zonal wind change pattern is completely different. In the SH, weakening of the polar night jet and both weakening of the poleward and strengthening of the equatorward flank of the subtropical jet down to the surface occur, changes which are almost opposite to those in B2 and Sb in the troposphere above southern high latitudes. This change in the subtropical jet in Sc means that the equatorward shift of the jet is associated with ozone recovery (e.g., Son et al., 2008). In the NH, the poleward flank of the polar night jet is slightly intensified above the middle stratosphere. In the tropics, there is almost no change in the troposphere and stratosphere, consistent with the temperature results.

#### **3.3 Evolution of the Brewer-Dobson circulation and the ozone hole area**

The GHG increase affects the mass exchange between the troposphere and the stratosphere, which can be evaluated as the upward mass flux across the tropopause. Most GCMs and

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

that the effect of the GHG increase on the ozone hole in the SH is limited.

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 365

ODSs will continue to induce an ozone hole over the Antarctic from late winter to spring in the future, as long as their abundances exceed certain values (e.g., SPARC, 2010). Figure 8 exhibits the evolutions of the annual maximum ozone hole area in the ODS time-evolving simulations (B21, B22, B23, and Sc). Different from these simulations, no ozone hole appears in the fixed-ODS simulation Sb, even if the GHGs-induced climate change progresses. Even though the inter-simulation variation is not especially small, the features of all of these simulations are very similar with regard to the long-term evolution of the ozone hole area: (1) initial appearance around 1980, (2) followed immediately by a rapid increase until the later 1990s, (3) then nearly saturated values until about 2010, and (4) followed by a gradual decrease, with disappearance around 2070. Features 1-3 qualitatively correspond to the observed features (e.g., WMO, 2011), aside from the short-term variations. The fact that the evolution of the ozone hole area in Sc is similar to that in the three B2 members indicates

(c) (d)

Fig. 8. Ozone hole area in SH. Evolution of the annual maximum ozone hole area in the SH

Figure 9 shows the observed and simulated zonal-mean zonal wind (averaged from 10°S to 10°N) between 100 and 5 hPa for the 20 years from 1990 to 2009. Monthly averages are subtracted from the values of each month to extract anomalies from the seasonal cycle, so that the semi-annual oscillation (SAO) in the upper stratosphere becomes smaller than or similar to the QBO. It is evident that all of the simulations reproduced the overall features of the observed QBO, such as the stronger westerly shear than easterly shear during the phase transition as well as the stalling of the descent of the easterly winds. Empirical orthogonal function (EOF) analysis revealed that EOF 1 and 2 explain major parts of the variance as in observations (e.g., Wallace et al.,1993) (not shown). The simulated QBOs realistically

in (a) B21, (b) B22, (c) B23, and (d) Sc. Units are 106 km2.

**4.1 Vertical structure and power spectrum of the QBO** 

**4. Characteristics of the QBO** 

(a) (b)

CCMs (e.g., Butchart et al., 2006; Austin & Li, 2006; Garcia & Randel, 2008) show an acceleration of BDC resulting from an increase in GHGs, but balloon measurements do not necessarily support these model results, even given the large uncertainties and sparse observations in both time and space (Engel et al., 2009). Figure 6 depicts the evolution of the global upward mass flux anomaly (49-month running mean) at 100 hPa, calculated from the residual mass stream function. It is evident that the simulations with the GHG increase scenarios (B21, B22, B23, and Sb) result in a nearly linear increase of the upward mass flux across 100 hPa, and the trend slopes are very similar at about 8×109 kgs-1 per century. On the other hand, the simulation results for the no-climate-change scenario Sc exhibit no trend at all. This feature can be also analyzed in the equatorial stratosphere, as displayed in Fig. 7, which depicts the low-frequency (period longer than 10 years) components of the residual vertical velocity \* *w* (in units of 10-4 ms-1, averaged from 10°S to 10°N) from 70 to 1 hPa. Note that some high-frequency components remain, due to leakage of the Lanczos filter. The magnitude of \* *w* is very small in the lower stratosphere, with a minimum at 50 hPa; above 50 hPa it increases with altitude, gradually up to about 5 hPa and steeply above that. This vertical structure does not change in the GHG-increase simulations or in the no-climatechange simulation during the whole integration period, consistent with equilibrium CO2 doubling simulations (e.g., Kawatani, et al., 2011).

Fig. 6. Upward mass flux at 100 hPa. Evolution of the upward mass flux (1010 kgs-1) at 100 hPa for (a) B21, (b) Sb, and (c) Sc. The 49-month running mean is applied to eliminate shortterm variations.

Fig. 7. Low-frequency residual vertical velocity. Time-pressure cross section of the low frequency (period longer than 120 months) component of the residual vertical velocity (averaged between 10S and 10N) from 70 to 1 hPa. Units are 10-4 ms-1.

CCMs (e.g., Butchart et al., 2006; Austin & Li, 2006; Garcia & Randel, 2008) show an acceleration of BDC resulting from an increase in GHGs, but balloon measurements do not necessarily support these model results, even given the large uncertainties and sparse observations in both time and space (Engel et al., 2009). Figure 6 depicts the evolution of the global upward mass flux anomaly (49-month running mean) at 100 hPa, calculated from the residual mass stream function. It is evident that the simulations with the GHG increase scenarios (B21, B22, B23, and Sb) result in a nearly linear increase of the upward mass flux across 100 hPa, and the trend slopes are very similar at about 8×109 kgs-1 per century. On the other hand, the simulation results for the no-climate-change scenario Sc exhibit no trend at all. This feature can be also analyzed in the equatorial stratosphere, as displayed in Fig. 7, which depicts the low-frequency (period longer than 10 years) components of the residual vertical velocity \* *w* (in units of 10-4 ms-1, averaged from 10°S to 10°N) from 70 to 1 hPa. Note that some high-frequency components remain, due to leakage of the Lanczos filter. The magnitude of \* *w* is very small in the lower stratosphere, with a minimum at 50 hPa; above 50 hPa it increases with altitude, gradually up to about 5 hPa and steeply above that. This vertical structure does not change in the GHG-increase simulations or in the no-climatechange simulation during the whole integration period, consistent with equilibrium CO2-

doubling simulations (e.g., Kawatani, et al., 2011).

(c)

term variations.

Fig. 6. Upward mass flux at 100 hPa. Evolution of the upward mass flux (1010 kgs-1) at 100 hPa for (a) B21, (b) Sb, and (c) Sc. The 49-month running mean is applied to eliminate short-

Fig. 7. Low-frequency residual vertical velocity. Time-pressure cross section of the low frequency (period longer than 120 months) component of the residual vertical velocity

(averaged between 10S and 10N) from 70 to 1 hPa. Units are 10-4 ms-1.

(a) (b)

ODSs will continue to induce an ozone hole over the Antarctic from late winter to spring in the future, as long as their abundances exceed certain values (e.g., SPARC, 2010). Figure 8 exhibits the evolutions of the annual maximum ozone hole area in the ODS time-evolving simulations (B21, B22, B23, and Sc). Different from these simulations, no ozone hole appears in the fixed-ODS simulation Sb, even if the GHGs-induced climate change progresses. Even though the inter-simulation variation is not especially small, the features of all of these simulations are very similar with regard to the long-term evolution of the ozone hole area: (1) initial appearance around 1980, (2) followed immediately by a rapid increase until the later 1990s, (3) then nearly saturated values until about 2010, and (4) followed by a gradual decrease, with disappearance around 2070. Features 1-3 qualitatively correspond to the observed features (e.g., WMO, 2011), aside from the short-term variations. The fact that the evolution of the ozone hole area in Sc is similar to that in the three B2 members indicates that the effect of the GHG increase on the ozone hole in the SH is limited.

Fig. 8. Ozone hole area in SH. Evolution of the annual maximum ozone hole area in the SH in (a) B21, (b) B22, (c) B23, and (d) Sc. Units are 106 km2.

## **4. Characteristics of the QBO**

## **4.1 Vertical structure and power spectrum of the QBO**

Figure 9 shows the observed and simulated zonal-mean zonal wind (averaged from 10°S to 10°N) between 100 and 5 hPa for the 20 years from 1990 to 2009. Monthly averages are subtracted from the values of each month to extract anomalies from the seasonal cycle, so that the semi-annual oscillation (SAO) in the upper stratosphere becomes smaller than or similar to the QBO. It is evident that all of the simulations reproduced the overall features of the observed QBO, such as the stronger westerly shear than easterly shear during the phase transition as well as the stalling of the descent of the easterly winds. Empirical orthogonal function (EOF) analysis revealed that EOF 1 and 2 explain major parts of the variance as in observations (e.g., Wallace et al.,1993) (not shown). The simulated QBOs realistically

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

(c) (d)

the simulations.

**4.2 Forcings of the QBO** 

Fig. 10. Power spectra of zonal wind. Power spectra (ms-1) of zonal-mean zonal wind averaged from 10°S to 10°N for periods > 16 months: (a) ERA-Interim, (b) B21, (c) Sb, and (d) Sc. The analysis period was from 1958 to 2009 for ERA-Interim, and from 1960 to 2009 for

configuration among the QBO and its forcings are represented in Fig. 11.

Figure 11 displays the composite relations within the QBO cycle at 30 and 70 hPa for B21 between the QBO zonal wind and the three major forcings, parameterized gravity-wave forcing (GWF), resolved-wave forcing (Eliassen-Palm flux divergence, EPD), and vertical advection of zonal momentum ( \* *w* (/) *u z* , WUZ), calculated for the entire period of 1960- 2099. Note that all of the forcing terms and the zonal wind are band-passed (20-40 months) values with the QBO components extracted by a Lanczos filter. In making the composite figures, we assumed that each cycle expands in 2 phase space with a peak power period of 28 months, even though each cycle had a different period. Although the fine structures such as steep gradient during the phase transitions disappear due to the band-pass filtering, basic

AT 30 hPa, GWF is retarded by about one-seventh cycle relative to zonal wind, so that positive (eastward acceleration) GWF occurs approximately from after the initial stage of the westerly shear descent (equivalently, u/t > 0) to the initial stage of the easterly shear descent (u/t < 0); conversely, negative (westward acceleration) GWF continues from after

(a) (b)

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 367

QBO has less power than the simulated QBO in the upper stratosphere. On the basis of the power spectrum distribution, we defined the QBO as the components with a period between 20 and 40 months. In the higher frequency domain (periods shorter than 16 months), the power spectrum is a narrow monomodal distribution concentrated at an annual frequency below 10 hPa, and a bimodal distribution sharply concentrated at annual and semiannual frequencies above 10 hPa, in both the observation and the simulations (not shown). The QBO power is the largest of these two or three peak powers below the middle stratosphere.

descend to as low as 100 hPa, whereas their amplitudes are smaller than the observed amplitude, particularly below about 50 hPa in the lower stratosphere, a common bias among three-dimensional models (Takahashi, 1999; Hamilton et al., 1999; Scaife et al., 2000; Giorgetta et al., 2002; Butchart et al., 2003; Shibata & Deushi, 2005a; Watanabe et al., 2008; Kawatani et al., 2011). The differences among the simulations are not necessarily discernible in Fig. 9.

Fig. 9. Zonal wind evolution in the tropics. Evolution of zonal-mean zonal wind (averaged between 10S and 10N) from 100 to 5 hPa for the 20 years from 1990 to 2009: (a) ERA-Interim, (b) B21, (c) Sb, and (d) Sc.

Figure 10 depicts the power spectrum in the low-frequency domain (periods longer than 16 months) for the observed and simulated wind averaged 10°S to 10°N from 100 to 1 hPa. The observed power, extending from about 18 to 240 months, is evidently much broader than the simulated power. This is because the simulations include only the anthropogenic forcing of GHGs and ODSs without the natural forcings of the solar 11-year cycle and volcanic aerosols. The broad spectral peaks at around 28 months, which represent the QBO, are distributed similarly between the observation and the simulations, though the observed QBO has less power than the simulated QBO in the upper stratosphere. On the basis of the power spectrum distribution, we defined the QBO as the components with a period between 20 and 40 months. In the higher frequency domain (periods shorter than 16 months), the power spectrum is a narrow monomodal distribution concentrated at an annual frequency below 10 hPa, and a bimodal distribution sharply concentrated at annual and semiannual frequencies above 10 hPa, in both the observation and the simulations (not shown). The QBO power is the largest of these two or three peak powers below the middle stratosphere.

Fig. 10. Power spectra of zonal wind. Power spectra (ms-1) of zonal-mean zonal wind averaged from 10°S to 10°N for periods > 16 months: (a) ERA-Interim, (b) B21, (c) Sb, and (d) Sc. The analysis period was from 1958 to 2009 for ERA-Interim, and from 1960 to 2009 for the simulations.

## **4.2 Forcings of the QBO**

366 Greenhouse Gases – Emission, Measurement and Management

descend to as low as 100 hPa, whereas their amplitudes are smaller than the observed amplitude, particularly below about 50 hPa in the lower stratosphere, a common bias among three-dimensional models (Takahashi, 1999; Hamilton et al., 1999; Scaife et al., 2000; Giorgetta et al., 2002; Butchart et al., 2003; Shibata & Deushi, 2005a; Watanabe et al., 2008; Kawatani et al., 2011). The differences among the simulations are not necessarily discernible

Fig. 9. Zonal wind evolution in the tropics. Evolution of zonal-mean zonal wind (averaged between 10S and 10N) from 100 to 5 hPa for the 20 years from 1990 to 2009: (a) ERA-Interim,

Figure 10 depicts the power spectrum in the low-frequency domain (periods longer than 16 months) for the observed and simulated wind averaged 10°S to 10°N from 100 to 1 hPa. The observed power, extending from about 18 to 240 months, is evidently much broader than the simulated power. This is because the simulations include only the anthropogenic forcing of GHGs and ODSs without the natural forcings of the solar 11-year cycle and volcanic aerosols. The broad spectral peaks at around 28 months, which represent the QBO, are distributed similarly between the observation and the simulations, though the observed

in Fig. 9.

(b) B21, (c) Sb, and (d) Sc.

Figure 11 displays the composite relations within the QBO cycle at 30 and 70 hPa for B21 between the QBO zonal wind and the three major forcings, parameterized gravity-wave forcing (GWF), resolved-wave forcing (Eliassen-Palm flux divergence, EPD), and vertical advection of zonal momentum ( \* *w* (/) *u z* , WUZ), calculated for the entire period of 1960- 2099. Note that all of the forcing terms and the zonal wind are band-passed (20-40 months) values with the QBO components extracted by a Lanczos filter. In making the composite figures, we assumed that each cycle expands in 2 phase space with a peak power period of 28 months, even though each cycle had a different period. Although the fine structures such as steep gradient during the phase transitions disappear due to the band-pass filtering, basic configuration among the QBO and its forcings are represented in Fig. 11.

AT 30 hPa, GWF is retarded by about one-seventh cycle relative to zonal wind, so that positive (eastward acceleration) GWF occurs approximately from after the initial stage of the westerly shear descent (equivalently, u/t > 0) to the initial stage of the easterly shear descent (u/t < 0); conversely, negative (westward acceleration) GWF continues from after

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

\* \*

*u u*

*z z*

*L Q*

*L Q*

**5.1 Future change of the zonal wind forcing in the tropical stratosphere** 

Changes in the zonal wind forcings affect the background zonal wind, which in turn modifies the QBO through the interactions between the background zonal wind and the

zonal wind due to WUZ is expressed as

demonstrated later

*WUZ*

where |(X)Q| represents the QBO amplitude of X.

climate change caused by the GHG increase.

**5. Evolution of the QBO** 

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 369

similarly to zonal wind as stated before. Thus, since cross terms with annual or semiannual component cannot create the QBO term of broad-period range, the QBO acceleration of

> *u u uu w ww t z zz*

where the superscripts Q and L represent the QBO and the low-frequency components, respectively. The first term on the right-hand side of the approximation sign represents the contribution of the lower-frequency (background) vertical wind, corresponding to the interaction with BDC, and the second term is the interaction between the background zonal wind shear and the QBO secondary circulation. Henceforth the first term is referred to as BDT and the second term as SCT. Just below the maximum power altitude at 30 hPa, the magnitude of each term on the right hand side of Eq. (1) is, on average, as will be

~ . ( ), ~ . ( )

1 5 10 3 5 10

0 35 10 0 1 10

It is evident that the residual vertical velocity is one order smaller than the vertical shear of zonal wind for both the QBO and the low-frequency components. Because the QBO vertical shear of zonal wind is almost of opposite phase to the QBO vertical wind (e.g., Plumb & Bell, 1982; Andrews et al., 1987), and the low-frequency term of vertical shear of zonal wind and that of vertical wind are negative- and positive-definite, respectively, at 30 hPa throughout the whole period from 1960 to 2100, BDT and SCT exert acceleration in approximately the same direction with a ratio of about 8 to 1, indicating that BDT is the major acceleration term of the vertical momentum advection. The magnitude of BDT is also much larger than that of SCT at other altitudes, in particular in the lower stratosphere, because the vertical shear is larger in the QBO than in the low-frequency and the residual vertical wind is larger in the low-frequency than in the QBO. However, other features such as acceleration direction and trend differ depending on altitude and latitude. In the lower stratosphere at 50 hPa and below, for example, BDT and SCT do not always exert acceleration in the same direction because the low-frequency vertical shear did not hold but, in the middle of the integration period (1960-2099), changes its sign from negative to positive at 50 hPa and from positive to negative at 70 hPa over the equator as a result of the

*w ms w ms*

3 1 3 1

*s s*

3 1 3 1

~ . ( ), ~ . ( )

 \* \*\* ~ *<sup>Q</sup> <sup>Q</sup> Q L L Q*

(1)

Fig. 11. Composite relations among the QBO and its forcings. Composite relations among zonal wind (U, red solid curve with closed circles), parameterized gravity wave forcing (GWF, black dashed curve), resolved wave forcing (EPD, green solid curve), and vertical advection (WUZ, cyan solid curve) at (upper) 30 hPa and (lower) 70 hPa over two cycles for an average QBO cycle (28 months). All quantities are averaged from 10°S to 10°N and bandpass filtered (20-40 months). Units are ms-1 for U, 10-2 ms-1 day-1 for forcing/advection.

the initial stage of the easterly shear descent to the initial stage of the westerly shear descent. The amplitude of WUZ is three-quarters that of GWF amplitude and is almost of opposite phase, thus partially cancelling GWF. Therefore, GWF promotes the descent of the vertical shear, and WUZ hinders it. EPD lags GWF by about one-seventh cycle, and thus lags the zonal wind by approximately one-quarter cycle. Of these three forcings, GWF and WUZ are the largest and second largest terms, but as a result of the partial cancellation due to their being of almost opposite phase, the sum of GWF and WUZ has a similar amplitude to that of EPD. At other altitudes, a similar relation quantitatively holds true, particularly at 20 hPa and above, although the deviation from the relation at 30 hPa increases in the lower stratosphere (50-70 hPa). AT 50 hPa, GWF is nearly in-phase with zonal wind, whereas WUZ and EPD lag the zonal wind by about four- and three-sevenths cycle, respectively, and are thus out of phase with zonal wind. In addition, the amplitudes of GWF and WUZ decrease from around 0.13 and 0.10 ms-1 day-1, respectively, at 30 hPa to 0.04 and 0.03 ms-1 day-1, respectively, at 70 hPa, resulting in similar amplitudes among GWF, WUZ, and EPD at 70 hPa. The Differences in the phase relation among all of the simulations were very small.

The vertical advection WUZ consists of two terms. One is associated with low-frequency (<< the QBO frequency) vertical wind, and the other with the vertical wind from the secondary circulation of the QBO (e.g., Plumb & Bell, 1982). This is because WUZ is the product of the residual vertical wind and the vertical shear of zonal wind, both of which have two major peaks in power spectrum at the QBO and annual frequencies below the middle stratosphere, similarly to zonal wind as stated before. Thus, since cross terms with annual or semiannual component cannot create the QBO term of broad-period range, the QBO acceleration of zonal wind due to WUZ is expressed as

$$\left(\frac{\partial u}{\partial t}\right)\_{\text{MLZ}}^{Q} = -\left(\text{w}^\*\left(\frac{\partial u}{\partial z}\right)\right)^Q \sim -\left(\text{w}^\*\right)^L \left(\frac{\partial u}{\partial z}\right)^Q - \left(\text{w}^\*\right)^Q \left(\frac{\partial u}{\partial z}\right)^L \tag{1}$$

where the superscripts Q and L represent the QBO and the low-frequency components, respectively. The first term on the right-hand side of the approximation sign represents the contribution of the lower-frequency (background) vertical wind, corresponding to the interaction with BDC, and the second term is the interaction between the background zonal wind shear and the QBO secondary circulation. Henceforth the first term is referred to as BDT and the second term as SCT. Just below the maximum power altitude at 30 hPa, the magnitude of each term on the right hand side of Eq. (1) is, on average, as will be demonstrated later

$$
\left(\frac{\partial \boldsymbol{u}}{\partial \boldsymbol{z}}\right)^{\boldsymbol{L}} \sim -1.5 \times 10^{-3} (\text{s}^{-1}), \left| \left(\frac{\partial \boldsymbol{u}}{\partial \boldsymbol{z}}\right)^{\boldsymbol{Q}} \right| \sim +3.5 \times 10^{-3} (\text{s}^{-1})
$$

$$
\left(\boldsymbol{w}^{\star}\right)^{\boldsymbol{L}} \sim +0.35 \times 10^{-3} (\boldsymbol{m} \text{s}^{-1}), \left| \left(\boldsymbol{w}^{\star}\right)^{\boldsymbol{Q}} \right| \sim +0.1 \times 10^{-3} (\text{m} \text{s}^{-1})
$$

where |(X)Q| represents the QBO amplitude of X.

368 Greenhouse Gases – Emission, Measurement and Management

Fig. 11. Composite relations among the QBO and its forcings. Composite relations among zonal wind (U, red solid curve with closed circles), parameterized gravity wave forcing (GWF, black dashed curve), resolved wave forcing (EPD, green solid curve), and vertical advection (WUZ, cyan solid curve) at (upper) 30 hPa and (lower) 70 hPa over two cycles for an average QBO cycle (28 months). All quantities are averaged from 10°S to 10°N and bandpass filtered (20-40 months). Units are ms-1 for U, 10-2 ms-1 day-1 for forcing/advection.

the initial stage of the easterly shear descent to the initial stage of the westerly shear descent. The amplitude of WUZ is three-quarters that of GWF amplitude and is almost of opposite phase, thus partially cancelling GWF. Therefore, GWF promotes the descent of the vertical shear, and WUZ hinders it. EPD lags GWF by about one-seventh cycle, and thus lags the zonal wind by approximately one-quarter cycle. Of these three forcings, GWF and WUZ are the largest and second largest terms, but as a result of the partial cancellation due to their being of almost opposite phase, the sum of GWF and WUZ has a similar amplitude to that of EPD. At other altitudes, a similar relation quantitatively holds true, particularly at 20 hPa and above, although the deviation from the relation at 30 hPa increases in the lower stratosphere (50-70 hPa). AT 50 hPa, GWF is nearly in-phase with zonal wind, whereas WUZ and EPD lag the zonal wind by about four- and three-sevenths cycle, respectively, and are thus out of phase with zonal wind. In addition, the amplitudes of GWF and WUZ decrease from around 0.13 and 0.10 ms-1 day-1, respectively, at 30 hPa to 0.04 and 0.03 ms-1 day-1, respectively, at 70 hPa, resulting in similar amplitudes among GWF, WUZ, and EPD at 70 hPa. The Differences in the phase relation among all of the simulations were very

The vertical advection WUZ consists of two terms. One is associated with low-frequency (<< the QBO frequency) vertical wind, and the other with the vertical wind from the secondary circulation of the QBO (e.g., Plumb & Bell, 1982). This is because WUZ is the product of the residual vertical wind and the vertical shear of zonal wind, both of which have two major peaks in power spectrum at the QBO and annual frequencies below the middle stratosphere,

small.

It is evident that the residual vertical velocity is one order smaller than the vertical shear of zonal wind for both the QBO and the low-frequency components. Because the QBO vertical shear of zonal wind is almost of opposite phase to the QBO vertical wind (e.g., Plumb & Bell, 1982; Andrews et al., 1987), and the low-frequency term of vertical shear of zonal wind and that of vertical wind are negative- and positive-definite, respectively, at 30 hPa throughout the whole period from 1960 to 2100, BDT and SCT exert acceleration in approximately the same direction with a ratio of about 8 to 1, indicating that BDT is the major acceleration term of the vertical momentum advection. The magnitude of BDT is also much larger than that of SCT at other altitudes, in particular in the lower stratosphere, because the vertical shear is larger in the QBO than in the low-frequency and the residual vertical wind is larger in the low-frequency than in the QBO. However, other features such as acceleration direction and trend differ depending on altitude and latitude. In the lower stratosphere at 50 hPa and below, for example, BDT and SCT do not always exert acceleration in the same direction because the low-frequency vertical shear did not hold but, in the middle of the integration period (1960-2099), changes its sign from negative to positive at 50 hPa and from positive to negative at 70 hPa over the equator as a result of the climate change caused by the GHG increase.

#### **5. Evolution of the QBO**

#### **5.1 Future change of the zonal wind forcing in the tropical stratosphere**

Changes in the zonal wind forcings affect the background zonal wind, which in turn modifies the QBO through the interactions between the background zonal wind and the

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

Fig. 13. WUZ in the past and future. Same as Fig.12 except for WUZ.

Fig. 14. EPD in the past and future. Same as Fig.12 except for EPD. Units are 10-2 ms-1 per day. The contour interval is 2 for the values greater than -12, except that the outer contours in the left panels have the values -16, -32, and -64. The contour interval is 1 for values greater

Future changes in GWF and EPD are larger outside of the mid-tropics below the middle stratosphere in GHG-increase simulations, owing to the shift of the zero line of zonal wind (e.g., Deushi & Shibata, 2011). In contrast, future GWF and EPD changes are small in the


Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 371

QBO; thus, such changes are one of the crucial factors affecting the future QBO. Figures 12- 14 depict annual mean GWF, WUZ, and EPD in the past (30 years, 1970-1999) and their changes after 100 years (30 years, 2070-2099) in the low-latitude (30°S-30°N) stratosphere from 70 to 5 hPa. GWF and EPD generally exert negative (westward) acceleration in this domain, except in some regions: namely, in the upper stratosphere around 15° in both hemispheres, and at 50 hPa in the deep-tropics (10°S-10°N) for GWF, and at 30 hPa south of the equator for EPD. On the other hand, WUZ exerts positive (eastward) acceleration almost everywhere in this domain.

It is evident that the distributions of the forcing magnitude are commonly very different between the mid-tropics (15°S-15°N) and higher latitudes. In the mid-tropics, the magnitude of GWF is minimal below 50 hPa, and above 50 hPa it is maximal and more concentrated south of the equator. The magnitude of WUZ shows a similar pattern with a broader maximum above 50 hPa. The magnitude of EPD is also minimal in the mid-tropics, and the width of the minimum area becomes narrower with altitude; the magnitude is largest at 50 hPa, and then the minimum area becomes more concentrated downward toward the equator. The EPD minimum over the tropics reflects the suppression of the propagation of Rossby waves from the mid-latitudes and subtropics against the easterly background zonal wind (Fig. 3).

Fig. 12. GWF in the past and future. Annual and zonal mean GWF (left) in the past 30-year period (1970-1999) and (right) the changes in the future 30-year period (2070-2099) for (upper) B21, (middle) Sb, and (lower) Sc. The contour interval is 2x10-2 ms-1 per day for the left panels and 1x10-2 ms-1 per day for the right panels.

QBO; thus, such changes are one of the crucial factors affecting the future QBO. Figures 12- 14 depict annual mean GWF, WUZ, and EPD in the past (30 years, 1970-1999) and their changes after 100 years (30 years, 2070-2099) in the low-latitude (30°S-30°N) stratosphere from 70 to 5 hPa. GWF and EPD generally exert negative (westward) acceleration in this domain, except in some regions: namely, in the upper stratosphere around 15° in both hemispheres, and at 50 hPa in the deep-tropics (10°S-10°N) for GWF, and at 30 hPa south of the equator for EPD. On the other hand, WUZ exerts positive (eastward) acceleration almost

It is evident that the distributions of the forcing magnitude are commonly very different between the mid-tropics (15°S-15°N) and higher latitudes. In the mid-tropics, the magnitude of GWF is minimal below 50 hPa, and above 50 hPa it is maximal and more concentrated south of the equator. The magnitude of WUZ shows a similar pattern with a broader maximum above 50 hPa. The magnitude of EPD is also minimal in the mid-tropics, and the width of the minimum area becomes narrower with altitude; the magnitude is largest at 50 hPa, and then the minimum area becomes more concentrated downward toward the equator. The EPD minimum over the tropics reflects the suppression of the propagation of Rossby waves from the mid-latitudes and subtropics against the easterly background zonal

Fig. 12. GWF in the past and future. Annual and zonal mean GWF (left) in the past 30-year period (1970-1999) and (right) the changes in the future 30-year period (2070-2099) for (upper) B21, (middle) Sb, and (lower) Sc. The contour interval is 2x10-2 ms-1 per day for the

left panels and 1x10-2 ms-1 per day for the right panels.

everywhere in this domain.

wind (Fig. 3).

Fig. 13. WUZ in the past and future. Same as Fig.12 except for WUZ.

Fig. 14. EPD in the past and future. Same as Fig.12 except for EPD. Units are 10-2 ms-1 per day. The contour interval is 2 for the values greater than -12, except that the outer contours in the left panels have the values -16, -32, and -64. The contour interval is 1 for values greater -2, but for values less than -2, the contour interval is 2 in the right panels.

Future changes in GWF and EPD are larger outside of the mid-tropics below the middle stratosphere in GHG-increase simulations, owing to the shift of the zero line of zonal wind (e.g., Deushi & Shibata, 2011). In contrast, future GWF and EPD changes are small in the

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

Sc (cyan dashed).

three members.

magnitude.

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 373

Fig. 16. QBO amplitude evolution. Evolution of the QBO amplitude of zonal-mean zonal wind at 30 hPa in B21 (black solid), B22 (green solid), B23 (red solid), Sb (blue dashed), and

Figure 16 depicts the QBO amplitude of the zonal-mean zonal wind (averaged from 10°S to 10°N) at 30 hPa in all simulations. The QBO amplitudes of the GHG-increase simulations steadily decreases from about 16 ms-1 in the 1960s and 70s to about 12 ms-1 in the 2080s and 90s, resulting in, on average, a linear trend of about 0.3 ms-1 per decade. On the other hand, in the no-climate-change simulation Sc, the QBO amplitude does not seem to show a distinct trend. Nevertheless, close inspection reveals that the amplitude, like the GHG-increase simulations, decreases until about year 2000, and thereafter gradually increases. As a result, the overall trend is a small positive slope. Among all simulations, other features such as decadal variations are very different from one simulation to another, even within the B2

Different from the distinct decreasing trend of the QBO amplitude, the QBO period does not exhibit any significant trend as depicted by the local wavelet power spectrum of the QBO zonal wind at 20 hPa, where the power maximizes (Fig. 17). The QBO powers are mainly confined to between 22 and 32 months, and the peak power frequencies (periods) stabilize at 28 months, in spite of some excursions toward 20-24 months in the GHG-increase simulations. Similarly to the direct evaluation, the QBO powers show a common decreasing trend in the GHG-increase simulations. The no-climate-change simulation Sc, on the other hand, shows a very stable QBO without a substantial trend in either frequency or power

Fig. 17. QBO local wavelet power spectrum at 20 hPa. Evolution of the local wavelet power spectrum of the zonal-mean zonal wind averaged from 10°S to 10°N at 20 hPa in the QBO

period for (a) B21, (b) Sb, and (c) Sc. Units are m2s-2 and the contour interval is 20.

mid-tropical middle stratosphere, in particular at around 30 hPa, in GHG-increase simulations. On the other hand, in the mid-tropical middle stratosphere, the future WUZ changes are larger negative values with a large cell structure. In the no-climate-change simulation, the future changes in the QBO forcings are very small everywhere.

#### **5.2 Trend of the QBO zonal wind**

Figure 15 exhibits the QBO zonal wind (averaged from 10°S to 10°N) at 30 hPa for the whole period from 1960 to 2099 in simulations for B21, Sb, and Sc. It is evident that the QBO zonal winds show long-term decreasing trends in all of the GHG-increase simulations, whereas in the no-climate-change simulation Sc, the QBO zonal wind shows no long-term trend at all. To quantify the trends, the QBO amplitude was calculated by two independent methods, that is, direct and wavelet methods. The direct method assumes that the QBO amplitude is √2σ, analogous to a monochromatic wave, where σ is the root mean square of the QBO time series over three cycles, even though the QBO time series is not a single sinusoidal wave (Pascoe et al., 2005; Baldwin & Gray, 2005). The cycles are defined as periods from one zero point to the third consecutive zero point, and the QBO amplitude is assigned to the center time of the three cycles. In this way, the QBO amplitudes are first calculated discretely in time, about a half (14 months) cycle apart, and then monthly amplitudes are obtained through cubic interpolation with Lagrange polynomials. This procedure results in a QBO amplitude that is a low-passed (> 3 cycles~7 years) data with a temporal resolution of about 14 months. The wavelet method used a Morlet mother wavelet (plane wave modified by a Gaussian envelope) with non-dimensional frequency ω0=6 (e.g., Torrence & Compo, 1998), so the result of the wavelet analysis incorporates a mean information within approximately three cycles centered at time and frequency concerned.

Fig. 15. QBO wind evolution. Zonal-mean zonal wind averaged from 10°S to 10°N at 30 hPa for the whole period 1960-2099 for (a) B21, (b) Sb, and (c) Sc. A band-pass filter (20-40 months) was applied to obtain the QBO component.

mid-tropical middle stratosphere, in particular at around 30 hPa, in GHG-increase simulations. On the other hand, in the mid-tropical middle stratosphere, the future WUZ changes are larger negative values with a large cell structure. In the no-climate-change

Figure 15 exhibits the QBO zonal wind (averaged from 10°S to 10°N) at 30 hPa for the whole period from 1960 to 2099 in simulations for B21, Sb, and Sc. It is evident that the QBO zonal winds show long-term decreasing trends in all of the GHG-increase simulations, whereas in the no-climate-change simulation Sc, the QBO zonal wind shows no long-term trend at all. To quantify the trends, the QBO amplitude was calculated by two independent methods, that is, direct and wavelet methods. The direct method assumes that the QBO amplitude is √2σ, analogous to a monochromatic wave, where σ is the root mean square of the QBO time series over three cycles, even though the QBO time series is not a single sinusoidal wave (Pascoe et al., 2005; Baldwin & Gray, 2005). The cycles are defined as periods from one zero point to the third consecutive zero point, and the QBO amplitude is assigned to the center time of the three cycles. In this way, the QBO amplitudes are first calculated discretely in time, about a half (14 months) cycle apart, and then monthly amplitudes are obtained through cubic interpolation with Lagrange polynomials. This procedure results in a QBO amplitude that is a low-passed (> 3 cycles~7 years) data with a temporal resolution of about 14 months. The wavelet method used a Morlet mother wavelet (plane wave modified by a Gaussian envelope) with non-dimensional frequency ω0=6 (e.g., Torrence & Compo, 1998), so the result of the wavelet analysis incorporates a mean information within approximately

Fig. 15. QBO wind evolution. Zonal-mean zonal wind averaged from 10°S to 10°N at 30 hPa for the whole period 1960-2099 for (a) B21, (b) Sb, and (c) Sc. A band-pass filter (20-40

simulation, the future changes in the QBO forcings are very small everywhere.

**5.2 Trend of the QBO zonal wind** 

three cycles centered at time and frequency concerned.

months) was applied to obtain the QBO component.

Fig. 16. QBO amplitude evolution. Evolution of the QBO amplitude of zonal-mean zonal wind at 30 hPa in B21 (black solid), B22 (green solid), B23 (red solid), Sb (blue dashed), and Sc (cyan dashed).

Figure 16 depicts the QBO amplitude of the zonal-mean zonal wind (averaged from 10°S to 10°N) at 30 hPa in all simulations. The QBO amplitudes of the GHG-increase simulations steadily decreases from about 16 ms-1 in the 1960s and 70s to about 12 ms-1 in the 2080s and 90s, resulting in, on average, a linear trend of about 0.3 ms-1 per decade. On the other hand, in the no-climate-change simulation Sc, the QBO amplitude does not seem to show a distinct trend. Nevertheless, close inspection reveals that the amplitude, like the GHG-increase simulations, decreases until about year 2000, and thereafter gradually increases. As a result, the overall trend is a small positive slope. Among all simulations, other features such as decadal variations are very different from one simulation to another, even within the B2 three members.

Different from the distinct decreasing trend of the QBO amplitude, the QBO period does not exhibit any significant trend as depicted by the local wavelet power spectrum of the QBO zonal wind at 20 hPa, where the power maximizes (Fig. 17). The QBO powers are mainly confined to between 22 and 32 months, and the peak power frequencies (periods) stabilize at 28 months, in spite of some excursions toward 20-24 months in the GHG-increase simulations. Similarly to the direct evaluation, the QBO powers show a common decreasing trend in the GHG-increase simulations. The no-climate-change simulation Sc, on the other hand, shows a very stable QBO without a substantial trend in either frequency or power magnitude.

Fig. 17. QBO local wavelet power spectrum at 20 hPa. Evolution of the local wavelet power spectrum of the zonal-mean zonal wind averaged from 10°S to 10°N at 20 hPa in the QBO period for (a) B21, (b) Sb, and (c) Sc. Units are m2s-2 and the contour interval is 20.

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

**5.3 Trends in the forcings of the QBO zonal wind** 

space, owing to their different phases (Fig. 11).

(m/s/day)

(a)

(b)

(c)

forcing (EPD).

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 375

The QBO trends of the zonal wind amplitude are statistically significant below 20 hPa in the GHG-increase simulations and above 30 hPa in the no-climate-change simulation; therefore, the forcings of the QBO zonal wind themselves show similar trends. Figure 20 exhibits the evolutions of the QBO forcing terms, GWF, WUZ, and EPD, at 30 hPa in all simulations. As shown by the composite relations among the QBO and its forcings (Fig. 11), GWF has the largest amplitude and EPD the smallest amplitude during the whole period (1960-2099). Although the short-term variations are considerable, it is evident that the forcings show decreasing trends in the GHG-increase simulations and that they do not in the no-climatechange simulation. Similarly to the quantification of the trend of the QBO zonal wind amplitude, linear regression was used to calculate the trends of the forcings. In addition, to include different responses at different frequencies, that is, a larger response at a smaller frequency, even in the QBO narrow-frequency domain (period of 20-40 months) and to get more direct contribution to the trend of the QBO zonal wind, each raw forcing was integrated over time and band-passed to yield the corresponding zonal wind response (see Eq. (1)). For example, uGWF is the integrated QBO zonal wind (response wind) due to GWF. Following this procedure, the linear regressions were calculated. Note that the trends in the forcing amplitudes are not additive, even though forcings are additive in the temporal

Fig. 20. QBO forcing evolution. Same as Fig. 16 except for (a) the parameterized gravity forcing (GWF), (b) the momentum vertical advection (WUZ), and (c) the resolved wave

Figure 21 displays the vertical profiles of GWF and uGWF from 70 to 5 hPa for all simulations. The GHG-increase simulations exhibit the same statistically significant negative trend at 30 ant 70 hPa and, in most cases, at 50 hPa as well, with a maximum of about 0.2x10-2 ms-1 day-1

GWF\_amp

WUZ\_amp

EPD\_amp

Fig. 18. Vertical profile of the QBO local wavelet power spectrum. Evolution of the wavelet power of the QBO zonal-mean zonal wind (averaged from 10°S to 10°N) from 70 to 10 hPa for (a) B21, (b) Sb, and (c) Sc. Units are m2s-2 and the contour interval is 20.

Figure 18 exhibits the evolution of the QBO local wavelet power from 70 to 10 hPa for B21, Sb, and Sc simulations. Below 30 hPa, the QBO powers show an apparent decreasing trend in the GHG-increase simulations, whereas they do not at 10 hPa. The no-climate-change simulation Sc, on the other hand, does not seem to show a discernible trend for the QBO zonal wind at any altitude. To quantify linear trends in the QBO amplitude, the linear regression is calculated not for the monthly interpolated amplitude but for the original about half-cycle (~14 months) interval amplitude using the least squares method at each altitude.

Figure 19 depicts the vertical profile of the linear trend in the QBO zonal wind amplitude from 70 to 10 hPa for all simulations. The GHG-increase simulations all exhibit a statistically significant (at the 99% confidence level) negative trend below 20 hPa, except for that of B23 at 20 hPa. The maximum negative trends at 30 hPa are about 0.3 ms-1 per decade, and negative trend below and above 30 hPa are 0.2-0.1 ms-1 per decade. In the upper stratosphere, the trends are near zero or small positive, but the differences among the simulations are not small. The no-climate-change simulation exhibits a near-zero trend in the lower stratosphere below 50 hPa, and a small but statistically significant positive trend above 30 hPa. These results demonstrate that the GHG increase is responsible for the decrease in the QBO amplitude below 20 hPa and that ODS changes alone cause the increase above 30 hPa.

Fig. 19. Trends of the QBO amplitude. Vertical profiles of the linear trends (ms-1 per decade) in the QBO amplitude for the zonal-mean zonal wind (averaged from 10°S to 10°N) from 70 to 5 hPa in all simulations. Open circles and diamonds represent the trend being different from zero at the 99% and 95% confidence levels (*t*-test), respectively. The QBO amplitude was calculated by the direct method.

#### **5.3 Trends in the forcings of the QBO zonal wind**

374 Greenhouse Gases – Emission, Measurement and Management

Fig. 18. Vertical profile of the QBO local wavelet power spectrum. Evolution of the wavelet power of the QBO zonal-mean zonal wind (averaged from 10°S to 10°N) from 70 to 10 hPa

Figure 18 exhibits the evolution of the QBO local wavelet power from 70 to 10 hPa for B21, Sb, and Sc simulations. Below 30 hPa, the QBO powers show an apparent decreasing trend in the GHG-increase simulations, whereas they do not at 10 hPa. The no-climate-change simulation Sc, on the other hand, does not seem to show a discernible trend for the QBO zonal wind at any altitude. To quantify linear trends in the QBO amplitude, the linear regression is calculated not for the monthly interpolated amplitude but for the original about half-cycle

Figure 19 depicts the vertical profile of the linear trend in the QBO zonal wind amplitude from 70 to 10 hPa for all simulations. The GHG-increase simulations all exhibit a statistically significant (at the 99% confidence level) negative trend below 20 hPa, except for that of B23 at 20 hPa. The maximum negative trends at 30 hPa are about 0.3 ms-1 per decade, and negative trend below and above 30 hPa are 0.2-0.1 ms-1 per decade. In the upper stratosphere, the trends are near zero or small positive, but the differences among the simulations are not small. The no-climate-change simulation exhibits a near-zero trend in the lower stratosphere below 50 hPa, and a small but statistically significant positive trend above 30 hPa. These results demonstrate that the GHG increase is responsible for the decrease in the QBO amplitude

Fig. 19. Trends of the QBO amplitude. Vertical profiles of the linear trends (ms-1 per decade) in the QBO amplitude for the zonal-mean zonal wind (averaged from 10°S to 10°N) from 70 to 5 hPa in all simulations. Open circles and diamonds represent the trend being different from zero at the 99% and 95% confidence levels (*t*-test), respectively. The QBO amplitude

for (a) B21, (b) Sb, and (c) Sc. Units are m2s-2 and the contour interval is 20.

(~14 months) interval amplitude using the least squares method at each altitude.

below 20 hPa and that ODS changes alone cause the increase above 30 hPa.

was calculated by the direct method.

The QBO trends of the zonal wind amplitude are statistically significant below 20 hPa in the GHG-increase simulations and above 30 hPa in the no-climate-change simulation; therefore, the forcings of the QBO zonal wind themselves show similar trends. Figure 20 exhibits the evolutions of the QBO forcing terms, GWF, WUZ, and EPD, at 30 hPa in all simulations. As shown by the composite relations among the QBO and its forcings (Fig. 11), GWF has the largest amplitude and EPD the smallest amplitude during the whole period (1960-2099). Although the short-term variations are considerable, it is evident that the forcings show decreasing trends in the GHG-increase simulations and that they do not in the no-climatechange simulation. Similarly to the quantification of the trend of the QBO zonal wind amplitude, linear regression was used to calculate the trends of the forcings. In addition, to include different responses at different frequencies, that is, a larger response at a smaller frequency, even in the QBO narrow-frequency domain (period of 20-40 months) and to get more direct contribution to the trend of the QBO zonal wind, each raw forcing was integrated over time and band-passed to yield the corresponding zonal wind response (see Eq. (1)). For example, uGWF is the integrated QBO zonal wind (response wind) due to GWF. Following this procedure, the linear regressions were calculated. Note that the trends in the forcing amplitudes are not additive, even though forcings are additive in the temporal space, owing to their different phases (Fig. 11).

Fig. 20. QBO forcing evolution. Same as Fig. 16 except for (a) the parameterized gravity forcing (GWF), (b) the momentum vertical advection (WUZ), and (c) the resolved wave forcing (EPD).

Figure 21 displays the vertical profiles of GWF and uGWF from 70 to 5 hPa for all simulations. The GHG-increase simulations exhibit the same statistically significant negative trend at 30 ant 70 hPa and, in most cases, at 50 hPa as well, with a maximum of about 0.2x10-2 ms-1 day-1

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

result , an Sc ensemble simulation should be performed.

either, if the signs of the forcing trends are not the same.

simulations for (a) B21, (b) Sb, and (c) Sc. Units are 10-3 ms-1.

(a) (b) (c)

Fig. 24. Residual vertical velocity evolution at 30 hPa. Evolutions of the residual vertical velocity for the low-frequency component (period longer than 120 months, solid) and the amplitude of the residual vertical velocity of the QBO frequency (dashed) at 30 hPa in

induced zonal wind uEPD.

**6. Discussion** 

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 377

Fig. 23. EPD amplitude evolution Same as Fig.19 except for (left) EPD and (right) the EPD-

The GHG-increase simulations have thus far demonstrated that the QBO amplitude decreases linearly below 20 hPa with rates of 0.1-0.3 ms-1 per decade and increases above 10 hPa with rates of 0-0.1 ms -1 per decade (Fig. 19). As shown by the difference between the B2 and Sb simulations, the trends of the B2 members tend to be smaller below 20 hPa than the Sb trend, but the difference between them is not substantial. That is, the inter-member variations are too large to ascribe the difference between the B2 members and Sb to the forcing difference, that is, ODSs. On the other hand, the small but statistical significant increasing trends of less than 0.1 ms -1 per decade at 30 hPa and above exhibited by the noclimate-change simulation is likely an effect of the ODS decrease, although to confirm this

The cause of the QBO trend should reflect the trends of the forcings, GWF, EPD, and WUZ. However, because the forcings include not only external quantities but also internal ones related to the QBO, separating the forcings from the responses is not straightforward. For example, the propagation of the resolved waves and that of the parameterized gravity waves strongly depend on the background QBO wind, as does the deposition of their momentum on the QBO. The vertical advection WUZ is also affected by the QBO, because WUZ is explicitly expressed as the product of the QBO and the background wind field (Eq. (1)). In addition, the trends in the forcing amplitude are not additive because of the phase differences as stated before; thus, evaluation of their total effects is not straightforward

per decade for GWF at 70 hPa. At 20 hPa the trends are dispersed around zero, and at 10 and 5 hPa, the GWF shows statistically significant positive values of nearly the same magnitude. The trends of integrated zonal wind uGWF are qualitatively similar to those of GWF but the statistically significances are not necessarily so.

Fig. 21. GWF amplitude evolution. Same as Fig.19 except for (left) GWF and (right) the GWF-induced zonal wind uGWF, which was calculated by time-integration of GWF. Units are 10-2 ms-1day-1 per century for GWF and ms-1 per decade for uGWF.

For example, while most GHG-increase simulations show a statistical significant negative trend at 70 hPa and positive trend above 10 hPa both in uGWF and in GWF, the number of statistically significant trends decreases at 50 and 30 hPa and increases at 20 hPa in uGWF compared with GWF. The no-climate-change simulation exhibits statistically significant positive trends of about 0.1x10-2 ms-1 day-1 per decade at 20 and 30 hPa, and a negative trend of similar magnitude at 5 hPa for GWF.

The vertical transport of zonal momentum WUZ shows a statistically significant trend at almost all altitudes in the GHG-increase simulations (Fig. 22). A negative trend of about 0.1x10-2 ms-1 day-1 per decade is seen at 30 and 50 hPa, and a positive trend of nearly the same magnitude at 70 hPa and above 20 hPa. The no-climate-change simulation exhibits almost no trends at any altitude in both WUZ and uWUZ, and most apparent trends are thus not statistically significant. The resolved-wave forcing EPD and uEPD exhibit a vertically flat structure (Fig. 23), and the trend magnitudes are smaller than those of the other forcings in all simulations; nevertheless, most of the trends are statistically significant.

Fig. 22. WUZ amplitude evolution. Same as Fig.19 except for (left) WUZ and (right) the WUZ-induced zonal wind uWUZ.

Fig. 23. EPD amplitude evolution Same as Fig.19 except for (left) EPD and (right) the EPDinduced zonal wind uEPD.

## **6. Discussion**

376 Greenhouse Gases – Emission, Measurement and Management

per decade for GWF at 70 hPa. At 20 hPa the trends are dispersed around zero, and at 10 and 5 hPa, the GWF shows statistically significant positive values of nearly the same magnitude. The trends of integrated zonal wind uGWF are qualitatively similar to those of

Fig. 21. GWF amplitude evolution. Same as Fig.19 except for (left) GWF and (right) the GWF-induced zonal wind uGWF, which was calculated by time-integration of GWF. Units are

For example, while most GHG-increase simulations show a statistical significant negative trend at 70 hPa and positive trend above 10 hPa both in uGWF and in GWF, the number of statistically significant trends decreases at 50 and 30 hPa and increases at 20 hPa in uGWF compared with GWF. The no-climate-change simulation exhibits statistically significant positive trends of about 0.1x10-2 ms-1 day-1 per decade at 20 and 30 hPa, and a negative trend

The vertical transport of zonal momentum WUZ shows a statistically significant trend at almost all altitudes in the GHG-increase simulations (Fig. 22). A negative trend of about 0.1x10-2 ms-1 day-1 per decade is seen at 30 and 50 hPa, and a positive trend of nearly the same magnitude at 70 hPa and above 20 hPa. The no-climate-change simulation exhibits almost no trends at any altitude in both WUZ and uWUZ, and most apparent trends are thus not statistically significant. The resolved-wave forcing EPD and uEPD exhibit a vertically flat structure (Fig. 23), and the trend magnitudes are smaller than those of the other forcings in

all simulations; nevertheless, most of the trends are statistically significant.

Fig. 22. WUZ amplitude evolution. Same as Fig.19 except for (left) WUZ and (right) the

GWF but the statistically significances are not necessarily so.

10-2 ms-1day-1 per century for GWF and ms-1 per decade for uGWF.

of similar magnitude at 5 hPa for GWF.

WUZ-induced zonal wind uWUZ.

The GHG-increase simulations have thus far demonstrated that the QBO amplitude decreases linearly below 20 hPa with rates of 0.1-0.3 ms-1 per decade and increases above 10 hPa with rates of 0-0.1 ms -1 per decade (Fig. 19). As shown by the difference between the B2 and Sb simulations, the trends of the B2 members tend to be smaller below 20 hPa than the Sb trend, but the difference between them is not substantial. That is, the inter-member variations are too large to ascribe the difference between the B2 members and Sb to the forcing difference, that is, ODSs. On the other hand, the small but statistical significant increasing trends of less than 0.1 ms -1 per decade at 30 hPa and above exhibited by the noclimate-change simulation is likely an effect of the ODS decrease, although to confirm this result , an Sc ensemble simulation should be performed.

The cause of the QBO trend should reflect the trends of the forcings, GWF, EPD, and WUZ. However, because the forcings include not only external quantities but also internal ones related to the QBO, separating the forcings from the responses is not straightforward. For example, the propagation of the resolved waves and that of the parameterized gravity waves strongly depend on the background QBO wind, as does the deposition of their momentum on the QBO. The vertical advection WUZ is also affected by the QBO, because WUZ is explicitly expressed as the product of the QBO and the background wind field (Eq. (1)). In addition, the trends in the forcing amplitude are not additive because of the phase differences as stated before; thus, evaluation of their total effects is not straightforward either, if the signs of the forcing trends are not the same.

Fig. 24. Residual vertical velocity evolution at 30 hPa. Evolutions of the residual vertical velocity for the low-frequency component (period longer than 120 months, solid) and the amplitude of the residual vertical velocity of the QBO frequency (dashed) at 30 hPa in simulations for (a) B21, (b) Sb, and (c) Sc. Units are 10-3 ms-1.

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

result, BDT shows a slightly larger decreasing trend at 50 hPa than at 30 hPa.

vertical shear, which is very small as well as its tendency in the lower stratosphere.

radiation are needed.

At 70 hPa, the BDT components are a rapidly increasing \**<sup>L</sup> w* and a very slowly decreasing (/)*<sup>Q</sup> u z* , leading to an increasing trend in BDT. The SCT components are both increasing but very small, so SCT is much smaller than BDT. At 20 hPa, both SCT components are increasing, so SCT has a positive trend. Although (/)*<sup>Q</sup> u z* is decreasing, BDT has a positive trend comparable to the SCT trend. These results indicate that the BDT trend is of similar magnitude to the SCT trend at 30 and 20 hPa, while the SCT trend is much smaller than the BDT trend at 50 and 70 hPa. This altitude dependence of SCT derives mainly from the low-frequency

In this study the source strength of GWF was fixed in all simulations, irrespective of the ODS evolution and/or the GHG-induced global warming, wherein convective precipitation increased (e.g., Yukimoto et al., 2006; IPCC, 2007) as manifested in the strong warming in the tropical upper troposphere (Fig. 4). This fixation of source strength is different from the precedent simulations of equilibrium CO2 doubling. Giorgetta & Doege (2005) varied the GWF source by 0% to 20% and showed that a statistically significant reduction in the QBO period occurred in runs with a 10% or 20% increase. Kawatani et al. (2011) demonstrated that wave momentum fluxes increase by 10%–15% in the equatorial lower stratosphere, though those relevant to the QBO forcing do not increase as much. Since tropical convective precipitation is closely linked to gravity and equatorial wave generation (e.g., Horinouchi et al., 2003), we assume that the fixation of the GWF source differs more or less from reality. Thus, to specify appropriate changes in the GWF source, observations of relations between the QBO and other independent quantities such as SST/precipitation or outgoing longwave

For example, Taguchi (2010) demonstrated a connection between the QBO and El Niño-Southern Oscillation (ENSO): the QBO signals have a weaker amplitude and faster phase

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 379

decade (Fig. 22). To explore the cause of this, we evaluated the trend of each component, that is, BDT and SCT, of WUZ (Eq. 1). Figure 24 depicts the evolution of the residual vertical velocities of the low-frequency and the QBO at 30 hPa in simulations for B21, Sb, and Sc, and Fig. 25 displays the evolution of the vertical shears. The tendencies of the residual vertical velocity and the vertical shear at 50 hPa are very similar to those at 30 hPa, aside from the background levels. Figures 26 and 27 depict the same evolutions as Figs. 25 and 26, but at 70 hPa. The tendency of WUZ depends on those of BDT and SCT (Eq. (1)), each of which is further decomposed into two terms. As a result, to evaluate of the tendency of WUZ it is necessary to know the tendencies of four terms: low-frequency residual vertical velocity \**<sup>L</sup> w* , QBO vertical shear (/)*<sup>Q</sup> u z* , QBO residual vertical velocity \**<sup>Q</sup> w* , and lowfrequency vertical shear (/)*<sup>L</sup> u z* . It should be noted that \**<sup>L</sup> w* always shows an increasing trend in all of the GHG-increase simulations, but the trends of the other three terms depend on altitude. At 30 hPa in the GHG-increase simulations, they are decreasing; thus, SCT also has a negative trend because of the decrease in \**<sup>Q</sup> w* and (/)*<sup>L</sup> u z* . On the other hand, BDT has both an increasing component \**<sup>L</sup> w* and a decreasing component (/)*<sup>Q</sup> u z* , resulting in a decreasing trend comparable to that of SCT. At 50 hPa the tendencies of the four terms are qualitatively similar to those at 30 hPa, but the SCT tendency is much smaller than the BDT tendency, in which the decrease in (/)*<sup>Q</sup> u z* is much larger than the increase in \**<sup>L</sup> w* . As a

Fig. 25. Zonal wind vertical shear evolution at 30 hPa. Same as Fig.24 except for the vertical shear of zonal wind. Units are 10-3 s-1.

Almost all of the trends in the forcings and their response winds in the GHG-increase simulations exhibit negative trends at 30 hPa, where the trend of the QBO wind is the largest negative; therefore, the forcings and responses are qualitatively consistent. At 50 hPa, the magnitude of the negative trend in WUZ is significantly larger than that of the GWF (small negative) or the EPD trend (nearly zero or small positive), which is also consistent with the negative trend in the QBO wind. At 20 hPa, the trends are on average nearly zero in GWF, have small positive values in WUZ, and small negative values in EPD, as are those of their responses, which is not consistent with the small but significant negative trend in the QBO wind. At 70 hPa, GWF exhibits a statistically significant negative trend, and its magnitude is largest compared with the magnitudes of the trends at other altitudes. WUZ, on the other hand, exhibits a statistically significant positive trend, the magnitude of which is about three-fourths that of GWF. EPD tends to have a small negative trend, but its response is nearly zero in the ensemble mean, indicating that EPD plays a minor role at 70 hPa.

Fig. 26. Residual vertical velocity evolution at 70 hPa. Same as Fig.24 except at 70 hPa.

Fig. 27. Zonal wind vertical shear evolution at 70 hPa. Same as Fig.25 except at 70 hPa.

WUZ trends in the GHG-increase simulations are statistically significant below the middle stratosphere but they show sharp contrasts in sign, being negative at 70 hPa and positive at 50 and 30 hPa, although their magnitudes are almost the same at 0.1x10-2 ms-1day-1 per

Fig. 25. Zonal wind vertical shear evolution at 30 hPa. Same as Fig.24 except for the vertical

Almost all of the trends in the forcings and their response winds in the GHG-increase simulations exhibit negative trends at 30 hPa, where the trend of the QBO wind is the largest negative; therefore, the forcings and responses are qualitatively consistent. At 50 hPa, the magnitude of the negative trend in WUZ is significantly larger than that of the GWF (small negative) or the EPD trend (nearly zero or small positive), which is also consistent with the negative trend in the QBO wind. At 20 hPa, the trends are on average nearly zero in GWF, have small positive values in WUZ, and small negative values in EPD, as are those of their responses, which is not consistent with the small but significant negative trend in the QBO wind. At 70 hPa, GWF exhibits a statistically significant negative trend, and its magnitude is largest compared with the magnitudes of the trends at other altitudes. WUZ, on the other hand, exhibits a statistically significant positive trend, the magnitude of which is about three-fourths that of GWF. EPD tends to have a small negative trend, but its response is

nearly zero in the ensemble mean, indicating that EPD plays a minor role at 70 hPa.

Fig. 26. Residual vertical velocity evolution at 70 hPa. Same as Fig.24 except at 70 hPa.

Fig. 27. Zonal wind vertical shear evolution at 70 hPa. Same as Fig.25 except at 70 hPa.

WUZ trends in the GHG-increase simulations are statistically significant below the middle stratosphere but they show sharp contrasts in sign, being negative at 70 hPa and positive at 50 and 30 hPa, although their magnitudes are almost the same at 0.1x10-2 ms-1day-1 per

(a) (b) (c)

(a) (b) (c)

(a) (b) (c)

shear of zonal wind. Units are 10-3 s-1.

decade (Fig. 22). To explore the cause of this, we evaluated the trend of each component, that is, BDT and SCT, of WUZ (Eq. 1). Figure 24 depicts the evolution of the residual vertical velocities of the low-frequency and the QBO at 30 hPa in simulations for B21, Sb, and Sc, and Fig. 25 displays the evolution of the vertical shears. The tendencies of the residual vertical velocity and the vertical shear at 50 hPa are very similar to those at 30 hPa, aside from the background levels. Figures 26 and 27 depict the same evolutions as Figs. 25 and 26, but at 70 hPa. The tendency of WUZ depends on those of BDT and SCT (Eq. (1)), each of which is further decomposed into two terms. As a result, to evaluate of the tendency of WUZ it is necessary to know the tendencies of four terms: low-frequency residual vertical velocity \**<sup>L</sup> w* , QBO vertical shear (/)*<sup>Q</sup> u z* , QBO residual vertical velocity \**<sup>Q</sup> w* , and lowfrequency vertical shear (/)*<sup>L</sup> u z* . It should be noted that \**<sup>L</sup> w* always shows an increasing trend in all of the GHG-increase simulations, but the trends of the other three terms depend on altitude. At 30 hPa in the GHG-increase simulations, they are decreasing; thus, SCT also has a negative trend because of the decrease in \**<sup>Q</sup> w* and (/)*<sup>L</sup> u z* . On the other hand, BDT has both an increasing component \**<sup>L</sup> w* and a decreasing component (/)*<sup>Q</sup> u z* , resulting in a decreasing trend comparable to that of SCT. At 50 hPa the tendencies of the four terms are qualitatively similar to those at 30 hPa, but the SCT tendency is much smaller than the BDT tendency, in which the decrease in (/)*<sup>Q</sup> u z* is much larger than the increase in \**<sup>L</sup> w* . As a result, BDT shows a slightly larger decreasing trend at 50 hPa than at 30 hPa.

At 70 hPa, the BDT components are a rapidly increasing \**<sup>L</sup> w* and a very slowly decreasing (/)*<sup>Q</sup> u z* , leading to an increasing trend in BDT. The SCT components are both increasing but very small, so SCT is much smaller than BDT. At 20 hPa, both SCT components are increasing, so SCT has a positive trend. Although (/)*<sup>Q</sup> u z* is decreasing, BDT has a positive trend comparable to the SCT trend. These results indicate that the BDT trend is of similar magnitude to the SCT trend at 30 and 20 hPa, while the SCT trend is much smaller than the BDT trend at 50 and 70 hPa. This altitude dependence of SCT derives mainly from the low-frequency vertical shear, which is very small as well as its tendency in the lower stratosphere.

In this study the source strength of GWF was fixed in all simulations, irrespective of the ODS evolution and/or the GHG-induced global warming, wherein convective precipitation increased (e.g., Yukimoto et al., 2006; IPCC, 2007) as manifested in the strong warming in the tropical upper troposphere (Fig. 4). This fixation of source strength is different from the precedent simulations of equilibrium CO2 doubling. Giorgetta & Doege (2005) varied the GWF source by 0% to 20% and showed that a statistically significant reduction in the QBO period occurred in runs with a 10% or 20% increase. Kawatani et al. (2011) demonstrated that wave momentum fluxes increase by 10%–15% in the equatorial lower stratosphere, though those relevant to the QBO forcing do not increase as much. Since tropical convective precipitation is closely linked to gravity and equatorial wave generation (e.g., Horinouchi et al., 2003), we assume that the fixation of the GWF source differs more or less from reality. Thus, to specify appropriate changes in the GWF source, observations of relations between the QBO and other independent quantities such as SST/precipitation or outgoing longwave radiation are needed.

For example, Taguchi (2010) demonstrated a connection between the QBO and El Niño-Southern Oscillation (ENSO): the QBO signals have a weaker amplitude and faster phase

Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

very small vertical shear trend of the background wind.

http://paos.colorado.edu/research/wavelets/.

10.1175/2009JAS3292.1, 2010.

doi:10.1029/2006GL026867.

doi:10.1029/2004GL022328.

**8. Acknowledgements** 

**9. References** 

Press.

uncertainties in the effects of the QBO in the future than in the QBO itself.

and tropopause height. *J. Atmos. Sci.*, 21, 479– 492.

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 381

The QBO was spontaneously generated with a period of about 28-month period in the equatorial stratosphere by MRI-CCM under both past and current conditions in all simulations. The QBO frequency was very stable without any substantial trend in all simulations, and the QBO amplitude showed a small but statistically significant decreasing trend below 20 hPa with a maximum (~0.3 ms-1 per decade) at 30 hPa in GHG-increase simulations. ODS decreases in the future (i.e., ozone recovery) led to very small increasing trend (<0.1 ms-1 per decade) in the QBO amplitude above 20 hPa. The causes of the QBO amplitude trend were largely changes in the parameterized gravity wave forcing and the vertical advection of zonal wind momentum. In the QBO trend, the vertical advection was brought about not only by the Brewer-Dobson circulation but also by the QBO secondary circulation above the middle stratosphere, while the Brewer-Dobson circulation was dominant in the lower stratosphere, mainly because of the very weak vertical shear and the

The QBO effects globally extend from the stratosphere to the troposphere, and thus the projected decreasing trend of the QBO amplitude in the CCM simulations would likely cause, more or less, changes in various phenomena in the whole atmosphere. However, since the current CCM simulations included only forcings of SST, GHGs, and ODSs, other forcings such as solar irradiance variability and volcanic aerosols would induce different responses in the QBO, depending on the timescale of these forcings. In addition, different scenarios in GHGs result in different SSTs, and thereby likely in different QBO trends. Therefore, there are larger

Some of the MRI-CCM simulations were performed by using the supercomputer at the National Institute for Environmental Studies, Japan. This work was supported in part by a Grant-in-Aid (20340129, 20340131, and 20244076) for Scientific Research (KAKENHI) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. Wavelet software was provided by C. Torrence & G. Compo, and is available at the URL:

Andrews, D. G., Holton, J. R. & Leovy, C. B. (1987). *Middle Atmosphere Dynamics,* Academic

Angell, J. K. & Korshover, J. (1964). Quasi-biennial variations in temperature, total ozone,

Anstey, J. A., Shepherd, T. G., Scinocca, J. F. (2010). Influence of the quasi-biennial oscillation

Austin, J. & Li, F. (2006). On the relationship between the strength of the Brewer-Dobson

Baldwin, M. P. & Gray, L. J. (2005). Tropical stratospheric zonal winds in ECMWF ERA-40

on the extratropical winter stratosphere in an atmospheric general circulation model and in reanalysis data, *J. Atmos. Sci.*, 67, 1402-1419, doi:

circulation and the age of stratospheric air. Geophys. Res. Lett., 33, L17807,

reanalysis, rocketsonde data, and rawinsonde data. *Geophys. Res. Lett.*, 32, L09806,

propagation under El Niño conditions, when the west-east contrast of SST and precipitation drastically changes in the tropical Pacific (e.g., Philander, 1990). However, since the tropical upwelling is also enhanced (Randel et al., 2009) during El Niño conditions, isolation of the SST/precipitation effect is required for evaluation of the GWF source in the troposphere. On the other hand, according to Fischer and Tung (2008), the QBO period does not show any trend but has been stable within 24-32 months for five decades (from 1953 to 2007) despite period variations due mainly to stalling of the westerly phase. These findings indicate that further investigation is required to identify the tropospheric factors affecting the QBO.

## **7. Conclusion**

To investigate future changes in the QBO in the tropical stratosphere, transient simulations were carried out for the period from 1960 to 2100 with MRI-CCM by inputting observed and projected GHGs and/or ODSs at the surface. SST/sea ice, generated by an MRI coupled ocean-atmosphere model, was also specified so as to be compatible with the GHGs abundances. Three types of simulations were performed to evaluate the effects of a GHGs increase and an ODSs decrease, both in combination and separately: The first type used the REF-B2 scenario of CCMVal-2, in which both GHGs and ODSs evolve following observed values for past simulations, and specified SRES A1B scenario for GHGs and the adjusted A1 scenario for ODSs for future simulations. The second type of simulation used an SCN-B2b scenario, in which GHGs were the same as in REF-B2 but ODSs were fixed at 1960 levels. The third type used an SCN-B2c scenario, in which ODSs were the same as in REF-B2 but GHGs/SST were fixed at 1960 levels. The REF-B2 simulation was carried out with three members, and the other two were performed with a single member.

The future climate change due to increasing GHGs was characterized as tropospheric warming as a result of SST warming (indirect CO2 effect), with maximum warming in the tropical upper troposphere, and cooling in the stratosphere (direct CO2 effect), with maximum cooling in the upper stratosphere. Zonal wind changes were characterized by westerly wind intensification from the upper region of the subtropical jet upward along the equatorward flank of the polar night jet axis, reaching a maximum at around 70 hPa and 40° in both hemispheres. In the SH, the westerly wind intensification also extends down to the surface at around 55°S, resulting in poleward spreading of the subtropical jet. In the tropics, westerly wind strengthening (eastward acceleration) occurred above 30 hPa and weakening (westward acceleration) at 50 hPa, below which very weak strengthening occurred down to the surface. The Brewer-Dobson circulation was similarly intensified in all of the GHGincrease simulations.

The future climate change in the no-climate-change simulation was characterized by strong warming in the upper stratosphere above about 5 hPa in both hemispheres, except in the southern high latitudes, where prominent warming occurred in the lower stratosphere at 300-30 hPa, centered at 100 hPa, with cooling above 30 hPa. This pattern of temperature change was due solely to the ODSs decrease, or, equivalently, ozone recovery. The zonal wind change characterized by slight intensification of the polar night jet in the poleward flank above the middle stratosphere in the NH. In the SH, weakening of the polar night jet and both weakening of the poleward and strengthening of the equatorward flank of the subtropical jet down to the surface occurred. In the tropics, there was almost no change, including in the Brewer-Dobson circulation.

The QBO was spontaneously generated with a period of about 28-month period in the equatorial stratosphere by MRI-CCM under both past and current conditions in all simulations. The QBO frequency was very stable without any substantial trend in all simulations, and the QBO amplitude showed a small but statistically significant decreasing trend below 20 hPa with a maximum (~0.3 ms-1 per decade) at 30 hPa in GHG-increase simulations. ODS decreases in the future (i.e., ozone recovery) led to very small increasing trend (<0.1 ms-1 per decade) in the QBO amplitude above 20 hPa. The causes of the QBO amplitude trend were largely changes in the parameterized gravity wave forcing and the vertical advection of zonal wind momentum. In the QBO trend, the vertical advection was brought about not only by the Brewer-Dobson circulation but also by the QBO secondary circulation above the middle stratosphere, while the Brewer-Dobson circulation was dominant in the lower stratosphere, mainly because of the very weak vertical shear and the very small vertical shear trend of the background wind.

The QBO effects globally extend from the stratosphere to the troposphere, and thus the projected decreasing trend of the QBO amplitude in the CCM simulations would likely cause, more or less, changes in various phenomena in the whole atmosphere. However, since the current CCM simulations included only forcings of SST, GHGs, and ODSs, other forcings such as solar irradiance variability and volcanic aerosols would induce different responses in the QBO, depending on the timescale of these forcings. In addition, different scenarios in GHGs result in different SSTs, and thereby likely in different QBO trends. Therefore, there are larger uncertainties in the effects of the QBO in the future than in the QBO itself.

## **8. Acknowledgements**

380 Greenhouse Gases – Emission, Measurement and Management

propagation under El Niño conditions, when the west-east contrast of SST and precipitation drastically changes in the tropical Pacific (e.g., Philander, 1990). However, since the tropical upwelling is also enhanced (Randel et al., 2009) during El Niño conditions, isolation of the SST/precipitation effect is required for evaluation of the GWF source in the troposphere. On the other hand, according to Fischer and Tung (2008), the QBO period does not show any trend but has been stable within 24-32 months for five decades (from 1953 to 2007) despite period variations due mainly to stalling of the westerly phase. These findings indicate that further investigation is required to identify the tropospheric factors affecting the QBO.

To investigate future changes in the QBO in the tropical stratosphere, transient simulations were carried out for the period from 1960 to 2100 with MRI-CCM by inputting observed and projected GHGs and/or ODSs at the surface. SST/sea ice, generated by an MRI coupled ocean-atmosphere model, was also specified so as to be compatible with the GHGs abundances. Three types of simulations were performed to evaluate the effects of a GHGs increase and an ODSs decrease, both in combination and separately: The first type used the REF-B2 scenario of CCMVal-2, in which both GHGs and ODSs evolve following observed values for past simulations, and specified SRES A1B scenario for GHGs and the adjusted A1 scenario for ODSs for future simulations. The second type of simulation used an SCN-B2b scenario, in which GHGs were the same as in REF-B2 but ODSs were fixed at 1960 levels. The third type used an SCN-B2c scenario, in which ODSs were the same as in REF-B2 but GHGs/SST were fixed at 1960 levels. The REF-B2 simulation was carried out with three

The future climate change due to increasing GHGs was characterized as tropospheric warming as a result of SST warming (indirect CO2 effect), with maximum warming in the tropical upper troposphere, and cooling in the stratosphere (direct CO2 effect), with maximum cooling in the upper stratosphere. Zonal wind changes were characterized by westerly wind intensification from the upper region of the subtropical jet upward along the equatorward flank of the polar night jet axis, reaching a maximum at around 70 hPa and 40° in both hemispheres. In the SH, the westerly wind intensification also extends down to the surface at around 55°S, resulting in poleward spreading of the subtropical jet. In the tropics, westerly wind strengthening (eastward acceleration) occurred above 30 hPa and weakening (westward acceleration) at 50 hPa, below which very weak strengthening occurred down to the surface. The Brewer-Dobson circulation was similarly intensified in all of the GHG-

The future climate change in the no-climate-change simulation was characterized by strong warming in the upper stratosphere above about 5 hPa in both hemispheres, except in the southern high latitudes, where prominent warming occurred in the lower stratosphere at 300-30 hPa, centered at 100 hPa, with cooling above 30 hPa. This pattern of temperature change was due solely to the ODSs decrease, or, equivalently, ozone recovery. The zonal wind change characterized by slight intensification of the polar night jet in the poleward flank above the middle stratosphere in the NH. In the SH, weakening of the polar night jet and both weakening of the poleward and strengthening of the equatorward flank of the subtropical jet down to the surface occurred. In the tropics, there was almost no change,

members, and the other two were performed with a single member.

**7. Conclusion** 

increase simulations.

including in the Brewer-Dobson circulation.

Some of the MRI-CCM simulations were performed by using the supercomputer at the National Institute for Environmental Studies, Japan. This work was supported in part by a Grant-in-Aid (20340129, 20340131, and 20244076) for Scientific Research (KAKENHI) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. Wavelet software was provided by C. Torrence & G. Compo, and is available at the URL: http://paos.colorado.edu/research/wavelets/.

## **9. References**


Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

doi:10.1029/2002GL014756.

10.1002/qj.49712757416.

*1*, University of Tokyo.

*J. Atmos. Sci.*, 60, 2765-2782.

2393–2418.

3829– 3846.

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 383

Fischer, P. & Tung, K. K. (2008). A reexamination of the QBO period modulation by the solar

Garcia, R. R. & Randel, W. J. (2008). Acceleration of the Brewer-Dobson circulation due to

Giorgetta, M. A. & Doege, M. C. (2005). Sensitivity of the quasi-biennial oscillation to CO2 doubling. *Geophys. Res. Lett.*, 32, L08701, doi:10.1029/2004GL021971. Giorgetta, M. A., Manzini, E. & Roeckner, E. (2002). Forcing of the quasi-biennial oscillation

Gray, L. J., Drysdale, E. F., Lawrence, B. N. & Dunkerton, T. J. (2001). Model studies of the

Hamilton, K. (1998). Effects of an imposed quasi-biennial oscillation in a comprehensive

Hamilton, K., Wilson, R. J. & Hemler, R. S. (1999). Middle atmosphere simulated with high

Hasumi, H. & S. Emori, Eds. (2004). K-1 coupled GCM (MIROC) description. *K-1 Tech. Rep.* 

Hines, C. O. (1997). Doppler-spread parameterization of gravity-wave momentum

Holton, J. R. & Tan, H. -C. (1980). The influence of the equatorial quasi-biennial oscillation

Holton, J. R. & Tan, H. -C. (1982). The quasi-biennial oscillation in the Northern Hemisphere

Horinouchi, T., Pawson, S., Shibata, K., Langematz, U., Manzini, E., Giorgetta, M. A., Sassi,

IPCC (2000). Special Report on Emissions Scenarios: A Special Report of Working Group III

IPCC (2007). Climate Change 2007: The Physical Science Basis. Contribution of Working

IPCC/TEAP (2005). IPCC/TEAP Special Report on Safeguarding the Ozone Layer and the

Kawatani, Y., Hamilton, K. & Watanabe, S. (2011). The quasi-biennial oscillation in a double CO2 climate. *J. Atmos. Sci.*, 68, 265–283, doi: 10.1175/2010JAS3623.1.

*Summary for Policymakers and Technical Summary*, WMO and UNEP.

and implementation. *J. Atmos. Sol. Terr. Phys.*, 59, 387–400.

lower stratosphere. *J. Meteorol. Soc. Japan*, 60, 140–148.

(eds)), Cambridge University Press, Cambridge, U.K.

United Kingdom and New York, NY, USA.

on the global circulation at 50 mb. *J. Atmos. Sci.*, 37, 2200–2208.

from a broad spectrum of atmospheric waves. *Geophys. Res. Lett.*, 29(8), 1245,

interannual variability of the northern-hemisphere stratospheric winter circulation: The role of the quasi-biennial oscillation. *Q. J. R. Meteorol. Soc.*, 127, 1413–1432, doi:

troposphere–stratosphere–mesosphere general circulation model. *J. Atmos. Sci.,* 55,

vertical and horizontal resolution versions of a GCM: Improvements in the cold pole bias and generation of a QBO-like oscillation in the tropics. *J. Atmos. Sci.*, 56,

deposition in the middle atmosphere. 2. Broad and quasi-monochromatic spectra,

F., Wilson, R. J., Hamilton, K., de Grandpre, J. & Scaife, A. A. (2003). Tropical cumulus convection and upward propagating waves in middle atmospheric GCMs.

of the Intergovernmental Panel on Climate Change (Nakicenovic, N. & R. Swart

Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor & H.L. Miller (eds.)). Cambridge University Press, Cambridge,

Global Climate System: Issues Related to Hydrofluorocarbons and Perfluorocarbons.

cycle. *J. Geophys. Res.*, 113, D07114, doi:10.1029/2007JD008983.

increases in greenhouse gases. *J. Atmos. Sci.*, 65, 2731-2739.


Baldwin, M. P., Gray, L. J., Dunkerton, T. J., Hamilton, K., Haynes, P. H., Randel, W. J.,

Butchart, N., Scaife, A. A., Austin, J., Hare, S. H. E. & Knight, J. R. (2003). Quasi-biennial

Butchart, N., Scaife, A. A., Bourqui, M., de Grandpre, J., Hare, S. H. E., Kettleborough, J.,

circulation, *Climate Dynamics*, 27, doi:10.1007/s00382-006-0162-4, 727-741. Chubachi, S. (1985). A special ozone observation at Syowa Station, Antarctica from February

Coughlin, K. & Tung, K. -K. (2001). QBO signal found at the extratropical surface through

Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P., Kobayashi, S., Andrae, U.,

Deushi, M. & Shibata, K. (2011). Impacts of increases in greenhouse gases and ozone

Duchon, C. E. (1979). Lanczos filtering in one and two dimensions. *J. Appl. Meteorol.,* 18,

Dunkerton, T. (2001). Quasi-biennial and subbiennial variations of stratospheric trace constituents derived from HALOE observations. *J. Atmos. Sci.*, 57, 7-25. Engel, A., Möbius, T., Bönisch, H., Schmidt, U., Heinz, R., Levin, I., Atlas, E., Aoki, S.,

coupled chemistry-climate models. *Bull. Am. Meteorol. Soc.*, 86, 1117–1133. Eyring, V., Chipperfield, M. P., Giorgetta, M. A., Kinnison, D. E., Manzini, E., Matthes, K.,

Farman, J. C., Gardiner, B. G. & Shanklin, J. D. (1985). Large losses of total ozone in Antarctica reveal seasonal ClOx/NOx interaction. *Nature*, 315, 207-210.

northern annular modes. *Geophys. Res. Lett.*, 28, 4563–4566.

*Q. J. R. Meteorol. Soc.* 137: 553–597. DOI:10.1002/qj.828.

*Rev. Geophys.*, 39, 179–229.

2010JD015361.

*Newsletter*, No. 30, p.20–26.

1016–1022.

108(D15), 4486, doi:10.1029/2002JD003004.

Holton, J. R., Alexander, M. J., Hirota, I., Horinouchi, T., Jones, D. B. A., Kinnersley, J. S. Marquardt, C., Sato, K. & Takahashi, M. (2001). The quasi-biennial oscillation,

oscillation in ozone in a coupled chemistry-climate model, *J. Geophys. Res.*,

Langematz, U., Manzini, E., Sassi, F., Shibata, K., Shindell, D. & Sigmond, M. (2006). Simulation of anthropogenic change in the strength of the Brewer-Dobson

1982 to January 1983. *Proceedings of the Quadrennial Ozone Symposium 1984*, 285-289.

Balmaseda, M. A., Balsamo, G., Bauer, P., Bechtold, P., Beljaars, A. C. M., van de Berg L., Bidlot J., Bormann N., Delsol C., Dragani R., Fuentes M., Geer A.J., Haimberger L., Healy S.B., Hersbach H., Hólm E.V., Isaksen L., Kållberg P., Köhler M., Matricardi M., McNally A. P., Monge-Sanz B. M., Morcrette J.-J., Park B-K, Peubey C., de Rosnay P., Tavolato C., Thépaut J.-N. & Vitart F. (2011). The ERA-Interim reanalysis: configuration and performance of the data assimilation system.

recovery on lower stratospheric circulation and the age of air: Chemistry-climate model simulations up to 2100. *J. Geophys. Res.*, 116, D10302, doi:10.1029/

Nakazawa, T., Sugawara, S., Moore, F., Hurst, D., Elkins, J., Schauffler, S., Andrews, A. & Boering, K. (2009). Age of stratospheric air unchanged within uncertainties over the past 30 years. *Nat. Geosci.*, 2, 28–31, doi:10.1038/NGEO388. Eyring, V., Harris, N. R. P., Rex, M., Shepherd, T. G., Fahey, D. W., Amanatidis, G. T., Austin,

J., Chipperfield, M. P., Dameris, M., Forster, P. M. De F., Gettelman, A., Graf, H. F., Nagashima, T., Newman, P. A., Pawson, S., Prather, M. J., Pyle, J. A., Salawitch, R. J., Santer, B. D. & Waugh, D. W. (2005). A strategy for process oriented validation of

Newman, P. A., Pawson, S., Shepherd, T. G. & Waugh, D. W. (2008). Overview of the New CCMVal reference and sensitivity simulations in support of upcoming ozone and climate assessments and the planned SPARC CCMVal. *SPARC* 


Future Changes in the Quasi-Biennial Oscillation Under a Greenhouse Gas

beta-plane. *Q. J. Roy. Meteorol. Soc*., 108, 335–352.

decomposition. *J. Atmos. Sci.,* 53, 2546-2559.

L15822, doi:10.1029/2009GL039343.

*Geophys. Res.*, 66, 813– 818.

doi:10.1029/2006GL026731.

*Res. Lett.*, 27, 3481–3484.

2005GL022885.

2005GL023433.

No.1526.

Increase and Ozone Recovery in Transient Simulations by a Chemistry-Climate Model 385

Plumb, R. A. & Bell, R. C. (1982). A model of the quasi-biennial oscillation on an equatorial

Randel, W. J. & Wu, F. (1996). Isolation of the ozone QBO in SAGE II data by singular value

Randel, W. J., Garcia, R. R., Calvo, N. & Marsh, D. (2009). ENSO influence on zonal mean

Reed, R. J., Campbell, W. J., Rasmussen, L. A. & Rogers, D. G. (1961). Evidence of a

Rex, M., Salawitch, R. J., Deckelmann, H., von der Gathen, P., Harris, N. R. P., Chipperfield,

Roeckner, E., Bäuml, G., Bonaventura, L., Brokopf, R., Esch, M., Giorgetta, M., Hagemann,

Shepherd, T. G. & Jonsson, A. I. (2008). On the attribution of stratospheric ozone and

Shibata, K., Deushi, M., Sekiyama, T. T., & Yoshimura, H. (2005). Development of an MRI

Shibata, K. & Deushi, M. (2005a). Partitioning between resolved wave forcing and

Shibata, K. & Deushi, M. (2005b). Radiative effect of ozone on the quasi-biennial oscillation

Shibata, K. & Deushi, M. (2008a). Long-term variations and trends in the simulation of the

Meteorological Research Institute. *Annales Geophysicae*, 26, 1299-1326. Shibata, K. & Deushi, M. (2008b). *Simulation of the stratospheric circulation and ozone during the* 

greenhouse gases. *Atmos. Chem. Phys.,* 8, 1435-1444.

*Meteorology and Geophysics*, 55, 75-119.

*Model description.* MPI-Rep. 349, Max-Planck-Inst., Hamburg, Germany. Scaife, A. A., N. Butchart, C. D. Warner, D. Stainforth, W. A. Norton & J. Austin (2000).

temperature and ozone in the tropical lower stratosphere. *Geophys. Res. Lett.*, 36,

downward propagating annual wind reversal in the equatorial stratosphere. *J.* 

M. P., Naujokat, B., Reimer, E., Allaart, M., Andersen, S. B., Bevilacqua, R., Braathen, G. O., Claude, H., Davies, J., De Backer, H., Dier, H., Dorokhov, V., Fast, H., Gerding, M., Godin-Beekmann, S., Hoppel, K., Johnson, B., Kyrö, E., Litynska, Z., Moore, D., Nakane, H., Parrondo, M. C., Risley Jr., A. D., Skrivankova, P., Stübi, R., Viatte, P., Yushkov, V. & Zerefos, C. (2006). Arctic winter 2005: Implications for stratospheric ozone loss and climate change. *Geophys. Res. Lett.*, 33, L23808,

S., Kirchner, I., Kornblueh, L., Manzini, E., Rhodin, A., Schlese, U., Schulzweida, U. & Tompkins, A. (2003). *The atmospheric general circulation model ECHAM 5. part I:* 

Realistic quasi-biennial oscillations in a simulation of the global climate. *Geophys.* 

temperature changes to changes in ozone-depleting substances and well-mixed

chemical transport model for the study of stratospheric chemistry. *Papers in* 

unresolved gravity wave forcing to the quasi-biennial oscillation as revealed with a coupled chemistry-climate model. *Geophys. Res. Lett.*, 32, L12820, doi:10.1029/

in the equatorial stratosphere. *Geophys. Res. Lett.*, 32, L24802, doi:10.1029/

middle atmosphere 1980-2004 by the chemistry-climate model of the

*recent past (1980-2004) with the MRI chemistry-climate model.* CGER's Supercomputer Monograph Report, Vol.13, National Institute for Environmental Studies, Japan. SPARC, V. Eyring, T. G. Shepherd, D. W. Waugh (Eds.) (2010). *SPARC CCMVal Report on the* 

*Evaluation of Chemistry-Climate Models*, SPARC Report No.5, WCRP-132, WMO/TD-


Kodama, C., Iwasaki, T., Shibata, K. & Yukimoto, S. (2007). Changes in the stratospheric mean

Labitzke, K. (1982). On the interannual variability of the middle stratosphere during the

Labitzke, K. & van Loon, H. (1999). *The Stratosphere: Phenomena, History, and Relevence.*

Labitzke, K., Kunze, M. & Brönnimann, S. (2006). Sunspots, the QBO, and the stratosphere in

Lee, H. & Smith, A. K. (2003). Simulation of the combined effects of solar cycle, quasi-

Lindzen, R. S. & Holton, J. R. (1968). A theory of the quasibiennial oscillation. *J. Atmos. Sci.*,

Lu, H., Baldwin, M. P., Gray, L. J. & Jarvis, M. J. (2008). Decadal-scale changes in the effect of

Lu, H., Gray, L. J., Baldwin, M. P., & Jarvis, M. J. (2009). Life cycle of the QBO-modulated 11-

Luo, M., Russell III, J. M. & Huang, T. Y. W. (1997). Halogen Occultation Experiment

Marshall, A. G. & Scaife, A. A. (2009). Impact of the QBO on surface winter climate. *J.* 

Naito, Y. & Hirota, I. (1997). Interannual variability of the northern winter stratospheric circulation related to the QBO and the solar cycle. *J. Meteorol. Soc. Japan*, 75, 925–937. Naoe, H. & Shibata, K. (2010). Equatorial quasi-biennial oscillation influence on northern winter extratropical circulation, *J. Geophys. Res.*, 115, doi:10.1029/2009JD012952. Naoe, H. & Shibata, K. (2012). Future changes in the influence of the quasi-biennial

Niwano, M. & Takahashi, M. (1998). The influence of the equatorial QBO on the Northern Hemisphere winter circulation of a GCM. *J. Meteor. Soc. Japan,* 76, 453–461. Pascoe C. L., Gray, L. J., Crooks, S. A., Juckes, M. N. & Baldwin, M. P. (2005). The quasi-

Philander, S. G. (1990). *El Niño, La Niña and the Southern Oscillation*, Academic Press, New

Plumb, R. A. (1977). The interaction of two internal waves with the mean flow: Implications for the theory of the quasi-biennial oscillation. *J. Atmos. Sci.*, 34, 1847–1858. Plumb, R. A. & McEwan, D. (1978). The instability of a forced standing wave in a viscous

induced effects. *J. Geophys. Res.*, 112, D16103, doi:10.1029 /2006JD008219. Kuroda, Y. & Yamazaki, K. (2010). Influence of the solar cycle and QBO modulation on the

northern winters. *J. Meteorol. Soc. Japan*, 60, 124–139.

the north polar region––20 years later. *Meteor. Z.,* 3, 355–363.

decades. *J. Geophys. Res.*, 108, 4049, doi:10.1029/2001JD001503.

the tropical stratosphere. *J. Geophys. Res.,* 102, 19187–19198.

model. *J. Geophys. Res.*, D03102, doi:10.1029/2011JD016255, 2012.

*Geophys. Res.*, 114, D18110, doi:10.1029/2009JD011737.

Springer-Verlag.

25, 1095–1107.

doi:10.1029/2007JD009647.

doi:10.1029/2004JD004941.

York, 293pp.

35, 1827–1839.

135, 1030-1043. DOI: 10.1002/qj.419.

meridional circulation due to increased CO2: Radiation- and sea surface temperature-

Southern Annular Mode. *Geophys. Res. Lett.*, 37, L12703, doi:10.1029/2010GL043252.

biennial oscillation, and volcanic forcing on stratospheric ozone changes in recent

the QBO on the northern stratospheric polar vortex. *J. Geophys. Res.*, 113, D10114,

year solar cycle signals in the Northern Hemispheric winter. *Q. J. R. Meteorol. Soc.*,

observations of the quasi-biennial oscillation and the effects of Pinatubo aerosols in

oscillation on the northern polar vortex simulated with an MRI chemistry climate

biennial oscillation: Analysis using ERA-40 data. *J. Geophys. Res.*, 110, D08105,

stratified fluid: A laboratory analogue of the quasi-biennial oscillation. *J. Atmos. Sci.,* 


**18**

*Mexico* 

Luis Brito Castillo

*Centro de Investigaciones Biológicas del Noroeste* 

**Regional Pattern of Trends in Long-Term** 

**Precipitation and Stream Flow Observations:** 

**Singularities in a Changing Climate in Mexico** 

The Intergovernmental Panel on Climate change has concluded that global warming is unequivocal (IPCC, 2007). However, there is still a debate regarding whether global warming is mainly due to anthropogenic greenhouse gas emissions (Zhao, 2011), particularly because, as Lindzen (2007) states, current models fail to describe many known climate changes, and because some sources of anthropogenic forcing contribution (like surface albedo, total aerosol, etc) are essentially unknown (Lindzen, 2007). It has been stated that the warming has penetrated into the world's oceans (Barnett et al., 2005) increasing sea surface temperatures (SSTs) by approximately 0.5°C since the mid-nineteenth century (Houghton et al. 2001). Higher SSTs are associated with increased water vapor in the lower troposphere (Trenberth, 2005) implying that perturbations in the global water cycle are expected to accompany this warming (Milly et al., 2005). The understanding of specific regional changes (i.e. trend analysis) in temperature, rainfall, and streamflow is very necessary for supporting the needs of an ever-broadening spectrum of decision-makers as they strive to deal with influences of Earth's climate at global to local scales. Trend analyses have revealed the existence of temperature trends with different signs in Mexico in general (Englehart & Douglas, 2004; Pavia et al., 2009), central Mexico (Brito-Castillo *et al*., 2009), and northwestern Mexico (Gutierrez-Ruacho et al., 2010; Weiss & Overpeck, 2005), which is partly associated with regional land use, changes in land cover (Englehart & Douglas, 2004), and changes in large-scale atmospheric flow patterns (Brito-Castillo *et al*., 2009). These studies have shown that, while significant trends do exist, they are at times in line with warming hypotheses and at times they are not. The complete understanding of these changes is particularly important, especially for river discharges, since availability of water is vital for human health, economic activity, ecosystem functions, and geophysical processes (Milly et al., 2005). In Mexico, trend analyses have mostly focused on maximum and minimum temperatures, while trend analyses of precipitation and streamflow have only partly been studied. Knowledge remains scarce, partly attributed to lack of long-term data (both in space and time), the difficulties encountered in providing a correct interpretation of changes attributed to human disturbances, such as diversion of water for irrigation of cropland and control of water for generating electric power, as well as the complexity of the processes that externally force hydroclimate variability (natural changes and low-frequency

**1. Introduction** 


## **Regional Pattern of Trends in Long-Term Precipitation and Stream Flow Observations: Singularities in a Changing Climate in Mexico**

Luis Brito Castillo *Centro de Investigaciones Biológicas del Noroeste Mexico* 

#### **1. Introduction**

386 Greenhouse Gases – Emission, Measurement and Management

Son, S. -W., Polvani, L. M., Waugh, D. W., Akiyoshi, H., Garcia, R., Kinnison, D., Pawson, S.,

Taguchi, M. (2010). Observed connection of the stratospheric quasi-biennial oscillation with

Takahashi, M. (1999). Simulation of the stratospheric quasi-biennial oscillation in a general

Torrence, C. & Compo, G. P. (1998). A Practical Guide to Wavelet Analysis. *Bull. Amer.* 

Uppala, S. M., Kållberg, P. W., Simmons, A. J., Andrae, U., da Costa Bechtold, V., Fiorino,

Veryard, R. G. & Ebdon, R. A. (1961). Fluctuations in tropical stratospheric winds. *Meteorol.* 

Wallace, J. M., Panetta, R. L. & Estberg, J. (1993). Representation of the equatorial stratospheric quasi-biennial oscillation in EOF phase space. *J. Atmos. Sci.*, 50, 1751–1762. Watanabe, S., Kawatani, Y., Tomikawa, Y., Miyazaki, K., Takahashi, M. & Sato, K. (2008).

WMO (2007). *Scientific Assessment of Ozone Depletion: 2006*, World Meteorological

WMO (2011). *Scientific Assessment of Ozone Depletion: 2010,* World Meteorological

Xiao, F. and Peng, X. (2004). A convexity preserving scheme for conservative advection transport. *J. Comput. Phys*., 198, 389–402, doi:10.1016/j.jcp.2004.01.013. Yamashita, Y., Akiyoshi, H. & Takahashi, M. (2011). Dynamical response in the Northern

CCSR/NIES CCM. *J. Geophys. Res., 116*, D06118, doi:10.1029/2010JD015016. Yukimoto, S., Noda, A., Kitoh, A., Hosaka, M., Yoshimura, H., Uchiyama, T., Shibata, K.,

General aspects of a T213L256 middle atmosphere general circulation model. *J.* 

Organization, Global Ozone Research and Monitoring Project, Report No. 50,

Organization, Global Ozone Research and Monitoring Project, Report No. 52,

Hemisphere midlatitude and high-latitude winter to the QBO simulated by

Arakawa, O. & Kusunoki, S. (2006). Present-day climate and climate sensitivity in the Meteorological Research Institute coupled GCM version 2.3 (MRI-CGCM2.3). *J.* 

circulation model. *Geophys. Res. Lett.*, 26, 1307–1310.

*Meteor. Soc.,* 131, 2961–3012, doi:10.1256/qj.04.176.

*Geophys. Res.*, 113, D12110, doi:10.1029/2008JD010026.

doi:10.1126/science.1155939.

doi:10.1029/2010JD014325.

*Meteor. Soc.*, 79, 61-78.

*Mag.*, 90, 125–143.

Geneva, Switzerland.

Geneva, Switzerland.

*Meteorol. Soc. Japan*, 84, 333-363.

Rozanov, E., Shepherd, T. G., & Shibata, K. (2008). The impact of stratospheric ozone recovery on the southern hemisphere westerly jet. *Science*, 320, 1486-1489,

El Niño–Southern Oscillation in radiosonde data. *J. Geophys. Res.*, 115, D18120,

M., Gibson, J. K., Haseler, J., Hernandez, A., Kelly, G. A., Li, X., Onogi, K., Saarinen, S., Sokka, N., Allan, R. P., Andersson, E., Arpe, K., Balmaseda, M. A., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann, N., Caires, S., Chevallier, F., Dethof, A., Dragosavac, M., Fisher, M., Fuentes, M., Hagemann, S., Holm, E., Hoskins, B. J., Isaksen, L., Janssen, P. A. E. M., Jenne, R., McNally, A. P., Mahfouf, J.-F., Morcrette, J.-J., Rayner, N. A., Saunders, R. W., Simon, P., Ster, A., Trenberth, K. E., Untch, A., Vasiljevic, D., Viterbo, P., & Woollen, J. (2005). The ERA-40 Reanalysis. *Q. J. Roy.* 

The Intergovernmental Panel on Climate change has concluded that global warming is unequivocal (IPCC, 2007). However, there is still a debate regarding whether global warming is mainly due to anthropogenic greenhouse gas emissions (Zhao, 2011), particularly because, as Lindzen (2007) states, current models fail to describe many known climate changes, and because some sources of anthropogenic forcing contribution (like surface albedo, total aerosol, etc) are essentially unknown (Lindzen, 2007). It has been stated that the warming has penetrated into the world's oceans (Barnett et al., 2005) increasing sea surface temperatures (SSTs) by approximately 0.5°C since the mid-nineteenth century (Houghton et al. 2001). Higher SSTs are associated with increased water vapor in the lower troposphere (Trenberth, 2005) implying that perturbations in the global water cycle are expected to accompany this warming (Milly et al., 2005). The understanding of specific regional changes (i.e. trend analysis) in temperature, rainfall, and streamflow is very necessary for supporting the needs of an ever-broadening spectrum of decision-makers as they strive to deal with influences of Earth's climate at global to local scales. Trend analyses have revealed the existence of temperature trends with different signs in Mexico in general (Englehart & Douglas, 2004; Pavia et al., 2009), central Mexico (Brito-Castillo *et al*., 2009), and northwestern Mexico (Gutierrez-Ruacho et al., 2010; Weiss & Overpeck, 2005), which is partly associated with regional land use, changes in land cover (Englehart & Douglas, 2004), and changes in large-scale atmospheric flow patterns (Brito-Castillo *et al*., 2009). These studies have shown that, while significant trends do exist, they are at times in line with warming hypotheses and at times they are not. The complete understanding of these changes is particularly important, especially for river discharges, since availability of water is vital for human health, economic activity, ecosystem functions, and geophysical processes (Milly et al., 2005). In Mexico, trend analyses have mostly focused on maximum and minimum temperatures, while trend analyses of precipitation and streamflow have only partly been studied. Knowledge remains scarce, partly attributed to lack of long-term data (both in space and time), the difficulties encountered in providing a correct interpretation of changes attributed to human disturbances, such as diversion of water for irrigation of cropland and control of water for generating electric power, as well as the complexity of the processes that externally force hydroclimate variability (natural changes and low-frequency

Regional Pattern of Trends in Long-Term Precipitation and




b)

c)

standardized values (i.e., z scores; see text for definitions).

z scores

z scores

analysis:

z scores

describe the overall change over time in a dataset (Wigley, 2006).

Stream Flow Observations: Singularities in a Changing Climate in Mexico 389

series is superimposed on multi-decadal variability with substantial deviations from linearity (i.e., the trend) in some years (1956 and 2000 in Figure 1b and in 1959 and 2004 in Figure 1c.). In these situations a detailed inspection of the series can help show more complex changes that are evident in data; still, linear trends provide the simplest and most convenient way to

1940 1950 1960 1970 1980 1990 2000 2010

<sup>1956</sup> <sup>2000</sup>

TDP>0

Precipitation

Streamflow

1940 1950 1960 1970 1980 1990 2000 2010

Fig. 1. Long-term fluctuations of different data sets: (a) Total annual days with precipitation >0 mm (TDP >0); (b) total annual precipitation (mm) and (c) mean annual streamflow (m³/s). a-b are from Ocozonautla, Chiapas weather station (Code: 07123; –93.374°W; 16.751°N; 838 masl); c is from La Y-Rio Lerma gauging station (Code: 12374; –99.5894°W; 19.40611°N; basin area 1582 km²). The line is the trend and the ± values define the 95% confidence intervals for the trends. For comparison all variables are expressed in

In this work, linear trends were quantified as the change of TDP > 0, precipitation, and streamflow per year by fitting a straight line to the data using the method of minimal least squares errors, with years (T) as the independent values and serial data (Y) within each year as the dependent values. The "best-fit" straight line to data is obtained by linear regression

Yest = b + T (2)

where ∑ represents the summation over i = 1…n; the subscripts "est" and "avg" are the estimate of Y (TDP > 0, precipitation or streamflow) by the fitted regression line and the

Estimates of the linear trend are sensitive to points at the start or end of the dataset. This is more a problem with small sample sizes (for trends over short time periods). For example, if

mean value, respectively, and b is the slope of the fitted line (the trend value).

b = [∑{(Ti - Tavg) (Yi - Yavg)}] / [∑{(Ti - Tavg)2}], (3)

<sup>1959</sup> <sup>2004</sup>

Trend: 4.73 ± 3.77 mm/yr

Trend: 0.03 ± 0.02 (m³/s)/yr

Trend: 0.64 ± 0.24 days/yr

internal variability, such as Pacific decadal oscillation (PDO) and Atlantic multidecadal oscillation (AMO). Given the limited knowledge of streamflow and precipitation trend structures in Mexico, in this chapter, we analyzed overall trends in Mexico and described the regional structures of the trends for the period 1920–2008. Evidence of real trends over different regions of Mexico is provided, sometimes revealing that oscillations of long-term high and low humid periods (depending on their arrangement) force the fitted linear trend model. This implies that declining or increasing trends are restrained by the length of the records. However, there are also cases of unambiguous multi-decadal trends in rainfall and streamflow time series that require more detailed analysis. Section 2 explains the statistical approaches used in the study of trends, with some real examples from precipitation and streamflow variables in Mexico. In Section 3, we describe seasonal precipitation characteristics over Mexico to illustrate the main climate differences across the country. Section 4 provides an introduction to the most recent findings related to global warming and changes in tropical cyclones, including a brief description of tropical cyclone contribution to rainfall in Mexico. In the remaining sections, we explain the main procedures for detecting trends used in this work. In Section 5 we describe climate data and quality control; in Section 6, we show a regionalization of Mexico by simply fractioning the territory into several domains with specific climatic conditions. Finally, in Sections 7 and 8, we discuss the main regional structures of trends detected in precipitation variables and their correspondence to trends in streamflow datasets. We hope that this chapter provides a useful introduction to problems related to trends in precipitation and streamflow variables in Mexico.

#### **2. The study of trends**

If data show continuous increase or decline change over time, we refer to this change as a trend (Figure 1). Linear trends are an example of data "statistics" like the average (or mean) and "standard deviation values", which is a measure of the variability within a data set around the mean value. Sometimes, to allow comparisons between the series having different scales of variability results, it is necessary to change the data scores to standardized values or z scores, which is the number of standard deviations away from the mean:

$$\mathbf{z}\_{i} = (\mathbf{x}\_{i} - \boldsymbol{\mu}) \;/\; \mathbf{o} \tag{1}$$

where: zi, is the standard score; xi is the observed score; µ is the mean value; σ is the standard deviation; i = 1…n; and n is the sample size.

When all the scores from a variable are transformed into standard scores (z scores), a new variable is obtained with a mean of zero and standard deviation of one.

Fitting a linear trend model to data is a common practice; however linear trends may be deceptive, particularly when datasets show non-linear changes over time (Wigley, 2006), which is a property of precipitation and streamflow time series. An example of this situation is displayed in Figure 1 for three different series expressed as z scores: (1) Total annual days with precipitation >0 mm (i.e. annual TDP > 0) (Figure 1a); (2) Total annual precipitation (mm) (Figure 1b & Figure 2) and (3) mean annual streamflow (m³/s) (Figure 1c). All variables show an initial declining trend (though incomplete), followed by a large upward step after 1958 that results in an overall increasing trend. It is evident that inter-annual variability of the time

internal variability, such as Pacific decadal oscillation (PDO) and Atlantic multidecadal oscillation (AMO). Given the limited knowledge of streamflow and precipitation trend structures in Mexico, in this chapter, we analyzed overall trends in Mexico and described the regional structures of the trends for the period 1920–2008. Evidence of real trends over different regions of Mexico is provided, sometimes revealing that oscillations of long-term high and low humid periods (depending on their arrangement) force the fitted linear trend model. This implies that declining or increasing trends are restrained by the length of the records. However, there are also cases of unambiguous multi-decadal trends in rainfall and streamflow time series that require more detailed analysis. Section 2 explains the statistical approaches used in the study of trends, with some real examples from precipitation and streamflow variables in Mexico. In Section 3, we describe seasonal precipitation characteristics over Mexico to illustrate the main climate differences across the country. Section 4 provides an introduction to the most recent findings related to global warming and changes in tropical cyclones, including a brief description of tropical cyclone contribution to rainfall in Mexico. In the remaining sections, we explain the main procedures for detecting trends used in this work. In Section 5 we describe climate data and quality control; in Section 6, we show a regionalization of Mexico by simply fractioning the territory into several domains with specific climatic conditions. Finally, in Sections 7 and 8, we discuss the main regional structures of trends detected in precipitation variables and their correspondence to trends in streamflow datasets. We hope that this chapter provides a useful introduction to problems related to trends in precipitation and streamflow variables

If data show continuous increase or decline change over time, we refer to this change as a trend (Figure 1). Linear trends are an example of data "statistics" like the average (or mean) and "standard deviation values", which is a measure of the variability within a data set around the mean value. Sometimes, to allow comparisons between the series having different scales of variability results, it is necessary to change the data scores to standardized values or z scores, which is the number of standard deviations away from

 zi = (xi – µ) / σ (1) where: zi, is the standard score; xi is the observed score; µ is the mean value; σ is the

When all the scores from a variable are transformed into standard scores (z scores), a new

Fitting a linear trend model to data is a common practice; however linear trends may be deceptive, particularly when datasets show non-linear changes over time (Wigley, 2006), which is a property of precipitation and streamflow time series. An example of this situation is displayed in Figure 1 for three different series expressed as z scores: (1) Total annual days with precipitation >0 mm (i.e. annual TDP > 0) (Figure 1a); (2) Total annual precipitation (mm) (Figure 1b & Figure 2) and (3) mean annual streamflow (m³/s) (Figure 1c). All variables show an initial declining trend (though incomplete), followed by a large upward step after 1958 that results in an overall increasing trend. It is evident that inter-annual variability of the time

in Mexico.

the mean:

**2. The study of trends** 

standard deviation; i = 1…n; and n is the sample size.

variable is obtained with a mean of zero and standard deviation of one.

series is superimposed on multi-decadal variability with substantial deviations from linearity (i.e., the trend) in some years (1956 and 2000 in Figure 1b and in 1959 and 2004 in Figure 1c.). In these situations a detailed inspection of the series can help show more complex changes that are evident in data; still, linear trends provide the simplest and most convenient way to describe the overall change over time in a dataset (Wigley, 2006).

Fig. 1. Long-term fluctuations of different data sets: (a) Total annual days with precipitation >0 mm (TDP >0); (b) total annual precipitation (mm) and (c) mean annual streamflow (m³/s). a-b are from Ocozonautla, Chiapas weather station (Code: 07123; –93.374°W; 16.751°N; 838 masl); c is from La Y-Rio Lerma gauging station (Code: 12374; –99.5894°W; 19.40611°N; basin area 1582 km²). The line is the trend and the ± values define the 95% confidence intervals for the trends. For comparison all variables are expressed in standardized values (i.e., z scores; see text for definitions).

In this work, linear trends were quantified as the change of TDP > 0, precipitation, and streamflow per year by fitting a straight line to the data using the method of minimal least squares errors, with years (T) as the independent values and serial data (Y) within each year as the dependent values. The "best-fit" straight line to data is obtained by linear regression analysis:

$$\mathbf{Y}\_{\text{est}} = \mathbf{b} + \mathbf{T} \tag{2}$$

$$\mathbf{b} = \left[ \sum \{ (\mathbf{T}\_{\text{i-}} \cdot \mathbf{T}\_{\text{avg}}) \left( \mathbf{Y}\_{\text{i-}} \cdot \mathbf{Y}\_{\text{avg}} \right) \} \right] \Big/ \left[ \sum \{ (\mathbf{T}\_{\text{i-}} \cdot \mathbf{T}\_{\text{avg}}) ^2 \} \right] \tag{3}$$

where ∑ represents the summation over i = 1…n; the subscripts "est" and "avg" are the estimate of Y (TDP > 0, precipitation or streamflow) by the fitted regression line and the mean value, respectively, and b is the slope of the fitted line (the trend value).

Estimates of the linear trend are sensitive to points at the start or end of the dataset. This is more a problem with small sample sizes (for trends over short time periods). For example, if

Regional Pattern of Trends in Long-Term Precipitation and

upward or downward trend (Wigley, 2006).

about the fitted line of the observed data.

**3. Rainfall distribution over Mexico** 

Stream Flow Observations: Singularities in a Changing Climate in Mexico 391

To decide if upward or downward trends are an indication of some underlying cause or merely a chance fluctuation, in this work, we use two approaches: (1) confidence intervals: b ± 2 SE (often called the two-sigma confidence interval) and (2) significance testing. The range b ± 2 SE is a good approximation to represent the 95% confidence interval and is used as a signed error magnitude. If –2SE < b < 2SE, it means that the 95% confidence interval includes the zero trend value and the null hypothesis that "there is no trend in data sets or b = 0" is true; otherwise (i.e. b > 2SE or b < –2SE), we say that the null hypothesis is unlikely to be true and we must accept the alternate hypothesis that "data show a real, externally forced

The second approach consists in trying to decide whether an observed data trend that is noticeably different from zero is sufficiently different that it could not have occurred by chance or that the probability (known as p-level) that it could have occurred by chance is very small. If the null hypothesis is rejected, then we say that the observed trend is "statistically significant" at some level of confidence. In this study, the confidence level we used to reject the null hypothesis is alpha level = 0.05. If the probability that the null hypothesis is true is small (i.e. p-level < 0.05), then the null hypothesis is unlikely to be correct. Such a low probability would mean that the observed trend is highly unusual and therefore a "significant result". The significance test used here is a "two-tailed" test, meaning that we are concerned only with whether the trend is different from zero, with no specification of whether

The significance of a trend and its confidence intervals depend on the standard error of the trend estimate. Equation (4) given above is, however, only correct if the individual scores are unrelated. The dependence between scores is referred to as "temporal autocorrelation" or "serial correlation". When data are auto-correlated, many statistics behave as if the sample size was less than the number of data points, n, so it is necessary to determine an

 neff = n(1–r1) / (1 + r1) (6) where r1 is the lag-1 autocorrelation coefficient calculated from the time series of residuals

If r1 is statistically significant (i.e. p-level of r1 is < 0.05), then in Equation (5), neff is used instead of n. It can be seen from Equation (5) that if neff is noticeably smaller than n, then the standard error of the trend estimate may be much larger than one would otherwise expect. This means that results that may show a significant trend if autocorrelation is ignored; it is

The climate of Mexico is influenced by the strength and position of the subtropical high pressure systems of the Northeast Pacific (the East Pacific High) and the North Atlantic (the Bermuda High) (Wang et al., 2007), and the variations in the location of the Intertropical Convergence Zone (ITCZ), which lies south of Mexico. The formation of the high pressure

the trend should be increasing (positive) or declining (negative) trend.

"effective sample size" (neff). In this study, neff was estimated by Equation (6):

frequently found to be non-significant when autocorrelation is accounted for.

where MSE is the mean square error and SE is the standard error of the trend estimate.

MSE = [∑((Y–Yest)2)] / (n – 2) (5)

we considered streamflow data points in Figure 1c from 1959 through 1983 (25 years), because there was an unusual increase in 1959 (which is also recorded at other gauging stations downstream along the river), the calculated trend may be an overestimate of the true underlying trend.

Other uncertainties that can be found in trend analysis is related to the presence of gaps in datasets and the apparently multi-decadal variability that seems to be "incomplete" in some series (Figure 2), giving the impression that the trends might not be real. In both cases shown in Figure 2, there are clear trends, upward trend in Figure 2a and downward trend in Figure 2b. However, after the 1970s, the overall trend in Figure 2a assumes a decreasing trend. Unfortunately, the dataset finished very early at the middle of the 1990s without enough data to prove if this second trend is statistically significant (see below for definition). In Figure 2b, on the contrary, there is an overall decreasing trend, but before 1970, it seems to be an increasing trend or no trend at all. Unfortunately, the dataset beginning in 1950 cannot prove this. Around the 1970s, the scores are greater than the mean value in both datasets (Figures 2a & b). Comparing the overall behavior between both data sets, it seems as though both datasets were complementary to each other, indicating the possibility that, if observations were available for both time series over the 1920-2010 period, it might be possible to calculate no trend at all. So far, we can say that, with this information, increasing (Figure 2a) or declining (Figure 2b) trends might be restrained by the length of the records.

For examples, given in Figure 2, results necessary to explore the behavior of surrounded stations is more certain.

Fig. 2. The same as Figure 1, but for total annual precipitation (mm). (a) Matías Romero (Code: 20068; –95.03°W; 16.88°N; 201 masl); (b) Paso de Arocha, Nayarit Code: 18025; – 105.10°W; 21.83°N; 30 masl). Heavy lines are 9-yr running means.

A measure of how well the straight line fits the data is the average value of the squares of the residuals (i.e. deviations about the fitted line) or the standard deviation of the trend estimate, or its squares, the variance of the distribution of b (Var(b)). The smaller this is, the better is the fit:

$$\text{Var}(\mathbf{b}) = (\mathbf{SE})^2 = \text{MSE} \ / \ [\sum [(\mathbf{T} - \mathbf{T}\_{\text{avg}})^2]] \tag{4}$$

we considered streamflow data points in Figure 1c from 1959 through 1983 (25 years), because there was an unusual increase in 1959 (which is also recorded at other gauging stations downstream along the river), the calculated trend may be an overestimate of the

Other uncertainties that can be found in trend analysis is related to the presence of gaps in datasets and the apparently multi-decadal variability that seems to be "incomplete" in some series (Figure 2), giving the impression that the trends might not be real. In both cases shown in Figure 2, there are clear trends, upward trend in Figure 2a and downward trend in Figure 2b. However, after the 1970s, the overall trend in Figure 2a assumes a decreasing trend. Unfortunately, the dataset finished very early at the middle of the 1990s without enough data to prove if this second trend is statistically significant (see below for definition). In Figure 2b, on the contrary, there is an overall decreasing trend, but before 1970, it seems to be an increasing trend or no trend at all. Unfortunately, the dataset beginning in 1950 cannot prove this. Around the 1970s, the scores are greater than the mean value in both datasets (Figures 2a & b). Comparing the overall behavior between both data sets, it seems as though both datasets were complementary to each other, indicating the possibility that, if observations were available for both time series over the 1920-2010 period, it might be possible to calculate no trend at all. So far, we can say that, with this information, increasing (Figure 2a) or declining (Figure 2b) trends might be restrained by the length of the records. For examples, given in Figure 2, results necessary to explore the behavior of surrounded

1920 1930 1940 1950 1960 1970 1980 1990 2000 2010

A measure of how well the straight line fits the data is the average value of the squares of the residuals (i.e. deviations about the fitted line) or the standard deviation of the trend estimate, or its squares, the variance of the distribution of b (Var(b)). The smaller this is, the

Var(b) = (SE)2 = MSE / [∑{(T – Tavg)2}] (4)

Fig. 2. The same as Figure 1, but for total annual precipitation (mm). (a) Matías Romero (Code: 20068; –95.03°W; 16.88°N; 201 masl); (b) Paso de Arocha, Nayarit Code: 18025; –

1920 1930 1940 1950 1960 1970 1980 1990 2000 2010


Trend: 11.2 ± 0.81 mm/yr

0

z scores

2

true underlying trend.

stations is more certain.

a)

b)

Trend: -6.41 ± 4.42 mm/yr

105.10°W; 21.83°N; 30 masl). Heavy lines are 9-yr running means.


better is the fit:

0

z scores

2

$$\text{MSE} = \left[ \sum ( (\text{Y-Y}\_{\text{est}})^2 ) \right] / \text{ (n-2)} \tag{5}$$

where MSE is the mean square error and SE is the standard error of the trend estimate.

To decide if upward or downward trends are an indication of some underlying cause or merely a chance fluctuation, in this work, we use two approaches: (1) confidence intervals: b ± 2 SE (often called the two-sigma confidence interval) and (2) significance testing. The range b ± 2 SE is a good approximation to represent the 95% confidence interval and is used as a signed error magnitude. If –2SE < b < 2SE, it means that the 95% confidence interval includes the zero trend value and the null hypothesis that "there is no trend in data sets or b = 0" is true; otherwise (i.e. b > 2SE or b < –2SE), we say that the null hypothesis is unlikely to be true and we must accept the alternate hypothesis that "data show a real, externally forced upward or downward trend (Wigley, 2006).

The second approach consists in trying to decide whether an observed data trend that is noticeably different from zero is sufficiently different that it could not have occurred by chance or that the probability (known as p-level) that it could have occurred by chance is very small. If the null hypothesis is rejected, then we say that the observed trend is "statistically significant" at some level of confidence. In this study, the confidence level we used to reject the null hypothesis is alpha level = 0.05. If the probability that the null hypothesis is true is small (i.e. p-level < 0.05), then the null hypothesis is unlikely to be correct. Such a low probability would mean that the observed trend is highly unusual and therefore a "significant result". The significance test used here is a "two-tailed" test, meaning that we are concerned only with whether the trend is different from zero, with no specification of whether the trend should be increasing (positive) or declining (negative) trend.

The significance of a trend and its confidence intervals depend on the standard error of the trend estimate. Equation (4) given above is, however, only correct if the individual scores are unrelated. The dependence between scores is referred to as "temporal autocorrelation" or "serial correlation". When data are auto-correlated, many statistics behave as if the sample size was less than the number of data points, n, so it is necessary to determine an "effective sample size" (neff). In this study, neff was estimated by Equation (6):

$$\mathbf{n}\_{\rm eff} \equiv \mathbf{n} (\mathbf{1} - \mathbf{r}\_1) \;/\; \; (\mathbf{1} + \mathbf{r}\_1) \tag{6}$$

where r1 is the lag-1 autocorrelation coefficient calculated from the time series of residuals about the fitted line of the observed data.

If r1 is statistically significant (i.e. p-level of r1 is < 0.05), then in Equation (5), neff is used instead of n. It can be seen from Equation (5) that if neff is noticeably smaller than n, then the standard error of the trend estimate may be much larger than one would otherwise expect. This means that results that may show a significant trend if autocorrelation is ignored; it is frequently found to be non-significant when autocorrelation is accounted for.

#### **3. Rainfall distribution over Mexico**

The climate of Mexico is influenced by the strength and position of the subtropical high pressure systems of the Northeast Pacific (the East Pacific High) and the North Atlantic (the Bermuda High) (Wang et al., 2007), and the variations in the location of the Intertropical Convergence Zone (ITCZ), which lies south of Mexico. The formation of the high pressure

Regional Pattern of Trends in Long-Term Precipitation and

southwest of Mexico over the ITCZ (Giovannettone & Barros, 2007).

Stream Flow Observations: Singularities in a Changing Climate in Mexico 393

propagating northward as the convective activity and easterlies strengthened in May-June. In southern and central Mexico, precipitation peaks in June and September with a decline in convective activity in late July-early August, causing a break period known as canicula (midsummer drought) (Magaña et al., 1999), as a result of the presence of an anticyclone

Fig. 3. Distribution of long-term mean values of daily precipitation per month. The mean value is calculated for individual stations with larger than 40 yrs of records over the 1948– 2000 period. In lower left map, GC = Gulf of California; Sin = Sinaloa; Mich = Michoacan; SMO = Sierra Madre Occidental; SMS = Sierra Madre del Sur, GM = Gulf of Mexico,

Poleward extension of the summer precipitation regime in northwest Mexico starts with an abrupt onset of precipitation in late June or early July and quickly spreads northward along the western slopes of the Sierra Madre Occidental. Precipitation increases coincide with increased vertical transport of moisture by convection and southerly winds through the Gulf of California. During July through early September, the monsoon system is fully developed

YP = Yucatan Peninsula.

systems is related to global scale circulations that function to move excess energy from the equatorial regions to poles affecting precipitation variability across Mexico on seasonal to decadal time scales. An unusually strong East Pacific High during the winter can steer storm systems far to the north, limiting cloud formation and precipitation with a drying effect over the river basins in much of northern Mexico (Brito-Castillo et al., 2003), while an unusually strong Bermuda High during the summer can enhance monsoon circulation and numerous transient disturbances (e.g. tropical storms, tropical easterly waves, inverted troughs, the Madden Julian Oscillation, and mid-latitude troughs) and bring above-average precipitation across much of Mexico (Douglas et al., 2007; Wang et al., 2007). Subtle changes to the position and strength of these circulation features appear to be connected to global scale shifts in circulation patterns related to changes in sea surface temperature patterns (e.g., El Niño-Southern Oscillation) affecting rainfall variability of Mexico at inter-annual timescales and can persist for years (e.g., the Pacific Decadal Oscillation, PDO), bringing extended periods of unusually wet or dry conditions at decadal timescales (Magaña et al., 2003; Pavia et al., 2006).

### **3.1 Seasonality of precipitation**

Mexico receives most of its annual rainfall during the summer, with a secondary peak occurring during the winter. From November through March (Figure 3 left maps), winter precipitation is associated with large-scale low pressure systems that traverse northern Mexico, drawing in moisture from the Pacific Ocean and producing widespread rain and snow at higher elevations of the Sierra Madre Occidental in northwestern Mexico, the northwestern portion of the Baja California Peninsula, the coastal plains of the Gulf of Mexico (CPGM), the Yucatan Peninsula (YP), and some portions of central Mexico (mainly in December and January), while the rest of the country generally experiences dry conditions. The effects of cold fronts coming from the northwest also cause lower temperatures in most of Mexico, but are more evident in the highlands of north and central Mexico, and on the CPGM, and the YP, where they are associated with strong northerly winds (nortes; outbreaks of cold weather), occasionally producing intense precipitation in the southeast, when the cold air masses are modified by the warm waters from the Gulf of Mexico. The large-scale forcing (1000 km-global) of regional climate variability, such as the location and strength of the subtropical westerly jet, determines the succession of weather events during winter in eastern Mexico. A change from zonal to more frequent meridional circulation has been linked to increased winter precipitation over the region. At the end of March, the strong midlatitude westerlies (jet stream) begin their northward retreat, and more tropical conditions affect the region, restraining wet conditions mostly over the CPGM and YP during April (Figure 3 left maps).

Summer precipitation in Mexico (May through October, see Figure 3 right maps) is, in large part, controlled by the North American Monsoon (NAM) system (Higgins et al., 2003), which is associated with thermal contrasts between the continent and adjacent eastern Pacific Ocean (Giovannettone & Barros, 2008; Higgins et al., 2003; Turrent & Cavazos, 2009).

Initially, greater rainfall occurs in southern Mexico in May, as the ITCZ assumes a northerly position during the Northern Hemisphere summer. This position, combined with modulation by easterly waves from the Caribbean Sea and the lower branch of easterly flow from over the Gulf of Mexico have a direct influence on the mountain ranges in southern Mexico (Giovannettone & Barros, 2007), initiating precipitation over the southern slopes and

systems is related to global scale circulations that function to move excess energy from the equatorial regions to poles affecting precipitation variability across Mexico on seasonal to decadal time scales. An unusually strong East Pacific High during the winter can steer storm systems far to the north, limiting cloud formation and precipitation with a drying effect over the river basins in much of northern Mexico (Brito-Castillo et al., 2003), while an unusually strong Bermuda High during the summer can enhance monsoon circulation and numerous transient disturbances (e.g. tropical storms, tropical easterly waves, inverted troughs, the Madden Julian Oscillation, and mid-latitude troughs) and bring above-average precipitation across much of Mexico (Douglas et al., 2007; Wang et al., 2007). Subtle changes to the position and strength of these circulation features appear to be connected to global scale shifts in circulation patterns related to changes in sea surface temperature patterns (e.g., El Niño-Southern Oscillation) affecting rainfall variability of Mexico at inter-annual timescales and can persist for years (e.g., the Pacific Decadal Oscillation, PDO), bringing extended periods of unusually wet or dry conditions at decadal timescales (Magaña et al., 2003; Pavia et al., 2006).

Mexico receives most of its annual rainfall during the summer, with a secondary peak occurring during the winter. From November through March (Figure 3 left maps), winter precipitation is associated with large-scale low pressure systems that traverse northern Mexico, drawing in moisture from the Pacific Ocean and producing widespread rain and snow at higher elevations of the Sierra Madre Occidental in northwestern Mexico, the northwestern portion of the Baja California Peninsula, the coastal plains of the Gulf of Mexico (CPGM), the Yucatan Peninsula (YP), and some portions of central Mexico (mainly in December and January), while the rest of the country generally experiences dry conditions. The effects of cold fronts coming from the northwest also cause lower temperatures in most of Mexico, but are more evident in the highlands of north and central Mexico, and on the CPGM, and the YP, where they are associated with strong northerly winds (nortes; outbreaks of cold weather), occasionally producing intense precipitation in the southeast, when the cold air masses are modified by the warm waters from the Gulf of Mexico. The large-scale forcing (1000 km-global) of regional climate variability, such as the location and strength of the subtropical westerly jet, determines the succession of weather events during winter in eastern Mexico. A change from zonal to more frequent meridional circulation has been linked to increased winter precipitation over the region. At the end of March, the strong midlatitude westerlies (jet stream) begin their northward retreat, and more tropical conditions affect the region, restraining wet conditions mostly over the CPGM

Summer precipitation in Mexico (May through October, see Figure 3 right maps) is, in large part, controlled by the North American Monsoon (NAM) system (Higgins et al., 2003), which is associated with thermal contrasts between the continent and adjacent eastern Pacific Ocean (Giovannettone & Barros, 2008; Higgins et al., 2003; Turrent & Cavazos, 2009). Initially, greater rainfall occurs in southern Mexico in May, as the ITCZ assumes a northerly position during the Northern Hemisphere summer. This position, combined with modulation by easterly waves from the Caribbean Sea and the lower branch of easterly flow from over the Gulf of Mexico have a direct influence on the mountain ranges in southern Mexico (Giovannettone & Barros, 2007), initiating precipitation over the southern slopes and

**3.1 Seasonality of precipitation** 

and YP during April (Figure 3 left maps).

propagating northward as the convective activity and easterlies strengthened in May-June. In southern and central Mexico, precipitation peaks in June and September with a decline in convective activity in late July-early August, causing a break period known as canicula (midsummer drought) (Magaña et al., 1999), as a result of the presence of an anticyclone southwest of Mexico over the ITCZ (Giovannettone & Barros, 2007).

Fig. 3. Distribution of long-term mean values of daily precipitation per month. The mean value is calculated for individual stations with larger than 40 yrs of records over the 1948– 2000 period. In lower left map, GC = Gulf of California; Sin = Sinaloa; Mich = Michoacan; SMO = Sierra Madre Occidental; SMS = Sierra Madre del Sur, GM = Gulf of Mexico, YP = Yucatan Peninsula.

Poleward extension of the summer precipitation regime in northwest Mexico starts with an abrupt onset of precipitation in late June or early July and quickly spreads northward along the western slopes of the Sierra Madre Occidental. Precipitation increases coincide with increased vertical transport of moisture by convection and southerly winds through the Gulf of California. During July through early September, the monsoon system is fully developed

Regional Pattern of Trends in Long-Term Precipitation and

than the negative phase.

**5. Climate data and quality control** 

Interval of observations (yr)

Stream Flow Observations: Singularities in a Changing Climate in Mexico 395

decades of the 1980s and 1990s appear to have favored fewer storms passing into the central part of the Baja California Peninsula. Diaz et al. (2008), indicate that tropical cyclone contributions to rainfall in the Baja California Peninsula and the State of Sinaloa display a statistically significant correlation with PDO, suggesting a larger contribution of tropical cyclones on annual precipitation in both areas occurs during the positive phase of the PDO

Perhaps, the main difficulty that one encounters when trying to discern regional structures of trends is the lack of sufficient quality records in space and time; this is not an exception in Mexico. According to available information, a little more than 6000 weather stations (Quintas, 2000) and 1475 gauging stations (Banco Nacional de Datos de Aguas Superficiales-BANDAS, 1997) have been put in operation around the country. Unfortunately, only 20% of both kinds of stations have > 40 yrs of records, with datasets on runoff within a river basin (Milly et al., 2005) even smaller than precipitation records (Table 1). The highest density of stations is from 1960 through 2004 (Figure 4) with an abrupt decline at the beginning of the 1980s in both sources of data and slight recovery in rainfall datasets in the 1990s. Considering just those stations with more than four decades of daily observations (between 1920 and 2008), it is still possible to have sufficient information to include all the country (Figure 5) and preserve the longest time series for analysis: 1161 for precipitation and 141 for streamflow data sets. The exception is the Baja California and Yucatan Peninsulas, northwestern Sonora (NWS), and the Altiplano Mexicano, where streamflow observations are scarce or non-existent (Figure 5). Selected basins range from very large (124,000 km²) to very small (25 km²), with an average of 14,500 km². For precipitation records, some parts of the central Baja California Peninsula; NWS; northern Mexico, and mid-level elevations of central Mexico in the Pacific drainage

> Less than 1 101 (6.8) 732 (12.1) 1 - 10 442 (30.0) 1139 (18.8) 11 - 20 314 (21.3) 1065 (17.6) 21 - 30 241 (16.3) 1187 (19.6) 31 - 40 165 (11.2) 773 (12.8) 41 - 50 123 (8.3) 664 (11.0) 51 - 60 63 (4.3) 325 (5.4) 61 - 70 25 (1.7) 139 (2.3) More than 70 1 (0.1) 37 (0.6) Total 1475 (100) 6061 (100)

Table 1. Density of gauging sites and weather stations with measurements of streamflow

Number of stations having data Gauging sites Weather stations

region (near 22° latitude) have no stations with long-term records.

and rainfall from 1920–2008. Percentages are shown in parentheses.

and the heaviest rainfalls occur west of the Sierra Madre Occidental. In late September to October, precipitation over northwest Mexico diminishes in response to a southward movement of the NAM upper level anticyclone to weakening of land surface heating and southerly flow and to the retraction of the North Atlantic Subtropical High eastward.

Other sources of moisture in Mexico during summertime are mesoscale convective systems (Farfan & Zhender, 1994) and tropical cyclones (Farfan, 2011; Jáuregui, 2003).

## **4. Tropical cyclones and their contribution to rainfall**

It has been established that sea surface temperature (SST) is increasing all over the world (Barnett et al., 2005). For tropical SSTs, the total increase has been on the order of 0.5°C since the mid-nineteenth century (Houghton et al. 2001). Warming signals have been found to be far stronger than would be expected from natural internal variability (i.e., external forcing such as solar variability or volcanic activity), attributing ocean warming signal to anthropogenic forcing (Barnett et al., 2005). A warmer tropical ocean brings increased water vapor in the lower troposphere and both tend to enhance moisture convergence for a given amount of mass convergence. This increases the energy available for atmospheric convection and rainfall rates in systems, such as thunderstorm and cyclones (Knutson et al., 2010) where moisture convergence is an important component of the water-vapor budget. However, the issue of attribution of increased hurricane frequency to increasing SST has been a matter of academic debate, since there is no sound theoretical basis for drawing any conclusions about how anthropogenic climate change affects hurricane number or tracks (Trenberth, 2005). It is known than over the 20th century, there is a nonlinear upward trend in SSTs with domination of multi-decadal variability being a pronounced trend in the past three to four decades (Trenberth, 2005). The study by Webster et al. (2005), analyzing global hurricane data, shows that, opposed to increasing SST, no global trend has yet emerged in the number of tropical storms or hurricanes, although model results and observations suggest a shift in hurricane intensities toward extreme hurricanes (Knutson et al., 2010; Webster et al., 2005; Yu et al., 2010).

With extensive coastline bordering both oceans, Mexico is particularly vulnerable to hurricanes. Moreover, the eastern Pacific Ocean is the most productive in hurricane activity of any basin per unit area worldwide (McBride, 1995). From May through October, an average of 15 named tropical cyclones (TC) form every summer season and provide changes in moisture content to western Mexico (Farfan, 2011). According to Jáuregui (2003), from 1951–2000, Pacific hurricane hits were more frequent on coastal areas of Sinaloa, southern Baja California Peninsula, and Michoacan. More recently, analysis by Farfan (2011) states that, from 1970 through 2009, the Baja California Peninsula was the most affected area, with 97% probability of being affected by at least one TC each season and 46% probability of having at least one TC move inland. On the Atlantic side, the Yucatan Peninsula and the northern state of Tamaulipas were the most exposed to these storms (Jáuregui, 2003). September is the month with the highest rate (33% compared to the eastern Pacific Ocean) of landfall in Mexico (Diaz et al., 2008; Farfan, 2011; Jáuregui, 2003). The study made by Jáuregui (2003) indicates that during 1951–2000, landfall hurricane trends are declining for both coasts of the country. Attempts to understand the links between SST changes and interannual variations of cyclone formations in western Mexico indicate that the strong El Niño decades of the 1980s and 1990s appear to have favored fewer storms passing into the central part of the Baja California Peninsula. Diaz et al. (2008), indicate that tropical cyclone contributions to rainfall in the Baja California Peninsula and the State of Sinaloa display a statistically significant correlation with PDO, suggesting a larger contribution of tropical cyclones on annual precipitation in both areas occurs during the positive phase of the PDO than the negative phase.

## **5. Climate data and quality control**

394 Greenhouse Gases – Emission, Measurement and Management

and the heaviest rainfalls occur west of the Sierra Madre Occidental. In late September to October, precipitation over northwest Mexico diminishes in response to a southward movement of the NAM upper level anticyclone to weakening of land surface heating and

Other sources of moisture in Mexico during summertime are mesoscale convective systems

It has been established that sea surface temperature (SST) is increasing all over the world (Barnett et al., 2005). For tropical SSTs, the total increase has been on the order of 0.5°C since the mid-nineteenth century (Houghton et al. 2001). Warming signals have been found to be far stronger than would be expected from natural internal variability (i.e., external forcing such as solar variability or volcanic activity), attributing ocean warming signal to anthropogenic forcing (Barnett et al., 2005). A warmer tropical ocean brings increased water vapor in the lower troposphere and both tend to enhance moisture convergence for a given amount of mass convergence. This increases the energy available for atmospheric convection and rainfall rates in systems, such as thunderstorm and cyclones (Knutson et al., 2010) where moisture convergence is an important component of the water-vapor budget. However, the issue of attribution of increased hurricane frequency to increasing SST has been a matter of academic debate, since there is no sound theoretical basis for drawing any conclusions about how anthropogenic climate change affects hurricane number or tracks (Trenberth, 2005). It is known than over the 20th century, there is a nonlinear upward trend in SSTs with domination of multi-decadal variability being a pronounced trend in the past three to four decades (Trenberth, 2005). The study by Webster et al. (2005), analyzing global hurricane data, shows that, opposed to increasing SST, no global trend has yet emerged in the number of tropical storms or hurricanes, although model results and observations suggest a shift in hurricane intensities toward extreme hurricanes (Knutson et al., 2010;

With extensive coastline bordering both oceans, Mexico is particularly vulnerable to hurricanes. Moreover, the eastern Pacific Ocean is the most productive in hurricane activity of any basin per unit area worldwide (McBride, 1995). From May through October, an average of 15 named tropical cyclones (TC) form every summer season and provide changes in moisture content to western Mexico (Farfan, 2011). According to Jáuregui (2003), from 1951–2000, Pacific hurricane hits were more frequent on coastal areas of Sinaloa, southern Baja California Peninsula, and Michoacan. More recently, analysis by Farfan (2011) states that, from 1970 through 2009, the Baja California Peninsula was the most affected area, with 97% probability of being affected by at least one TC each season and 46% probability of having at least one TC move inland. On the Atlantic side, the Yucatan Peninsula and the northern state of Tamaulipas were the most exposed to these storms (Jáuregui, 2003). September is the month with the highest rate (33% compared to the eastern Pacific Ocean) of landfall in Mexico (Diaz et al., 2008; Farfan, 2011; Jáuregui, 2003). The study made by Jáuregui (2003) indicates that during 1951–2000, landfall hurricane trends are declining for both coasts of the country. Attempts to understand the links between SST changes and interannual variations of cyclone formations in western Mexico indicate that the strong El Niño

southerly flow and to the retraction of the North Atlantic Subtropical High eastward.

(Farfan & Zhender, 1994) and tropical cyclones (Farfan, 2011; Jáuregui, 2003).

**4. Tropical cyclones and their contribution to rainfall** 

Webster et al., 2005; Yu et al., 2010).

Perhaps, the main difficulty that one encounters when trying to discern regional structures of trends is the lack of sufficient quality records in space and time; this is not an exception in Mexico. According to available information, a little more than 6000 weather stations (Quintas, 2000) and 1475 gauging stations (Banco Nacional de Datos de Aguas Superficiales-BANDAS, 1997) have been put in operation around the country. Unfortunately, only 20% of both kinds of stations have > 40 yrs of records, with datasets on runoff within a river basin (Milly et al., 2005) even smaller than precipitation records (Table 1). The highest density of stations is from 1960 through 2004 (Figure 4) with an abrupt decline at the beginning of the 1980s in both sources of data and slight recovery in rainfall datasets in the 1990s. Considering just those stations with more than four decades of daily observations (between 1920 and 2008), it is still possible to have sufficient information to include all the country (Figure 5) and preserve the longest time series for analysis: 1161 for precipitation and 141 for streamflow data sets. The exception is the Baja California and Yucatan Peninsulas, northwestern Sonora (NWS), and the Altiplano Mexicano, where streamflow observations are scarce or non-existent (Figure 5). Selected basins range from very large (124,000 km²) to very small (25 km²), with an average of 14,500 km². For precipitation records, some parts of the central Baja California Peninsula; NWS; northern Mexico, and mid-level elevations of central Mexico in the Pacific drainage region (near 22° latitude) have no stations with long-term records.


Table 1. Density of gauging sites and weather stations with measurements of streamflow and rainfall from 1920–2008. Percentages are shown in parentheses.

Regional Pattern of Trends in Long-Term Precipitation and

Stream Flow Observations: Singularities in a Changing Climate in Mexico 397

For all selected datasets, quality control was performed to remove implausible values (i.e., values arising from data entry errors evident as data points with extreme deviations from the overall mean of the series without replication at surrounding station datasets). For streamflow time series, significance tests for variances (*F* distribution) and means (*t*-test distribution) were performed to test for the null hypothesis about the similarity of these parameters inside the series. Hypothesis testing was performed for each series, dividing the series in two equally long parts and comparing the mean and variance of each part. Then, the results were accepted if there were statistically significant differences between the parameters at the 0.05 alpha level. When the series showed significant differences between

the parameters, this was considered an indication of change in the long term period.

Fig. 5. Spatial distribution of 1161 selected weather stations (a), and 141 selected gauging sites (b) having more than 40 years of records. The maps are divided in domains: N = North; C = Center; SE = Southeast, and YP = Yucatan Peninsula. Density of stations is displayed in parentheses over each domain. Lines in (a) divide the main drainage areas of Mexico: PW = Pacific watershed; GMW = Gulf of Mexico watershed; NCB = northern closed basins; AP = Altiplano Mexicano, and main mountain arrangements: SMO = Sierra Madre Occidental; SMOr = Sierra Madre Oriental; TMVB = Trans-Mexican Volcanic Belt, and SMS = Sierra Madre del Sur. Polygons displayed inside the map in (b) are the catchment areas of

each identified gauge point.


Table 2. The 1921–2008 mean annual and seasonal precipitation (in mm) over each of the domains derived from weather stations displayed in Figure 5. Shown in parentheses is the proportion of seasonal precipitation relative to annual precipitation. N = north domain; C = center domain; SE = southeast domain; YP = Yucatan Peninsula domain.

From daily observations, we calculated total annual (preceding May to following April, the year is of the last month) precipitation and seasonal accumulations for summer (May– October), and winter (November–April) because they are the main precipitation seasons in Mexico (see Figure 3). For each precipitation dataset, we also estimated TDP >0 (expressed in n-days – total days with precipitation >0 mm) for annual and seasonal components. For streamflow records, we estimated mean annual and seasonal streamflow for the same months as precipitation time series. The mean magnitudes of streamflow records, derived from the 141 gauging stations are 295 m³/s, 374 m³/s, and 59.1 m³/s for mean annual, summer, and winter streamflow, respectively. In all variables, a year was considered valid if, at least 85% of the daily records were available (i.e., >310 daily observations for annual series; >156 for summer series, and >153 for winter series), otherwise the year was considered a gap.

Fig. 4. Series displaying daily density observations of precipitation (a) and streamflow (b) from 1920–2008 derived from 6061 weather and 1554 gauging stations in Mexico.

 N C SE YP Annual 402 899 1716 1220 Summer 311 (78) 791 (88) 1446 (84) 955 (79) Winter 89 (22) 109 (12) 288 (16) 257 (21)

Table 2. The 1921–2008 mean annual and seasonal precipitation (in mm) over each of the domains derived from weather stations displayed in Figure 5. Shown in parentheses is the proportion of seasonal precipitation relative to annual precipitation. N = north domain; C =

From daily observations, we calculated total annual (preceding May to following April, the year is of the last month) precipitation and seasonal accumulations for summer (May– October), and winter (November–April) because they are the main precipitation seasons in Mexico (see Figure 3). For each precipitation dataset, we also estimated TDP >0 (expressed in n-days – total days with precipitation >0 mm) for annual and seasonal components. For streamflow records, we estimated mean annual and seasonal streamflow for the same months as precipitation time series. The mean magnitudes of streamflow records, derived from the 141 gauging stations are 295 m³/s, 374 m³/s, and 59.1 m³/s for mean annual, summer, and winter streamflow, respectively. In all variables, a year was considered valid if, at least 85% of the daily records were available (i.e., >310 daily observations for annual series; >156 for summer series, and >153 for winter series),

Fig. 4. Series displaying daily density observations of precipitation (a) and streamflow (b)

from 1920–2008 derived from 6061 weather and 1554 gauging stations in Mexico.

center domain; SE = southeast domain; YP = Yucatan Peninsula domain.

otherwise the year was considered a gap.

For all selected datasets, quality control was performed to remove implausible values (i.e., values arising from data entry errors evident as data points with extreme deviations from the overall mean of the series without replication at surrounding station datasets). For streamflow time series, significance tests for variances (*F* distribution) and means (*t*-test distribution) were performed to test for the null hypothesis about the similarity of these parameters inside the series. Hypothesis testing was performed for each series, dividing the series in two equally long parts and comparing the mean and variance of each part. Then, the results were accepted if there were statistically significant differences between the parameters at the 0.05 alpha level. When the series showed significant differences between the parameters, this was considered an indication of change in the long term period.

Fig. 5. Spatial distribution of 1161 selected weather stations (a), and 141 selected gauging sites (b) having more than 40 years of records. The maps are divided in domains: N = North; C = Center; SE = Southeast, and YP = Yucatan Peninsula. Density of stations is displayed in parentheses over each domain. Lines in (a) divide the main drainage areas of Mexico: PW = Pacific watershed; GMW = Gulf of Mexico watershed; NCB = northern closed basins; AP = Altiplano Mexicano, and main mountain arrangements: SMO = Sierra Madre Occidental; SMOr = Sierra Madre Oriental; TMVB = Trans-Mexican Volcanic Belt, and SMS = Sierra Madre del Sur. Polygons displayed inside the map in (b) are the catchment areas of each identified gauge point.

Regional Pattern of Trends in Long-Term Precipitation and

2009).

1920

700 800 900 1000 1100 1200 Total annual precipitation (mm)

Total annual precipitation (mm)

1930

1940

1950

1960

1970

1980

1990

2000

**a)**

**b)**

**c)**

**d)**

magnitude and variability in each data set.

2010

Total annual precipitation (mm)

Fig. 6. Mean total annual (May–April) precipitation (left series: a–d; mm), and mean total seasonal precipitation in Mexico for summer (May–October; middle series: e-h) and winter (November–April: right series: i–l), from weather stations displayed in Figure 6 over Ndomain (a, e, i); C-domain (b, f, j); SE-domain (c, g, k) and YP-domain (d, h, l). Heavy lines are 9-yr running means and horizontal lines display the mean values over the period for each variable (see Table 2). Note the different ranges in Y-scales due to differences in

Total summer precipitation (mm)

Total summer precipitation (mm)

Total annual precipitation (mm)

Stream Flow Observations: Singularities in a Changing Climate in Mexico 399

The multi-decadal behavior observed over the SE and YP domains display different structures. Over both domains, dry and wet periods in mean total annual and seasonal precipitation, (see Figures 6c, d, g, h, k & l), appear to last longer, a situation that is unfavorable for Mexico, since over precipitation in both domains is larger (see Table 2) and the effects over the country's national economy might be large, as occurred in the past (Mendoza & Velasco, 2006). Over the SE domain, for example, the early dry period that started in 1926 lasted almost three decades (until the 1950s) and is evident in mean total annual (Figure 6c) and summer (Figure 6g) precipitation. In mean total winter precipitation (Figure 6k), this early dry period ended some years before (mid 1940s). The dry period was followed by a prolonged wet period that occurred from the 1950s onward (Figures 6c, g, k) with a near-normal period at the end of the 1980s-beginning of the 1990s. Comparing the multi-decadal variability between N and SE domains (i.e., Figures 6a, c, e, g, i & k), an apparent opposite pattern emerges (i.e., droughts in the N domain generally coincide with anomalously wet conditions in the SE domain and vice versa) that is related to tropical SST conditions (Matias & Magaña, 2009). The interaction between easterly waves and trade winds over the Gulf of Mexico and the Caribbean Sea appears to be crucial to explain spatial patterns of droughts that affected Mexico (Matias & Magaña,

Annual (May-April) Summer (May-October) Winter (November-April)

200 250 300 350 400 450 Total summer precipitation (mm)

**e)**

**f)**

**g)**

**h)**

Total summer precipitation (mm)

Total winter precipitation (mm)

Total winter precipitation (mm)

**j)**

**k)**

**l)**

**i)**

Total winter precipitation (mm)

Total winter precipitation (mm)

Trend analysis was performed for precipitation (total annual and total seasonal observations and their correspondent TDP >0) and streamflow data sets over Mexico from 1161 weather stations and 141 gauging stations (shown in Figure 5), using the procedure explained above. For precipitation time series, our interest was on trends observed in the sequence of days with recorded rainfall (TDP >0). This was done to prove if a station had a declining or increasing trend in total annual and seasonal precipitation corresponding to a downward or upward trend in TDP >0.

## **6. Regionalization**

To simplify the analysis of the regional structure of the trends observed in precipitation and streamflow, the country was divided into polygons of different sizes in the domains (N, C, SE, and YP). Density of weather stations across the domains is displayed in Figure 5. The C domain has the highest density of stations (700 weather stations and 85 gauging sites), while N domain, the largest area, has the lowest density (270 and 29). Climate conditions are different across each of the domains, with a gradient from arid and semiarid climates over the N domain; sub-humid climates over the C domain, except over the Altiplano Mexicano and middle-to-high elevations in the southeastern part of the C domain, where semi-arid and humid climates dominate, respectively; humid climate over the SE domain, and sub-humid climates over the YP domain (Garcia, 2004). The 1921–2008 mean annual and seasonal precipitation over each domain displays the intra-seasonal differences inside and among the domains (Table 2). As can be seen in Table 2, the arid and semi-arid N domain (annual mean = 402 mm) contrasts with the humid SE domain (1716 mm). The larger winter contribution to annual total is observed over the N domain (22%) followed by the YP domain (21%), the SE domain (16%), and the C domain (12%). This agrees with the larger influence of the intrusion of cold fronts that produce winter precipitation over the N, SE, and YP domains (Figure 3). Each domain is characterized by high annual and seasonal inter-annual variability that is superimposed on multi-decadal variability (Figure 6). Length and magnitude of long-term (multi-decadal) wet and dry periods have been different between the domains, although they overlap in some years. For example (see Figures 6a & b), the prolonged 1940–55 and 1996-onward dry periods in mean total annual precipitation over the C and N domains (early dry period lasting longer in the N domain until 1960, while the actual dry period started earlier in the C domain in 1980); the moderate 1960–1980 wet period between dry periods and the earlier wet period that ended in 1940 over both domains are evidence that wet and dry periods can affect more than half of the country at the same time, affecting not only mean total annual precipitation but also mean total seasonal precipitation (see Figures 6e, f, i & j). These multi-decadal annual and seasonal fluctuations in time series of N and C domains are linked to the PDO shifts, from cold to warm PDO in the mid-1970s and from warm to a temporarily cold PDO at the end of the 1990s (Pavia et al., 2006). Moreover, in the northwestern portion of the N-domain (known as the monsoon region, Higgins et al., 2003), it is known that winter precipitation is in phase with the PDO (Brito-Castillo et al., 2002; Pavia et al., 2006). Stahle et al. (2009) propose that the recent drought (or the 2000s drought) that has affected the N and C domains (Figures 6a, b, e, f, i & j), is associated with a downward trend in precipitation projected for the regions (Seager et al., 2007), based on anthropogenic global warming.

Trend analysis was performed for precipitation (total annual and total seasonal observations and their correspondent TDP >0) and streamflow data sets over Mexico from 1161 weather stations and 141 gauging stations (shown in Figure 5), using the procedure explained above. For precipitation time series, our interest was on trends observed in the sequence of days with recorded rainfall (TDP >0). This was done to prove if a station had a declining or increasing trend in total annual and seasonal precipitation corresponding to a downward or

To simplify the analysis of the regional structure of the trends observed in precipitation and streamflow, the country was divided into polygons of different sizes in the domains (N, C, SE, and YP). Density of weather stations across the domains is displayed in Figure 5. The C domain has the highest density of stations (700 weather stations and 85 gauging sites), while N domain, the largest area, has the lowest density (270 and 29). Climate conditions are different across each of the domains, with a gradient from arid and semiarid climates over the N domain; sub-humid climates over the C domain, except over the Altiplano Mexicano and middle-to-high elevations in the southeastern part of the C domain, where semi-arid and humid climates dominate, respectively; humid climate over the SE domain, and sub-humid climates over the YP domain (Garcia, 2004). The 1921–2008 mean annual and seasonal precipitation over each domain displays the intra-seasonal differences inside and among the domains (Table 2). As can be seen in Table 2, the arid and semi-arid N domain (annual mean = 402 mm) contrasts with the humid SE domain (1716 mm). The larger winter contribution to annual total is observed over the N domain (22%) followed by the YP domain (21%), the SE domain (16%), and the C domain (12%). This agrees with the larger influence of the intrusion of cold fronts that produce winter precipitation over the N, SE, and YP domains (Figure 3). Each domain is characterized by high annual and seasonal inter-annual variability that is superimposed on multi-decadal variability (Figure 6). Length and magnitude of long-term (multi-decadal) wet and dry periods have been different between the domains, although they overlap in some years. For example (see Figures 6a & b), the prolonged 1940–55 and 1996-onward dry periods in mean total annual precipitation over the C and N domains (early dry period lasting longer in the N domain until 1960, while the actual dry period started earlier in the C domain in 1980); the moderate 1960–1980 wet period between dry periods and the earlier wet period that ended in 1940 over both domains are evidence that wet and dry periods can affect more than half of the country at the same time, affecting not only mean total annual precipitation but also mean total seasonal precipitation (see Figures 6e, f, i & j). These multi-decadal annual and seasonal fluctuations in time series of N and C domains are linked to the PDO shifts, from cold to warm PDO in the mid-1970s and from warm to a temporarily cold PDO at the end of the 1990s (Pavia et al., 2006). Moreover, in the northwestern portion of the N-domain (known as the monsoon region, Higgins et al., 2003), it is known that winter precipitation is in phase with the PDO (Brito-Castillo et al., 2002; Pavia et al., 2006). Stahle et al. (2009) propose that the recent drought (or the 2000s drought) that has affected the N and C domains (Figures 6a, b, e, f, i & j), is associated with a downward trend in precipitation projected for the regions (Seager et al., 2007),

upward trend in TDP >0.

based on anthropogenic global warming.

**6. Regionalization** 

The multi-decadal behavior observed over the SE and YP domains display different structures. Over both domains, dry and wet periods in mean total annual and seasonal precipitation, (see Figures 6c, d, g, h, k & l), appear to last longer, a situation that is unfavorable for Mexico, since over precipitation in both domains is larger (see Table 2) and the effects over the country's national economy might be large, as occurred in the past (Mendoza & Velasco, 2006). Over the SE domain, for example, the early dry period that started in 1926 lasted almost three decades (until the 1950s) and is evident in mean total annual (Figure 6c) and summer (Figure 6g) precipitation. In mean total winter precipitation (Figure 6k), this early dry period ended some years before (mid 1940s). The dry period was followed by a prolonged wet period that occurred from the 1950s onward (Figures 6c, g, k) with a near-normal period at the end of the 1980s-beginning of the 1990s. Comparing the multi-decadal variability between N and SE domains (i.e., Figures 6a, c, e, g, i & k), an apparent opposite pattern emerges (i.e., droughts in the N domain generally coincide with anomalously wet conditions in the SE domain and vice versa) that is related to tropical SST conditions (Matias & Magaña, 2009). The interaction between easterly waves and trade winds over the Gulf of Mexico and the Caribbean Sea appears to be crucial to explain spatial patterns of droughts that affected Mexico (Matias & Magaña, 2009).

Fig. 6. Mean total annual (May–April) precipitation (left series: a–d; mm), and mean total seasonal precipitation in Mexico for summer (May–October; middle series: e-h) and winter (November–April: right series: i–l), from weather stations displayed in Figure 6 over Ndomain (a, e, i); C-domain (b, f, j); SE-domain (c, g, k) and YP-domain (d, h, l). Heavy lines are 9-yr running means and horizontal lines display the mean values over the period for each variable (see Table 2). Note the different ranges in Y-scales due to differences in magnitude and variability in each data set.

Regional Pattern of Trends in Long-Term Precipitation and

confidence intervals for the trends.

confidence intervals for the trends.

confidence intervals for the trends.

Stream Flow Observations: Singularities in a Changing Climate in Mexico 401

than over the Pacific drainage area (except in total winter precipitation, where there is no increasing precipitation trends on either side of the main drainage areas), and over the main drainage area of the Rio Lerma-Santiago basin; a tendency for upward mean seasonal precipitation and downward mean annual precipitation trends over the western half of the YP domain; and finally, a tendency for mean annual and mean summer downward precipitation trends and a mean upward winter precipitation trend over the SE domain. Over the SE domain, individual stations located through the high mountain

areas show a tendency for upward trends in annual and seasonal accumulations.

Domain N(+) N(-) Total Mean Trend (mm/yr) N 28 7 35 2.54 ± 1.36 C 26 75 101 –3.79 ± 1.57 SE 9 14 23 –3.99 ± 5.40 YP 6 7 13 –0.40 ± 3.76 Overall 69 103 172 –2.38 ± 1.31 Table 3. Mean annual precipitation trends derived from all weather stations having

statistically significant trends and located in the North (N), Central (C), Southeast (SE), and Yucatan Peninsula (YP) domains, respectively. N(+) and N(-) are the number of stations displaying upward and downward trends, respectively; the ± values define the 95%

Domain N(+) N(-) Total Mean Trend (mm/yr) N 22 3 25 2.39 ± 0.26 C 38 69 107 –2.14 ± 1.47 SE 8 12 20 –3.03 ± 4.93 YP 6 3 9 1.90 ± 3.86 Overall 74 87 161 –1.32 ± 1.22 Table 4. Mean summer precipitation trends derived from all the weather stations having statistically significant trends and located in the North (N), Central (C), Southeast (SE), and Yucatan Peninsula (YP) domains respectively. N(+) and N(-) are the number of stations displaying upward and downward trends, respectively; the ± values define the 95%

Domain N(+) N(-) Total Mean Trend (mm/yr) N 16 6 22 –0.68 ± 0.74 C 10 36 46 –1.13 ± 0.65 SE 9 2 11 0.68 ± 2.03 YP 4 1 5 –0.91 ± 2.14 Overall 39 45 84 –0.30 ± 0.54 Table 5. Mean winter precipitation trends derived from all the weather stations having statistically significant trends located in the North (N), Central (C), Southeast (SE), and Yucatan Peninsula (YP) domains, respectively. N(+) and N(-) are the number of stations displaying upward and downward trends, respectively; the ± values define the 95%

### **7. Regional structures of precipitation trends**

The primary goal in detecting trends in streamflow is to identify changes associated with specific physical processes (causal factors) or combinations of processes. But this procedure requires a complete knowledge of the components of the balance equation. Instead, in this study, we start identifying regional trends in precipitation records and then corroborate the corresponding regional trends in streamflow to confirm if they are spatially consistent with one another. This procedure helps, in some way to avoid misleading interpretations, particularly in streamflow datasets where diversion of water for irrigation, control of rivers by damming (when the gauge station is located downstream), or pumping of underground water results in a strong (non-climatic) decline of surface water through time, implying that the causal factor for the declining trend is mostly anthropogenic, and probably with very localized (i.e., non-regional) effects.

Tables 3–5 show the mean magnitude of annual and seasonal trends over each of the domains. In all cases, except in the C domain, total annual precipitation showed a larger number of statistically significant trends than seasonal accumulations. Throughout Mexico, the number of stations displaying downward precipitation trends is larger than stations displaying upward trends, resulting in overall mean annual and summer declining trends. However, the statistical uncertainty in the mean winter trend, based on the 95% confidence interval, includes the zero trend (–0.30 ± 0.54 mm/yr; Table 5). The decrease number of stations having significant trends in winter and their high and uneven spread over the domains (i.e. N, SE and YP) are factors resulted in increasing error in estimating the mean trend, therefore the magnitude of the standard error of trends, derived from weather stations having statistically significant trends in total winter precipitation is very large, indicating that with the available information there is not sufficient evidence to conclude there is an overall mean winter declining trend in Mexico. This conclusion is also valid for the N, SE, and YP domains for mean winter trends (Table 5) and for the SE and YP domains for mean annual (Table 3) and summer (Table 4) trends. In all these cases, the standard error of the trends exceeds the mean magnitude of their respective trends. A different situation occurs over the N and C domains, where the statistical uncertainty of mean annual (Table 3) and summer (Table 4) trends and over the C domain for mean winter trend (Table 5) is smaller than the mean magnitude of the trend. In these cases, we conclude that the mean annual and summer trends are toward increasing precipitation over the N domain and, on the contrary, toward declining precipitation over the C domain (including the mean winter precipitation trend). This last conclusion can be easily verified in Figure 7 that shows the distribution of weather stations over each of the domains that display statistically significant annual (Figure 7a) and seasonal (Figures 7b & c) trends. In all maps of Figure 7, some basic coherent regional trend patterns can be identified for annual and seasonal accumulations over the N domain, such as the tendency for higher precipitation over the high elevations of northeastern Mexico (verified by the line indicating the position of the continental divide, i.e., the limits of the Sierra Madre Oriental), and over middle to high elevations of northwestern Mexico, including the Baja California Peninsula; a tendency for downward precipitation trends over the C domain, with a larger number of stations displaying declining trends over the middle-to-high elevations of the Gulf of Mexico drainage area

The primary goal in detecting trends in streamflow is to identify changes associated with specific physical processes (causal factors) or combinations of processes. But this procedure requires a complete knowledge of the components of the balance equation. Instead, in this study, we start identifying regional trends in precipitation records and then corroborate the corresponding regional trends in streamflow to confirm if they are spatially consistent with one another. This procedure helps, in some way to avoid misleading interpretations, particularly in streamflow datasets where diversion of water for irrigation, control of rivers by damming (when the gauge station is located downstream), or pumping of underground water results in a strong (non-climatic) decline of surface water through time, implying that the causal factor for the declining trend is mostly anthropogenic, and probably with very

Tables 3–5 show the mean magnitude of annual and seasonal trends over each of the domains. In all cases, except in the C domain, total annual precipitation showed a larger number of statistically significant trends than seasonal accumulations. Throughout Mexico, the number of stations displaying downward precipitation trends is larger than stations displaying upward trends, resulting in overall mean annual and summer declining trends. However, the statistical uncertainty in the mean winter trend, based on the 95% confidence interval, includes the zero trend (–0.30 ± 0.54 mm/yr; Table 5). The decrease number of stations having significant trends in winter and their high and uneven spread over the domains (i.e. N, SE and YP) are factors resulted in increasing error in estimating the mean trend, therefore the magnitude of the standard error of trends, derived from weather stations having statistically significant trends in total winter precipitation is very large, indicating that with the available information there is not sufficient evidence to conclude there is an overall mean winter declining trend in Mexico. This conclusion is also valid for the N, SE, and YP domains for mean winter trends (Table 5) and for the SE and YP domains for mean annual (Table 3) and summer (Table 4) trends. In all these cases, the standard error of the trends exceeds the mean magnitude of their respective trends. A different situation occurs over the N and C domains, where the statistical uncertainty of mean annual (Table 3) and summer (Table 4) trends and over the C domain for mean winter trend (Table 5) is smaller than the mean magnitude of the trend. In these cases, we conclude that the mean annual and summer trends are toward increasing precipitation over the N domain and, on the contrary, toward declining precipitation over the C domain (including the mean winter precipitation trend). This last conclusion can be easily verified in Figure 7 that shows the distribution of weather stations over each of the domains that display statistically significant annual (Figure 7a) and seasonal (Figures 7b & c) trends. In all maps of Figure 7, some basic coherent regional trend patterns can be identified for annual and seasonal accumulations over the N domain, such as the tendency for higher precipitation over the high elevations of northeastern Mexico (verified by the line indicating the position of the continental divide, i.e., the limits of the Sierra Madre Oriental), and over middle to high elevations of northwestern Mexico, including the Baja California Peninsula; a tendency for downward precipitation trends over the C domain, with a larger number of stations displaying declining trends over the middle-to-high elevations of the Gulf of Mexico drainage area

**7. Regional structures of precipitation trends** 

localized (i.e., non-regional) effects.

than over the Pacific drainage area (except in total winter precipitation, where there is no increasing precipitation trends on either side of the main drainage areas), and over the main drainage area of the Rio Lerma-Santiago basin; a tendency for upward mean seasonal precipitation and downward mean annual precipitation trends over the western half of the YP domain; and finally, a tendency for mean annual and mean summer downward precipitation trends and a mean upward winter precipitation trend over the SE domain. Over the SE domain, individual stations located through the high mountain areas show a tendency for upward trends in annual and seasonal accumulations.


Table 3. Mean annual precipitation trends derived from all weather stations having statistically significant trends and located in the North (N), Central (C), Southeast (SE), and Yucatan Peninsula (YP) domains, respectively. N(+) and N(-) are the number of stations displaying upward and downward trends, respectively; the ± values define the 95% confidence intervals for the trends.


Table 4. Mean summer precipitation trends derived from all the weather stations having statistically significant trends and located in the North (N), Central (C), Southeast (SE), and Yucatan Peninsula (YP) domains respectively. N(+) and N(-) are the number of stations displaying upward and downward trends, respectively; the ± values define the 95% confidence intervals for the trends.


Table 5. Mean winter precipitation trends derived from all the weather stations having statistically significant trends located in the North (N), Central (C), Southeast (SE), and Yucatan Peninsula (YP) domains, respectively. N(+) and N(-) are the number of stations displaying upward and downward trends, respectively; the ± values define the 95% confidence intervals for the trends.

Regional Pattern of Trends in Long-Term Precipitation and


TPS






affecting the overall trend.

Trend: 0.13 ± 0.13 days/yr

Trend: 2.32 ± 1.26 mm/yr

Trend: 0.52 ± 0.18 days/yr

Trend: 1.21 ± 0.33 days/yr

Trend: 0.29 ± 0.15 days/yr

Trend: 0.22 ± 0.19 days/yr

Trend: 0.28 ± 0.15 days/yr

downward (circles) trends.

Trend: 8.15 ± 3.23 mm/yr

Trend: 4.49 ± 3.46 mm/yr

Trend: 3.31 ± 1.72 mm/yr

Trend: 8.08 ± 4.16 mm/yr

Trend: 8.38 ± 4.00 mm/yr


b)

a)


c)


d)


e)


f)


Stream Flow Observations: Singularities in a Changing Climate in Mexico 403

as TDP >0 changes is highly possible and the occurrence of heavy rainfall (from tropical cyclones, for example) only temporarily changes the total annual precipitation without

> -2 0 2 z scores

> > h)

g)

TDP>0 series


i)


j)

k)

l)




Fig. 8. Examples of observed time series of selected variables (expressed as z scores) displaying upward (left series) and downward (right series) total annual precipitation trends. A couple of series have been plotted from the same selected station, i.e., an upper –

TDP >0 (total days with rainfall > 0 mm; lines with ×s) series, and lower TPS-total precipitation series (lines with •). For each variable, the magnitude of the trend with the 95% confidence intervals with their respective units is shown. The map to the right displays the position of the selected stations corresponding to the variables with upward (points) and

Trend: -10.5 ± 6.33 mm/yr

Trend: -8.41 ± 5.77 mm/yr

Trend: -0.86 ± 0.43 days/yr

Trend: -10.8 ± 7.67 mm/yr

Trend: -3.82 ± 3.35 mm/yr

Trend: -4.04 ± 2.23 mm/yr

Trend: -20.67 ± 12.47mm/yr







Trend: -0.52 ± 0.28 days/yr

Trend: -0.83 ± 0.22 days/yr

Trend: -0.92 ± 0.37 days/yr

Trend: -1.16 ± 0.28 days/yr

Trend: -1.38 ± 0.25 days/yr

Fig. 7. Maps displaying the distribution of weather stations having statistically significant total annual (a) total summer (b) and total winter (c) precipitation trends in Mexico. Points indicate the position of stations having an increasing precipitation trend while open circles indicate the position of those stations having a declining precipitation trend. Symbols inside the maps are proportional to the magnitude of the trends displayed in the bottom left corner scale of the upper map. The mean trends, derived from the total number of stations overall Mexico, are shown outside the maps, while the mean trends, derived from the stations over each domain, are displayed inside the polygon of each domain (in mm/yr). The ± values define the 95% confidence intervals for the trends. Lines inside the maps show the continental divides with the Pacific drainage areas to the left and Gulf of Mexico drainage areas to the right. Abbreviations in upper left map are the same as in Figure 5a. LSB= Lerma Santiago basin.

To clarify the internal structure of the series, Figures 8–10 show observed time series (expressed as z scores) of selected variables having general significant upward and downward trends. A couple of series have been plotted from each selected station: (1) TDP >0 and (2) total precipitation. Total annual precipitation series are displayed in Figure 8 and total seasonal precipitation time series are displayed in Figures 9 and 10. All series display large inter-annual variability with substantial deviations from linearity in several years. The correspondence between TDP >0 and total annual precipitation trends (having both variables similar general upward or downward trend) is evident in Figure 8 for each station, indicating that an overall upward or downward precipitation trend matches a similar upward or downward trend in TDP >0 (i.e., the decline in total annual precipitation corresponds to a decline in the number of rainy days and vice versa). What is the cause of the overall increasing rainy (dry) days over the N and C domains? It is clear that, if a change of TDP >0 is underway, then a change in annual rainfall accumulation in the same direction



Winter (Nov-April)

**c)**


0.9±2.1

0.7±2.0

Annual (May-April)

Overall trend = -2.38±1.31 mm/yr


**SE-domain**

**N-domain**

SMS

SMOr

AP


2.5±1.4

Baja California Peninsula

> -1 mm × yr -28 mm × yr 1 mm × yr 46 mm × yr

2.4±0.3

Santiago basin.


**C-domain**

TMVB

LSB

SMO


Summer (May-Oct)


1.9±3.9

Fig. 7. Maps displaying the distribution of weather stations having statistically significant total annual (a) total summer (b) and total winter (c) precipitation trends in Mexico. Points indicate the position of stations having an increasing precipitation trend while open circles indicate the position of those stations having a declining precipitation trend. Symbols inside the maps are proportional to the magnitude of the trends displayed in the bottom left corner scale of the upper map. The mean trends, derived from the total number of stations overall Mexico, are shown outside the maps, while the mean trends, derived from the stations over each domain, are displayed inside the polygon of each domain (in mm/yr). The ± values define the 95% confidence intervals for the trends. Lines inside the maps show the

continental divides with the Pacific drainage areas to the left and Gulf of Mexico drainage areas to the right. Abbreviations in upper left map are the same as in Figure 5a. LSB= Lerma

To clarify the internal structure of the series, Figures 8–10 show observed time series (expressed as z scores) of selected variables having general significant upward and downward trends. A couple of series have been plotted from each selected station: (1) TDP >0 and (2) total precipitation. Total annual precipitation series are displayed in Figure 8 and total seasonal precipitation time series are displayed in Figures 9 and 10. All series display large inter-annual variability with substantial deviations from linearity in several years. The correspondence between TDP >0 and total annual precipitation trends (having both variables similar general upward or downward trend) is evident in Figure 8 for each station, indicating that an overall upward or downward precipitation trend matches a similar upward or downward trend in TDP >0 (i.e., the decline in total annual precipitation corresponds to a decline in the number of rainy days and vice versa). What is the cause of the overall increasing rainy (dry) days over the N and C domains? It is clear that, if a change of TDP >0 is underway, then a change in annual rainfall accumulation in the same direction


**b)**

**a)**

**PY-domain**

as TDP >0 changes is highly possible and the occurrence of heavy rainfall (from tropical cyclones, for example) only temporarily changes the total annual precipitation without affecting the overall trend.

Fig. 8. Examples of observed time series of selected variables (expressed as z scores) displaying upward (left series) and downward (right series) total annual precipitation trends. A couple of series have been plotted from the same selected station, i.e., an upper – TDP >0 (total days with rainfall > 0 mm; lines with ×s) series, and lower TPS-total precipitation series (lines with •). For each variable, the magnitude of the trend with the 95% confidence intervals with their respective units is shown. The map to the right displays the position of the selected stations corresponding to the variables with upward (points) and downward (circles) trends.

Regional Pattern of Trends in Long-Term Precipitation and



h)

g)

TDP>0 series

TPS


i)


j)


k)


l)


that for the particular time series the trend is not statistically significant.






Trend: 3.02 ± 2.15 mm/yr

Trend: 1.63 ± 1.24 mm/yr

Trend: 0.46 ± 0.40 mm/yr

Trend: 0.92 ± 0.8 mm/yr

Trend: 1.65 ± 1.22 mm/yr

Trend: 1.83 ± 1.4 mm/yr


b)

a)


c)


d)


e)


f)


No Trend

No Trend

Trend: 0.44 ± 0.16 days/yr

Trend: 0.47 ± 0.15 days/yr

No Trend

No Trend

Stream Flow Observations: Singularities in a Changing Climate in Mexico 405

Trend: -1.31 ± 0.85 mm/yr

Trend: -2.11 ± 1.51 mm/yr

Trend: -2.46 ± 2.18 mm/yr

Trend: -2.41 ± 2.28 mm/yr

Trend: -6.76 ± 6.56 mm/yr

Trend: -2.24 ± 1.92 mm/yr

Fig. 10. The same as Figure 8, but for total winter precipitation series. No Trend indicates

This conclusion is not completely valid for total summer (Figure 9) and total winter (Figure 10) precipitation variables because some stations have a general upward or downward seasonal precipitation trend that does not match the corresponding behavior in TDP >0 variables for the same station. This occurs for example, in Figures 9a, 9b, & 9e. Note that the







No Trend

No Trend

Trend: -0.37 ± 0.23 days/yr

No Trend

Trend: -0.37 ± 0.24 days/yr

Trend: -0.64 ± 0.26 days/yr

Fig. 9. The same as Figure 8, but for total summer precipitation series. No Trend indicates that for the particular time series the trend is not statistically significant.


b)

a)


c)


d)


e)

No Trend


f)


No Trend

No Trend

Trend: 0.32 ± 0.13 days/yr

Trend: 0.27 ± 0.17 days/yr

Trend: 0.24 ± 0.08 days/yr

Trend: 5.84 ± 2.58 mm/yr

Trend: 0.74 ± 0.59 mm/yr

Trend: 6.81 ± 2.94 mm/yr

Trend: 8.17 ± 3.84 mm/yr

Trend: 1.26 ± 1.14 mm/yr

Trend: 7.05 ± 2.82 mm/yr


TPS

TDP>0 series


h)

g)


i)


j)


k)


l)


that for the particular time series the trend is not statistically significant.

Trend: -7.72 ± 5.2 mm/yr

Trend: -7.88 ± 4.72 mm/yr

Trend: -1.14 ± 0.21 days/yr

Trend: -7.77 ± 4.81 mm/yr

Trend: -0.25 ± 0.13 days/yr

Trend: -0.70 ± 0.18 days/yr

Trend: -4.28 ± 2.71 mm/yr

Trend: -3.53 ± 2.25 mm/yr

Trend: -7.37 ± 4.05 mm/yr

Fig. 9. The same as Figure 8, but for total summer precipitation series. No Trend indicates

Trend: -0.63 ± 0.27 days/yr

Trend: -0.82 ± 0.24 days/yr







Trend: -0.48 ± 0.26 days/yr






Fig. 10. The same as Figure 8, but for total winter precipitation series. No Trend indicates that for the particular time series the trend is not statistically significant.

This conclusion is not completely valid for total summer (Figure 9) and total winter (Figure 10) precipitation variables because some stations have a general upward or downward seasonal precipitation trend that does not match the corresponding behavior in TDP >0 variables for the same station. This occurs for example, in Figures 9a, 9b, & 9e. Note that the

Regional Pattern of Trends in Long-Term Precipitation and

Rio Lerma-Santiago basin

Rio Lerma-Santiago basin

Stream Flow Observations: Singularities in a Changing Climate in Mexico 407

Black: linear increase White: linear decline

Rio Lerma-Santiago basin **c)**

**a)**

Black: linear increase White: linear decline

Black: linear increase White: linear decline

Fig. 11. Maps showing the distribution of river basin polygons displaying significant upward (black polygons) and downward (white polygons) streamflow trends for mean

The distribution of river basins having significant upward and downward mean annual and mean seasonal streamflow trends is shown in Figure 11. A tendency for a large number of drainage areas to have downward annual and seasonal trends is evident in basins located in the C domain, particularly those in the Lerma-Santiago river basin in central Mexico. This result matches downward trends observed in total annual and seasonal precipitation over the C domain, indicating that overall decreases in rainfall over the C domain have resulted in lower values for streamflow, mainly over Pacific drainage areas. Small basins surrounding the Lerma-Santiago river basin display upward trends in mean annual and seasonal streamflow. Although in the C domain, there are also weather stations displaying significant upward annual and summer trends (See Figures 7a & b), the position of those stations does not necessarily correspond to the position of the small basins. Moreover, mean winter streamflow in some basins display upward trends (Figure 11c), while no weather station displays similar significant upward trends over those areas (Figure 7c). These results suggest that the significant annual and seasonal upward trends in precipitation and streamflow over the C domain might be a reflection of local rather than regional factors. In addition it may be also due to there are 1161 for precipitation and only 141 for streamflow data sets. Not enough streamflow datasets and the large dataset ratio for precipitation to streamflow may cause some errors in the analysis. Over northwestern Mexico in the N domain, large drainage areas display upward and downward mean annual and mean seasonal trends with no obvious regionally coherent pattern (Figures 11a-c). Although we established that the middle-to-high elevations in northwestern Mexico in the N domain have upward mean annual and mean seasonal precipitation trends (Figures 7a-c), the decreasing trends observed in streamflow over these areas do not match this pattern.

annual (a) mean summer (b) and mean winter (c) components.

**b)**

upper TDP >0 time series in each case does not display statistically significant trends (indicated in Figure 9 by the legend "No Trend"). This is in contrast with the upward trend in the lower total summer precipitation series. In Figures 9a & 9b, a group of years with high precipitation values occurred in recent years, while the corresponding TDP >0 point values show a declining trend in the last two decades, an indication that some external source of humidity has provided the complementary rainfalls for the high summer values recorded in these stations (i.e., less summer rainy days correspond to higher summer rainfalls). This source of precipitation might derive from tropical cyclones, as suggested by Arriaga-Ramírez & Cavazos (2010).

The situation is a little different in winter, where the main source of precipitation proceeds from the passage of cold fronts, enhanced by the southern movement of the subtropical westerly jet stream, which is sometimes associated with El Niño events (Magaña et al., 2003). It is established that there is a direct link between warm SSTs during El Niño years and high winter precipitation in Mexico and vice versa (a cold SSTs during La Niña years and low winter precipitation). In this sense, the high values of winter precipitation during the 1990s, evident in some series (Figures 10a, b, e & f) that display an upward total winter precipitation trend and no trend at all in corresponding the TDP >0 time series might be a reflection of the strong-to-moderate El Niño years of 1991, 1994, 1997, and 2002 that provided necessary atmospheric conditions for production of the complementary winter precipitations in the datasets, since the corresponding TDP >0 time series decreases during the same period. The situation is a little harder to explain in series displaying a decreasing total winter precipitation trend (Figures 10g, h & j), and no trend at all in corresponding TDP >0 time series. The low winter precipitation observed in recent years in these time series might have been influenced by the strong-to-moderate La Niña conditions of 1988, 1998, 1999, and 2007 enhancing the overall decreasing trend, i.e., similar number of rainy days experienced lower precipitation values in recent years.

## **8. Annual and seasonal regional trends observed in streamflow series**

## **8.1 Hypothesis testing**

The results of hypothesis testing (not shown) indicate that changes in variances of all datasets are more common in winter than in summer or annual components. However, a significant change in means or variances of the two halves of the series do not necessarily result in a significant general trend for the series since the number of stations with significant changes in means or variances are always larger than those stations displaying a significant trend. This conclusion is also valid for each domain. On the contrary, similar means or similar variances inside the series do not necessarily mean that the corresponding series does not display a significant overall trend.

### **8.2 Significant trends in streamflow**

Mean winter streamflow show a larger number of stations displaying significant trends than mean summer or mean annual components. Upward trends in streamflow are more common in winter, while downward trends are more common in summer, than the other components.

upper TDP >0 time series in each case does not display statistically significant trends (indicated in Figure 9 by the legend "No Trend"). This is in contrast with the upward trend in the lower total summer precipitation series. In Figures 9a & 9b, a group of years with high precipitation values occurred in recent years, while the corresponding TDP >0 point values show a declining trend in the last two decades, an indication that some external source of humidity has provided the complementary rainfalls for the high summer values recorded in these stations (i.e., less summer rainy days correspond to higher summer rainfalls). This source of precipitation might derive from tropical cyclones, as suggested by Arriaga-

The situation is a little different in winter, where the main source of precipitation proceeds from the passage of cold fronts, enhanced by the southern movement of the subtropical westerly jet stream, which is sometimes associated with El Niño events (Magaña et al., 2003). It is established that there is a direct link between warm SSTs during El Niño years and high winter precipitation in Mexico and vice versa (a cold SSTs during La Niña years and low winter precipitation). In this sense, the high values of winter precipitation during the 1990s, evident in some series (Figures 10a, b, e & f) that display an upward total winter precipitation trend and no trend at all in corresponding the TDP >0 time series might be a reflection of the strong-to-moderate El Niño years of 1991, 1994, 1997, and 2002 that provided necessary atmospheric conditions for production of the complementary winter precipitations in the datasets, since the corresponding TDP >0 time series decreases during the same period. The situation is a little harder to explain in series displaying a decreasing total winter precipitation trend (Figures 10g, h & j), and no trend at all in corresponding TDP >0 time series. The low winter precipitation observed in recent years in these time series might have been influenced by the strong-to-moderate La Niña conditions of 1988, 1998, 1999, and 2007 enhancing the overall decreasing trend, i.e., similar number of rainy

days experienced lower precipitation values in recent years.

series does not display a significant overall trend.

**8.2 Significant trends in streamflow** 

**8. Annual and seasonal regional trends observed in streamflow series** 

The results of hypothesis testing (not shown) indicate that changes in variances of all datasets are more common in winter than in summer or annual components. However, a significant change in means or variances of the two halves of the series do not necessarily result in a significant general trend for the series since the number of stations with significant changes in means or variances are always larger than those stations displaying a significant trend. This conclusion is also valid for each domain. On the contrary, similar means or similar variances inside the series do not necessarily mean that the corresponding

Mean winter streamflow show a larger number of stations displaying significant trends than mean summer or mean annual components. Upward trends in streamflow are more common in winter, while downward trends are more common in summer, than the other

Ramírez & Cavazos (2010).

**8.1 Hypothesis testing** 

components.

Fig. 11. Maps showing the distribution of river basin polygons displaying significant upward (black polygons) and downward (white polygons) streamflow trends for mean annual (a) mean summer (b) and mean winter (c) components.

The distribution of river basins having significant upward and downward mean annual and mean seasonal streamflow trends is shown in Figure 11. A tendency for a large number of drainage areas to have downward annual and seasonal trends is evident in basins located in the C domain, particularly those in the Lerma-Santiago river basin in central Mexico. This result matches downward trends observed in total annual and seasonal precipitation over the C domain, indicating that overall decreases in rainfall over the C domain have resulted in lower values for streamflow, mainly over Pacific drainage areas. Small basins surrounding the Lerma-Santiago river basin display upward trends in mean annual and seasonal streamflow. Although in the C domain, there are also weather stations displaying significant upward annual and summer trends (See Figures 7a & b), the position of those stations does not necessarily correspond to the position of the small basins. Moreover, mean winter streamflow in some basins display upward trends (Figure 11c), while no weather station displays similar significant upward trends over those areas (Figure 7c). These results suggest that the significant annual and seasonal upward trends in precipitation and streamflow over the C domain might be a reflection of local rather than regional factors. In addition it may be also due to there are 1161 for precipitation and only 141 for streamflow data sets. Not enough streamflow datasets and the large dataset ratio for precipitation to streamflow may cause some errors in the analysis. Over northwestern Mexico in the N domain, large drainage areas display upward and downward mean annual and mean seasonal trends with no obvious regionally coherent pattern (Figures 11a-c). Although we established that the middle-to-high elevations in northwestern Mexico in the N domain have upward mean annual and mean seasonal precipitation trends (Figures 7a-c), the decreasing trends observed in streamflow over these areas do not match this pattern.

Regional Pattern of Trends in Long-Term Precipitation and

streamflow and precipitation changes in northwestern Mexico.

necessary to prove these.

**10. Acknowledgments** 

**11. References** 

Stream Flow Observations: Singularities in a Changing Climate in Mexico 409

match the mean annual and seasonal trends in streamflow. However, we also see that trend structures exhibit some spatial patterns that are not readily explained, such as the trend for declining streamflow in the N domain over large drainage areas of the western mainland and increasing streamflow in the C domain that does not match the opposite behavior of the trends observed in precipitation. The possibility that these differences might be related to more localized external factor and errors resulted from not enough streamflow data sets and the large dataset ratio for precipitation are suggested. However, more detailed analyses are

Finally, we conclude that over the last nine decades, Mexico has experienced an overall decrease in mean total annual and summer precipitation, based on analysis of 1161 weather stations across Mexico. Mean total precipitation changes were derived from 172 annual precipitation series and 161 summer rainfall series, displaying significant trends in both directions, declining trends for 103 annual and 87 summer series and increasing trends for 69 annual and 74 summer series. The total change from 1920–2008 has been about 21 mm in mean annual accumulation and 9 mm in mean summer accumulation. The most affected areas from declining precipitation are the middle-to-high elevations in the Gulf of Mexico drainage areas in eastern Mexico and over the upper Rio Lerma-Santiago basin in central Mexico. Several stations in northern and northwestern Mexico (including the Baja California Peninsula), have an overall increase in mean annual and mean summer rainfall accumulations with large variations over the areas. Although the number of stations having declines in mean total winter precipitation is larger than those having increases in mean total winter precipitation, the uncertainties among the trends are so large that the double mean square trend error overpasses the mean magnitude of the trends. In essence, the data does not provide evidence for regional mean changes in winter. The analysis of 141 streamflow series provided more evidence for declines in total annual and summer precipitation, mainly in the Río Lerma-Santiago Basin, although some river basins in northwestern Mexico also display declining trends that do not match the upward trend observed in precipitation. Unfortunately, no weather stations with long datasets are available there, mainly in the highlands of those basins. This makes it difficult to relate

This work was carried out with the support of the National Council on Science and Technology of Mexico (CONACYT grant J50757-F) and of the Centro de Investigaciones

Arriaga-Ramírez, S. & Cavazos, T. (2010). Regional trends of daily precipitation indices in

BANDAS (Banco Nacional de Datos de Aguas Superficiales). (1997). Instituto Mexicano de

northwest Mexico and southwest United States. *Journal of Geophysical Research*, Vol.

Tecnología del Agua, Período 1997-2004, CD-rom data base, Morelos, México.

Biologicas del Noroeste (CIBNOR grant PC 0.3). Ira Fogel provided editorial services.

115, D14111, doi:10.1029/2009JD013248, ISSN 0148-0227

Available from:<http://www.imta.gob.mx>

Unfortunately, over the drainage areas where significant downward streamflow trends are evident in northwestern Mexico, there are no weather stations (See Figure 5). This makes it impossible to prove a correspondence between precipitation and streamflow variables in western part of the mainland in the N domain.

Finally, some selected time series with significant upward and downward mean annual streamflow trends are displayed in Figure 12. As occurred with total precipitation series, streamflow series also show high inter-annual variability, with substantial deviations from the linear trend. From the time series, it is evident that the dry period during the 1950s (Figures 12a, b, e, i, j & l), and the dry period during the 1990s (Figure 12a, c, h & i) affected a large part of Mexico.

Fig. 12. Examples of observed time series (expressed as z scores) of selected gauging stations displaying upward (left series) and downward (right series) mean annual streamflow trends (lines). The magnitude of each trend with 95% confidence intervals is shown for each variable.

#### **9. Conclusions**

Regional changes in rainfall and streamflow in Mexico have significant trends of different sign (Figures 7 & 11), both declining and increasing trends. While isolated differences in trends occur in rainfall and streamflow over large areas, some basic coherent patterns can be identified, such as the trend for lower precipitation in the C domain and a trend for higher precipitation in the N domain. In the C domain, decreasing annual and seasonal trends match the mean annual and seasonal trends in streamflow. However, we also see that trend structures exhibit some spatial patterns that are not readily explained, such as the trend for declining streamflow in the N domain over large drainage areas of the western mainland and increasing streamflow in the C domain that does not match the opposite behavior of the trends observed in precipitation. The possibility that these differences might be related to more localized external factor and errors resulted from not enough streamflow data sets and the large dataset ratio for precipitation are suggested. However, more detailed analyses are necessary to prove these.

Finally, we conclude that over the last nine decades, Mexico has experienced an overall decrease in mean total annual and summer precipitation, based on analysis of 1161 weather stations across Mexico. Mean total precipitation changes were derived from 172 annual precipitation series and 161 summer rainfall series, displaying significant trends in both directions, declining trends for 103 annual and 87 summer series and increasing trends for 69 annual and 74 summer series. The total change from 1920–2008 has been about 21 mm in mean annual accumulation and 9 mm in mean summer accumulation. The most affected areas from declining precipitation are the middle-to-high elevations in the Gulf of Mexico drainage areas in eastern Mexico and over the upper Rio Lerma-Santiago basin in central Mexico. Several stations in northern and northwestern Mexico (including the Baja California Peninsula), have an overall increase in mean annual and mean summer rainfall accumulations with large variations over the areas. Although the number of stations having declines in mean total winter precipitation is larger than those having increases in mean total winter precipitation, the uncertainties among the trends are so large that the double mean square trend error overpasses the mean magnitude of the trends. In essence, the data does not provide evidence for regional mean changes in winter. The analysis of 141 streamflow series provided more evidence for declines in total annual and summer precipitation, mainly in the Río Lerma-Santiago Basin, although some river basins in northwestern Mexico also display declining trends that do not match the upward trend observed in precipitation. Unfortunately, no weather stations with long datasets are available there, mainly in the highlands of those basins. This makes it difficult to relate streamflow and precipitation changes in northwestern Mexico.

## **10. Acknowledgments**

This work was carried out with the support of the National Council on Science and Technology of Mexico (CONACYT grant J50757-F) and of the Centro de Investigaciones Biologicas del Noroeste (CIBNOR grant PC 0.3). Ira Fogel provided editorial services.

## **11. References**

408 Greenhouse Gases – Emission, Measurement and Management

Unfortunately, over the drainage areas where significant downward streamflow trends are evident in northwestern Mexico, there are no weather stations (See Figure 5). This makes it impossible to prove a correspondence between precipitation and streamflow variables in

Finally, some selected time series with significant upward and downward mean annual streamflow trends are displayed in Figure 12. As occurred with total precipitation series, streamflow series also show high inter-annual variability, with substantial deviations from the linear trend. From the time series, it is evident that the dry period during the 1950s (Figures 12a, b, e, i, j & l), and the dry period during the 1990s (Figure 12a, c, h & i) affected a

western part of the mainland in the N domain.

Trend: 0.76 ± 0.57 (m³/s)/yr

Trend: 0.82 ± 0.60 (m³/s)/yr

Trend: 0.02 ± 0.01 (m³/s)/yr

Trend: 0.07 ± 0.05 (m³/s)/yr

Trend: 0.03 ± 0.01 (m³/s)/yr

Trend: 0.08 ± 0.04 (m³/s)/yr

z scores


Fig. 12. Examples of observed time series (expressed as z scores) of selected gauging stations displaying upward (left series) and downward (right series) mean annual streamflow trends (lines). The magnitude of each trend with 95% confidence intervals is shown for each

Regional changes in rainfall and streamflow in Mexico have significant trends of different sign (Figures 7 & 11), both declining and increasing trends. While isolated differences in trends occur in rainfall and streamflow over large areas, some basic coherent patterns can be identified, such as the trend for lower precipitation in the C domain and a trend for higher precipitation in the N domain. In the C domain, decreasing annual and seasonal trends



g)

h)

i)

j)

k)

l)

Trend: -0.11 ± 0.09 (m³/s)/yr

Trend: -0.08 ± 0.03 (m³/s)/yr

Trend: -0.02 ± 0.01 (m³/s)/yr

Trend: -0.25 ± 0.21 (m³/s)/yr

Trend: -0.31 ± 0.15 (m³/s)/yr

Trend: -0.18 ± 0.15 (m³/s)/yr




z scores



z scores

large part of Mexico.

a)

b)

c)

d)

e)

f)


variable.

**9. Conclusions** 




Regional Pattern of Trends in Long-Term Precipitation and

http://www.ipcc.ch/press/index.htm

No. 7+8, pp. 937-950, ISSN: 0958-305X

pp. 1175-1188, ISSN 0894-8755

347-350, ISSN: 0028-0836

1329, ISSN: 1097-0088

0836

0894-8755

0016-7169

Stream Flow Observations: Singularities in a Changing Climate in Mexico 411

Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K.

IPCC (2007). Intergovernmental Panel on Climate Change. In *Climate Change 2007: The* 

Jáuregui, E. (2003). Climatology of landfalling hurricanes and tropical storms in Mexico. *Atmosfera*, Vol. 16, No. 4, (October 2003), pp. 193-204, ISSN 0187-6236 Knutson, Th.R., McBride, J.L Greenhouse Gases / Book 1., Chan, J., Emanuel, K., Hollan, G.,

Lindzen, R.S. (2007). Taking greenhouse warming seriously. *Energy & Environment*, Vol. 18,

Magaña, V., Amador, J. & Medina, S (1999). The midsummer drought over Mexico and

Magaña, V.O., Vázquez, J.L, Pérez, J.L. & Pérez, J.B. (2003). Impact of El Niño on

Matías, M. & Magaña, V. (2010). Regional aspects of prolonged meteorological droughts

McBride, J.L. (1995). *Tropical cyclone formation, global perspective on tropical cyclones*. Elsberry,

Mendoza, B. & Velasco, V. (2006). A study of historical droughts in southeastern Mexico. *Journal of Climate*, Vol. 19, No. 12 (June 2006), pp. 2916-2934, ISSN 0894-8755 Milly, P.C.D., Dunne, K.A & Vecchia, A.V. (2005). Global pattern of trends in streamflow

Pavia, E.G., Graef F. & Reyes, J. (2009). Annual and seasonal surface air temperature trends

Pavia, E.G., Graef, F. & Reyes, J. (2006). PDO-ENSO effects in the climate of Mexico. *Journal of Climate*, Vol. 19, pp. 6433-6438, doi:10.1175/JCL14045.1, ISSN 0894-8755 Quintas, I. (2000). ERIC II. In: *Documentación de la base de datos climatológica y del programa* 

*program)*. Instituto Mexicano de Tecnología del Agua (IMTA). 54 pp. Seager, R., Tign, M., Held, I., Kushnir, Y., Lu, J., Vecchi, G., Huan, H., Harnik, N., Leetmaa,

(May 2007), pp. 1181-7784, ISSN 0036-8075

R.L. (ed.). World Meteorological Organization, pp. 106-197

University Press, ISBN: 9780521014953, Cambridge, U.K.

& Johnson, C.A. (2001). *Climate Change 2001: The Scientific Basis*, Cambridge

*physical science basis. Contribution of working group I to the fourth assessment report of the IPCC (AR4)*, Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Avery, K.B., Tignor, M. & Miller, H.L. (eds). Cambridge University Press, New York,

Landsea, Ch., Held, I., Kossin, J.P., Srivastava A.K. & Sugi, M. (2010). Tropical cyclones and climate change. *Nature*, Vol. 3, (March 2010), pp. 157-163, ISSN: 0028-

Central America. *Journal of Climate*, Vol. 12, No. 6, (June 1999), pp. 1577-1588, ISSN

precipitation in Mexico. *Geofísica Internacional*, Vol. 42, No. 3, pp. 313-330, ISSN:

over Mexico and Central America. *Journal of Climate*, Vol. 23, No. 5 (March 2010),

and water availability in a changing climate. *Nature*, Vol. 438, (November 2005), pp.

in Mexico. *International Journal of Climatology,* Vol. 29, No. 9, (July 2009), pp. 1324-

*extractor (ERIC II: Documentation of the climatologic database anda data extraction* 

A., Lau, N., Li, C. Velez, J. & Naik, N. (2007). Model projections of an imminent transition to a more arid climate in southwestern North America. *Science*, Vol. 316,


Barnett, T.P., Pierce, D.W., AchutaRao, K.M., Gleckler, P.J., Santer, B.D., Gregory, J.M. &

Brito-Castillo, L., Díaz, C.S.C. & Ulloa-Herrera, R.S. (2009). Observed tendencies in

Brito-Castillo, L., Douglas, A.V., Leyva-Contreras, A. & Lluch-Belda, D. (2003). The effect of

Brito-Castillo, L., Leyva-Contreras, A., Douglas, A.V. & Lluch-Belda, D. (2002). Pacific

Díaz, S.C., Salinas-Zavala, C.A. & Hernández-Vázquez, S. (2008). Variability of rainfall from

Douglas, A.V. & Englehart, Ph. J. (2007). A Climatological perspective of transient synoptic

Englehart, Ph.J. & Douglas, A.V. (2004). Characterizing regional-scale variations in monthly

*Forecasting*, Vol. 9, No. 4, (December 1994), pp. 630-639, ISSN: 0882-8156 Farfan, L.M. (2011). Eastern Pacific tropical cyclones and their impact over western Mexico.

Garcia, E. (2004). *Modificaciones al sistema de clasificación climática de Koppen (para adaptarlo a* 

Giovannettone, J.P. & Barros, A.P. (2008). A remote sensing survey of the role of landform

Gutierrez-Ruacho, O.G., Brito-Castillo, L., Díaz, C.S.C. & Watts Ch. (2010). Trends in rainfall

Higgins, R.W., Douglas, A.V., Hahmann, A., Berbery, E.H., Gutzler, D., Shuttleworth, J.,

7\_9. ISBN: 978-3-642-17957-0, Springer-Verlag Berlin Heidelberg

*Atmosfera*, Vol. 21, No. 2, (April 2008), pp. 213-223, ISSN 0187-6236

oceans. *Science*, Vol. 309, (July 2005), pp. 284-287, ISSN 0036-8075

ISSN: 1097-0088

136, ISSN 0187-6236

1954, ISSN 0894-8755

ISBN: 970-32-1010-4, México, D.F.

(July 2010),pp. 133-142, ISSN: 1616-1572

16, No. 1, (January 2003), pp. 29-65, ISSN 0187-6236

1525-7541

2003), pp. 751-768, ISSN: 1097-0088

Washington, W.M. (2005). Penetration of human-induced warming into the world´s

maximum and minimum temperatures in Zacatecas, Mexico and possible causes. *International Journal of Climatology,* Vol. 29, No. 2, (February 2009), pp. 211-221,

large-scale circulation on precipitation and streamflow in the Gulf of California continental watershed. *International Journal of Climatology*, Vol. 23, No. 7, (June

Decadal Oscillation and the filled capacity of dams on the rivers of the Gulf of California continental watershed. *Atmosfera*, Vol. 15, No. 2, (April 2002), pp. 121-

tropical cyclones in northwestern Mexico and its relation to SOI and PDO.

features during NAME 2004. *Journal of Climate*, Vol. 20, No. 9, (May 2007), pp. 1947-

and seasonal surface air temperature over Mexico. *International Journal of Climatology,* Vol. 24, No. 15, (December 2004), pp. 1897-1090, ISSN: 1097-0088 Farfan, L.M. & Zehnder J.A. (1994). Moving and stationary mesoscale convective systems

over Northwest Mexico during the southwest area monsoon project. *Weather and* 

In: *Experimental and Theoretical Advances in Fluid Dynamics*, Klapp et al. (eds.), Environmental Science and Engineering, pp. 135-148, doi: 10.1007/978-3-642-17958-

*las condiciones climáticas de México)*. 5ta edición, Instituto de Geografía, UNAM,

on the organization of orographic precipitation in Central and southern Mexico. *Journal of Hydrometeorology*, Vol. 9, No. 6, (December 2008), pp. 1257-1283, ISSN:

and extreme temperatures in Northwestern Mexico. *Climate Research,* Vol. 42, No. 2,

Stensrud, D., Amador J., Carbone, R., Cortez, M., Douglas, M., Lobato, R., Meitin, J., Ropelewski, Ch., Schemm, J., Schubbert, S. & Zhang, Ch. (2003). Progress in Pan American CLIVAR research: the North American monsoon system. *Atmosfera*, Vol.


**19**

*Canada* 

Richard Larouche

*Children's Hospital of Eastern Ontario* 

**The Environmental and Population Health** 

In 2007, the prestigious medical journal *The Lancet* published a series of papers entitled *Energy and Health*. In his commentary, Richard Horton argued that "*The current debate about the impact of human beings on our planet – especially with respect to climate change – is one of the most important issues of our time. But that debate is presently unbalanced and too narrow*." He considered that the relationship between energy and health had been clearly neglected by governments and international institutions such as the World Health Organization and the World Bank. One of the proposed pathways to reduce our global footprint while improving our health is to switch from an automobile-dependent transportation system to a lowcarbon system characterized by an increased utilization of modes of transportation such as

On a worldwide basis, transport activities account for 23% of CO2 emissions and this proportion is steadily increasing (International Transport Forum, 2010). Moreover, worldwide CO2 emissions from transport have increased by 45% from 1990 to 2007 (International Transport Forum, 2010). In parallel with their rapid economic growth, transport-related CO2 emissions are increasing markedly in developing countries such as China and India (Woodcock et al., 2007). For example, in China, total motorized travel distance has increased almost five-folds between 1980 and 2003 (Zhou & Szyliowicz, 2006). Within the transport sector, private vehicles represent the second most important contributor to greenhouse gases (GHG) emissions behind road freight (Chapman, 2007). Existing road infrastructure cannot cope with the rapid increase in the number of vehicles; thereby generating traffic jams which, aside from the increased GHG emissions, have dire economic consequences. For example, Levy et al. (2010) estimated that fuel wasted and time lost in traffic jams in 83 American cities cost 60 billion US\$. Moreover, they evaluated that mortality associated with ambient exposure to fine particulate matter (PM2.5) cost an additional 31 billion US\$ in these cities. Although technological innovations may contribute to reduce the intensity of CO2 emissions from transport in a long term perspective, more immediate changes in transport behaviors are crucial to achieve a stabilization of the

walking, cycling and public transport (Horton, 2007; Woodcock et al., 2007).

absolute emissions in order to mitigate climate change (Chapman, 2007).

Data from the 1995 Nationwide Personal Transportation Survey in the US indicate that private cars were used for about 90% of trips of 0.5 to 3 miles, of which an important

**1. Introduction** 

**Benefits of Active Transport: A Review** 


## **The Environmental and Population Health Benefits of Active Transport: A Review**

Richard Larouche *Children's Hospital of Eastern Ontario Canada* 

## **1. Introduction**

412 Greenhouse Gases – Emission, Measurement and Management

Stahle, D.W., Cook, E.R., Villanueva-Diaz, J., Fye, F.K, Burnette, D.J., Griffin, R.D., Acuña-

Turrent, C. & Cavazos, T. (2009). Role of the land-sea thermal contrast in the interannual

Wang, C., Lee, S.K & Enfield, D.B. (2007). Impact of the Atlantic warm pool on the summer

Webster, P.J., Hollan, G.J., Curry, J.A. & Chang, H.-R. (2005). Changes in tropical cyclone

Weiss, J.L. & Overpeck, J.T. (2005). Is the Sonoran Desert losing its cool?. *Global Change Biology,* Vol. 11, No. 12, (December 2005), pp. 2065-2077., ISSN: 1365-2486 Wigley, T.M.L. (2006). Appendix A: Statistical Issues Regarding Trends. *In Temperature* 

Yu, J., Wang, Y. & Hamilton, K. (2010). Response of tropical cyclone potential intensity to al

Zhao, X. (2011). Is global warming mainly due to anthropogenic greenhouse gas emissions?.

2005), pp. 1753-1754, ISSN 0036-8075

2007), pp. 5021-5040, ISSN 0894-8755

DC. 165 pp.

L02808, doi:10.1029/2008GL036299, ISSN: 0094-8276

(September 2005), pp. 1844-1846, ISSN: 0036-8075

(March 2010), pp. 1354-1373, ISSN: 0894-8755

21, (August 2011), pp. 1985-1992, ISSN: 1556-7036

Soto, R., Seager, R. & Heim, Jr., R.R. (2009). Early 21st-century drought in Mexico, *Eos Translational*, AGU, Vol. 90, No. 11, pp. 89-100, doi:10.1029/2009EO110001 Trenberth, K. (2005). Uncertainty in Hurricane and global warming. *Science*, Vol. 308, (June

modulation of the North American Monsoon. *Geophysical Research Letters*, Vol. 36,

climate of the western hemisphere. *Journal of Climate*, Vol. 20, No. 20, (October

number, duration and intensity in a warming environment. *Science*, Vol. 309,

*Trends in the Lower Atmosphere: Steps for Understanding and Reconciling Differences.* TR Karl, SJ Hassol, CD Miller, and WL Murray (Eds.)*.* A Report by the Climate Change Science Program and the Subcommitttee on Global Change Research,Washington,

global warming scenario in the IPCC AR4 CGCMs. *Journal of Climate*, Vol. 23, No. 6,

*Energy Sources, Part A: Recovery, Utilization, and Environmental Effects*, Vol. 33, No.

In 2007, the prestigious medical journal *The Lancet* published a series of papers entitled *Energy and Health*. In his commentary, Richard Horton argued that "*The current debate about the impact of human beings on our planet – especially with respect to climate change – is one of the most important issues of our time. But that debate is presently unbalanced and too narrow*." He considered that the relationship between energy and health had been clearly neglected by governments and international institutions such as the World Health Organization and the World Bank. One of the proposed pathways to reduce our global footprint while improving our health is to switch from an automobile-dependent transportation system to a lowcarbon system characterized by an increased utilization of modes of transportation such as walking, cycling and public transport (Horton, 2007; Woodcock et al., 2007).

On a worldwide basis, transport activities account for 23% of CO2 emissions and this proportion is steadily increasing (International Transport Forum, 2010). Moreover, worldwide CO2 emissions from transport have increased by 45% from 1990 to 2007 (International Transport Forum, 2010). In parallel with their rapid economic growth, transport-related CO2 emissions are increasing markedly in developing countries such as China and India (Woodcock et al., 2007). For example, in China, total motorized travel distance has increased almost five-folds between 1980 and 2003 (Zhou & Szyliowicz, 2006).

Within the transport sector, private vehicles represent the second most important contributor to greenhouse gases (GHG) emissions behind road freight (Chapman, 2007). Existing road infrastructure cannot cope with the rapid increase in the number of vehicles; thereby generating traffic jams which, aside from the increased GHG emissions, have dire economic consequences. For example, Levy et al. (2010) estimated that fuel wasted and time lost in traffic jams in 83 American cities cost 60 billion US\$. Moreover, they evaluated that mortality associated with ambient exposure to fine particulate matter (PM2.5) cost an additional 31 billion US\$ in these cities. Although technological innovations may contribute to reduce the intensity of CO2 emissions from transport in a long term perspective, more immediate changes in transport behaviors are crucial to achieve a stabilization of the absolute emissions in order to mitigate climate change (Chapman, 2007).

Data from the 1995 Nationwide Personal Transportation Survey in the US indicate that private cars were used for about 90% of trips of 0.5 to 3 miles, of which an important

The Environmental and Population Health Benefits of Active Transport: A Review 415

"traffic", "physical activity", "physical fitness", "obes\*", "overweight", "diabetes", "cardiovascular disease\*", "cancer", "hypertension", "all-cause mortality", "correlates", "determinants". Other articles were added from the author's personal library. In addition, reference lists of previous reviews were scanned to identify potentially relevant literature. Articles were selected on the basis that they provide additional elements to the comprehension of the explored phenomena. To be included in the review, articles generally needed to have examined AT, either as an outcome variable or as an exposure. However, in some cases, studies that examined the association between environments recognized as conductive to AT were also considered eligible. There were no restriction with respect to date of publication; however, most included studies were published during the last 10 years.

For the section on the impact of transport practices on pollution and GHG, priority was given to the natural experiments in which the impact of transport policies was rigorously examined before and after the intervention. To complement this rather scarce literature, simulations and cross-sectional studies were also included. For the section on the healthrelated outcomes of AT, priority was given to the studies that used an experimental design, which provide the highest level of evidence. These studies were complemented by prospective cohort studies, large cross-sectional studies and systematic reviews. Finally, as a systematic examination of the correlates of AT was beyond the scope of this book chapter, a simplified ecological model was used to categorize the most important factors into five

Personal

Fig. 1. Simplified ecological model for categorizing the correlates of active transport into the

In central London, a congestion charging scheme was introduced by the mayor in February 2003. Drivers are charged £8 (about US\$16) to enter the downtown core during working hours. While the scheme had little impact on the concentrations of CO2 and PM10 in Greater

**3. Impact of transport practices on pollution and greenhouse gases** 

Interpersonal

Communities / Organizations

Public policies

Built / Physical environment

Studies published in English or French were considered eligible.

levels of influence, as illustrated in Figure 1.

different levels of influence.

**3.1 Natural experiments** 

proportion could be substituted by walking, cycling and public transport (de Nazelle et al., 2010). This would represent an interesting strategy to reduce GHG emissions considering the disproportionate contribution of short trips to pollutant emissions due notably to cold starts (Frank et al., 2000). Further, since most individuals must travel to work or to school on a daily basis, active transport (e.g. using active modes of transportation such as walking and cycling) is increasingly regarded as a promising alternative to increase physical activity (PA) levels (Larouche & Trudeau, 2010; Shephard, 2008).

From a public health perspective, there is clear evidence of a strong relationship between PA and several health-related illnesses including cardiovascular diseases (Manson et al., 2002), diabetes (Sigal et al., 2006), cancer (Friedenreich & Orenstein, 2002), and all-cause mortality (Paffenbarger et al., 1993). Moreover, the prevalence of obesity has increased throughout the world to such an extent that it is now considered a pandemic (Lobstein et al., 2004; Wang & Lobstein, 2006). Recent research has also shown that the majority of the population fails to accumulate enough daily PA (Colley et al 2011a, b; Troiano et al., 2008). During the last decades, the rates of AT to/from school in children and adolescents has decreased markedly in several countries (Buliung et al, 2009, Grize et al., 2010; McDonald, 2007; van der Ploeg et al., 2008). Further, exposure to ambient air pollution is associated with respiratory and cardiovascular morbidity and mortality (Brook et al., 2010; Brunekreef & Holgate, 2002).

To implement more effective strategies to promote active transport (AT), it is necessary to develop a greater understanding of the factors associated with individuals' mode of transport, that is the correlates of AT1. Current theoretical models suggest that the choice of travel mode is based on several factors including characteristics of the individual (i.e. perceived barriers to active transport), of the social environment (i.e. social support), of public policies (i.e. laws and taxes) and of the built environment (i.e. population density and land-use mix) (Panter et al., 2008; Sallis et al., 2006). These models are typically based on the social ecological theory (i.e. Bronfenbrenner, 1979) which predicts that human behavior result from the interactions within and between these multiple "levels of influence".

Consequently, the current review is organized in four main sections. First, the impact of various transportation strategies to reduce emissions of GHG and other pollutants are described. Second, the impacts of AT on physical activity, physical fitness and other healthrelated outcomes are summarized. Third the correlates of AT are briefly examined to identify key variables to consider for the promotion of AT as a strategy to reduce GHG emissions and increase physical activity. Finally, the fourth section examines the interrelationships between the environmental and public health impacts of AT.

## **2. Methods**

Potentially relevant articles were identified through databases including Medline, PsychInfo, Google Scholar and the National Transportation Library. Keywords such as "active transport\*", "active commuting", "active travel" were used in addition to the following exposures or outcomes: "greenhouse gas\*", "carbon dioxide", "climate change",

<sup>1</sup> As detailed in the current review, most of the research that has examined the factors associated with AT have used a cross-sectional methodology. Therefore, the term "correlates" is used rather than the term "determinants" which involves a causal relationship (Bauman et al., 2002).

proportion could be substituted by walking, cycling and public transport (de Nazelle et al., 2010). This would represent an interesting strategy to reduce GHG emissions considering the disproportionate contribution of short trips to pollutant emissions due notably to cold starts (Frank et al., 2000). Further, since most individuals must travel to work or to school on a daily basis, active transport (e.g. using active modes of transportation such as walking and cycling) is increasingly regarded as a promising alternative to increase physical activity (PA)

From a public health perspective, there is clear evidence of a strong relationship between PA and several health-related illnesses including cardiovascular diseases (Manson et al., 2002), diabetes (Sigal et al., 2006), cancer (Friedenreich & Orenstein, 2002), and all-cause mortality (Paffenbarger et al., 1993). Moreover, the prevalence of obesity has increased throughout the world to such an extent that it is now considered a pandemic (Lobstein et al., 2004; Wang & Lobstein, 2006). Recent research has also shown that the majority of the population fails to accumulate enough daily PA (Colley et al 2011a, b; Troiano et al., 2008). During the last decades, the rates of AT to/from school in children and adolescents has decreased markedly in several countries (Buliung et al, 2009, Grize et al., 2010; McDonald, 2007; van der Ploeg et al., 2008). Further, exposure to ambient air pollution is associated with respiratory and cardiovascular morbidity and mortality (Brook et al., 2010; Brunekreef & Holgate, 2002).

To implement more effective strategies to promote active transport (AT), it is necessary to develop a greater understanding of the factors associated with individuals' mode of transport, that is the correlates of AT1. Current theoretical models suggest that the choice of travel mode is based on several factors including characteristics of the individual (i.e. perceived barriers to active transport), of the social environment (i.e. social support), of public policies (i.e. laws and taxes) and of the built environment (i.e. population density and land-use mix) (Panter et al., 2008; Sallis et al., 2006). These models are typically based on the social ecological theory (i.e. Bronfenbrenner, 1979) which predicts that human behavior

result from the interactions within and between these multiple "levels of influence".

relationships between the environmental and public health impacts of AT.

term "determinants" which involves a causal relationship (Bauman et al., 2002).

**2. Methods** 

Consequently, the current review is organized in four main sections. First, the impact of various transportation strategies to reduce emissions of GHG and other pollutants are described. Second, the impacts of AT on physical activity, physical fitness and other healthrelated outcomes are summarized. Third the correlates of AT are briefly examined to identify key variables to consider for the promotion of AT as a strategy to reduce GHG emissions and increase physical activity. Finally, the fourth section examines the inter-

Potentially relevant articles were identified through databases including Medline, PsychInfo, Google Scholar and the National Transportation Library. Keywords such as "active transport\*", "active commuting", "active travel" were used in addition to the following exposures or outcomes: "greenhouse gas\*", "carbon dioxide", "climate change",

1 As detailed in the current review, most of the research that has examined the factors associated with AT have used a cross-sectional methodology. Therefore, the term "correlates" is used rather than the

levels (Larouche & Trudeau, 2010; Shephard, 2008).

"traffic", "physical activity", "physical fitness", "obes\*", "overweight", "diabetes", "cardiovascular disease\*", "cancer", "hypertension", "all-cause mortality", "correlates", "determinants". Other articles were added from the author's personal library. In addition, reference lists of previous reviews were scanned to identify potentially relevant literature.

Articles were selected on the basis that they provide additional elements to the comprehension of the explored phenomena. To be included in the review, articles generally needed to have examined AT, either as an outcome variable or as an exposure. However, in some cases, studies that examined the association between environments recognized as conductive to AT were also considered eligible. There were no restriction with respect to date of publication; however, most included studies were published during the last 10 years. Studies published in English or French were considered eligible.

For the section on the impact of transport practices on pollution and GHG, priority was given to the natural experiments in which the impact of transport policies was rigorously examined before and after the intervention. To complement this rather scarce literature, simulations and cross-sectional studies were also included. For the section on the healthrelated outcomes of AT, priority was given to the studies that used an experimental design, which provide the highest level of evidence. These studies were complemented by prospective cohort studies, large cross-sectional studies and systematic reviews. Finally, as a systematic examination of the correlates of AT was beyond the scope of this book chapter, a simplified ecological model was used to categorize the most important factors into five levels of influence, as illustrated in Figure 1.

Fig. 1. Simplified ecological model for categorizing the correlates of active transport into the different levels of influence.

## **3. Impact of transport practices on pollution and greenhouse gases**

#### **3.1 Natural experiments**

In central London, a congestion charging scheme was introduced by the mayor in February 2003. Drivers are charged £8 (about US\$16) to enter the downtown core during working hours. While the scheme had little impact on the concentrations of CO2 and PM10 in Greater

The Environmental and Population Health Benefits of Active Transport: A Review 417

Similarly, American researchers examined the impact of five different education policy scenarios on vehicle emissions during children's school commuting in St-Paul, Minnesota using data from a school travel survey and a school enrollment data set (Marshall et al., 2010). In the first scenario, children where modeled as attending the school that they actually attended. The second scenario randomly assigned children to any school in the district. In the third scenario, children were assigned to the closest school in their neighborhood. In the fourth scenario, the district was subdivided into three regions and most children were assigned to a school in the district where they lived. Finally, the fifth scenario is similar to the first one, except that all children living within 1 mile of school were modeled as active commuters. Emissions of CO2, CO, NOx, PM10 and volatile organic compounds were predicted using the US Environmental Protection Agency's MOBILE6 model2. The third scenario would be the most effective, achieving 3- to 8-fold reductions in the pollutant emissions with a 7-fold reduction in transport costs. Further, because only 27% of children lived within 1 mile of their actual school, this scenario would achieve an even greater mode share of AT. However, this scenario would go against the habit of free choice;

Using data from the 1995 Nationwide Personal Transportation Survey, de Nazelle et al. (2010) estimated the impact of conversing short car trips (defined as < 3 miles) to alternative modes of transportation on pollutant emissions. Their estimates suggest that, across the US, daily emissions reductions would amount to 30-70 tons for volatile organic compounds, 15- 35 tons for nitrogen oxides, and 400-900 tons for carbon monoxide. Daily reductions in GHG emissions would vary between 20 000 to 46 000 tons of CO2 equivalent. This simulation is based on a very small decrease in vehicle miles travelled (0.8% to 1.8%), while it is estimated for example that a 10% increase in the price of gas can reduce consumption by 2.5% (Kaza et al., 2011). Thus, one could argue that there is a potential for greater reductions of driving.

Combining data from the Neighborhood Quality of Life Study and the King County Land Use, Transportation, Air Quality, and Health Study, Frank and colleagues (2006) examined the interactions between land use, pollution and health. Their analyses indicate that a 5% increase in the index of walkability3 was associated with decreases in vehicle miles travelled (6.5%), NOx emissions (5.6%) and volatile organic compounds (5.5%). Such an increase in walkability was also associated with lower body mass indices (-0.228 kg/m2) and with 32.1% more minutes of weekly AT. Similar analyses based on the SMARTRAQ study in the Atlanta area have shown that increased residential density, street connectivity and transit accessibility were associated with increased walking and decreased driving (Frank et al., 2010). In contrast, increasing the land-use mix was associated with reductions in both driving and walking, probably because individuals travel shorter distances in this type of

Further, a study from the US Environmental Protection Agency (1999) compared the impact of new residences in already built-up areas (e.g. infill development) vs. new residences built

3 Walkability refers to the ability to walk to certain destinations within the neighborhood. This concept

thus, it would likely be a challenge to gather public acceptability.

**3.3 Observational studies** 

environment (Frank et al., 2010).

2 This model can be found at www.epa.gov/otaq/mobile.htm

is further described in section 5.1 of this book chapter.

London, small decreases were observed in the central wards. Estimated years of life gained were 183 per 100 000 individuals. Interestingly, air pollution and mortality reductions were greater in more deprived areas, but emissions nevertheless remained higher in these areas compared to the more affluent areas (Tonne et al., 2008). A similar scheme was implemented in Stockholm and it resulted in a 15% decrease in vehicle miles travelled accompanied by 8.5-14% reductions of the emissions of nitric oxides, PM10 and CO2 in the inner city (Johansson et al., 2009). However, as in London, changes in emissions were lower in the Greater Stockholm (1-3%), suggesting that the impact of congestion charging schemes is mostly localized.

For the 1996 Summer Olympic Games in Atlanta, a transportation strategy was developed to minimize traffic congestion and to ensure that spectators could commute to the Olympic events in a reasonable amount of time. This strategy included the development of a public transportation system operating 24 hours a day, the addition of 1000 buses for park-and-ride services, the closure of the downtown core to private automobile travel, and modified delivery schedules (Friedman et al., 2001). Local businesses were also encouraged to implement alternative work schedules. Using an ecological study design, Friedman and colleagues (2001) examined the impact of these changes in traffic on air quality and childhood asthma events. They observed a 22.5% reduction in weekday peak traffic that was correlated (r = 0.36) with a 27.9% reduction in ozone concentration. Reductions in concentrations of carbon monoxide, PM10 and nitrogen dioxide were also noted. Lower rates of childhood asthma acute care events were also reported.

Similarly, a combination of aggressive measures was put in place in order to improve air quality for the 2008 Beijing Olympic Games. Private vehicles with odd-numbered license plates were allowed to circulate on Beijing's roads one day and those with even numbers were allowed on the next day, while vehicles that failed to meet the European No. 1 standard for exhaust emissions were banned. Power plants and large factories were also required to reduce their emissions (Wang et al., 2009). These measures have led to reductions in the concentration of sulfur oxide, carbon monoxide, nitric oxides and ozone of 61%, 25%, 21% and 26% respectively. It is however unclear whether such policies could be implemented at a larger scale and for longer periods of time.

#### **3.2 Simulations**

Woodcock and colleagues (2009) examined the impact of different transportation scenarios on carbon emissions and public health for the year 2030. Using the cities of London and Delhi as examples, they modeled the impact of five scenarios: 1) Business as usual; 2) Lower carbon-emission motor vehicles; 3) Increased active travel; 4) "Towards sustainable transport (e.g. a combination of scenarios 2 and 3); and 5) Increased active travel, but shorter distances travelled. In London, the scenario #1 would lead to a 4% increase in CO2 emissions from 1990 levels while scenarios 2 to 5 would lead to 35-60% reductions of emissions. In Delhi, due to the projected increases in population and in the level of motorization, the predicted CO2 emissions from all scenarios would represent major increases in comparison with 1990 levels. However, with scenario #4, the increase would be 199% compared to 526% for the business as usual scenario. The "Towards sustainable transport" scenario would have the greatest impact on mortality in London (7 439 disability-adjusted life-years) and Delhi (12 995 disability-adjusted life-years). Increasing the mode share of AT would have a greater impact than relying on technology (e.g. lower-emissions motor vehicles).

Similarly, American researchers examined the impact of five different education policy scenarios on vehicle emissions during children's school commuting in St-Paul, Minnesota using data from a school travel survey and a school enrollment data set (Marshall et al., 2010). In the first scenario, children where modeled as attending the school that they actually attended. The second scenario randomly assigned children to any school in the district. In the third scenario, children were assigned to the closest school in their neighborhood. In the fourth scenario, the district was subdivided into three regions and most children were assigned to a school in the district where they lived. Finally, the fifth scenario is similar to the first one, except that all children living within 1 mile of school were modeled as active commuters. Emissions of CO2, CO, NOx, PM10 and volatile organic compounds were predicted using the US Environmental Protection Agency's MOBILE6 model2. The third scenario would be the most effective, achieving 3- to 8-fold reductions in the pollutant emissions with a 7-fold reduction in transport costs. Further, because only 27% of children lived within 1 mile of their actual school, this scenario would achieve an even greater mode share of AT. However, this scenario would go against the habit of free choice; thus, it would likely be a challenge to gather public acceptability.

Using data from the 1995 Nationwide Personal Transportation Survey, de Nazelle et al. (2010) estimated the impact of conversing short car trips (defined as < 3 miles) to alternative modes of transportation on pollutant emissions. Their estimates suggest that, across the US, daily emissions reductions would amount to 30-70 tons for volatile organic compounds, 15- 35 tons for nitrogen oxides, and 400-900 tons for carbon monoxide. Daily reductions in GHG emissions would vary between 20 000 to 46 000 tons of CO2 equivalent. This simulation is based on a very small decrease in vehicle miles travelled (0.8% to 1.8%), while it is estimated for example that a 10% increase in the price of gas can reduce consumption by 2.5% (Kaza et al., 2011). Thus, one could argue that there is a potential for greater reductions of driving.

#### **3.3 Observational studies**

416 Greenhouse Gases – Emission, Measurement and Management

London, small decreases were observed in the central wards. Estimated years of life gained were 183 per 100 000 individuals. Interestingly, air pollution and mortality reductions were greater in more deprived areas, but emissions nevertheless remained higher in these areas compared to the more affluent areas (Tonne et al., 2008). A similar scheme was implemented in Stockholm and it resulted in a 15% decrease in vehicle miles travelled accompanied by 8.5-14% reductions of the emissions of nitric oxides, PM10 and CO2 in the inner city (Johansson et al., 2009). However, as in London, changes in emissions were lower in the Greater Stockholm (1-3%), suggesting that the impact of congestion charging schemes is

For the 1996 Summer Olympic Games in Atlanta, a transportation strategy was developed to minimize traffic congestion and to ensure that spectators could commute to the Olympic events in a reasonable amount of time. This strategy included the development of a public transportation system operating 24 hours a day, the addition of 1000 buses for park-and-ride services, the closure of the downtown core to private automobile travel, and modified delivery schedules (Friedman et al., 2001). Local businesses were also encouraged to implement alternative work schedules. Using an ecological study design, Friedman and colleagues (2001) examined the impact of these changes in traffic on air quality and childhood asthma events. They observed a 22.5% reduction in weekday peak traffic that was correlated (r = 0.36) with a 27.9% reduction in ozone concentration. Reductions in concentrations of carbon monoxide, PM10 and nitrogen dioxide were also noted. Lower rates

Similarly, a combination of aggressive measures was put in place in order to improve air quality for the 2008 Beijing Olympic Games. Private vehicles with odd-numbered license plates were allowed to circulate on Beijing's roads one day and those with even numbers were allowed on the next day, while vehicles that failed to meet the European No. 1 standard for exhaust emissions were banned. Power plants and large factories were also required to reduce their emissions (Wang et al., 2009). These measures have led to reductions in the concentration of sulfur oxide, carbon monoxide, nitric oxides and ozone of 61%, 25%, 21% and 26% respectively. It is however unclear whether such policies could be

Woodcock and colleagues (2009) examined the impact of different transportation scenarios on carbon emissions and public health for the year 2030. Using the cities of London and Delhi as examples, they modeled the impact of five scenarios: 1) Business as usual; 2) Lower carbon-emission motor vehicles; 3) Increased active travel; 4) "Towards sustainable transport (e.g. a combination of scenarios 2 and 3); and 5) Increased active travel, but shorter distances travelled. In London, the scenario #1 would lead to a 4% increase in CO2 emissions from 1990 levels while scenarios 2 to 5 would lead to 35-60% reductions of emissions. In Delhi, due to the projected increases in population and in the level of motorization, the predicted CO2 emissions from all scenarios would represent major increases in comparison with 1990 levels. However, with scenario #4, the increase would be 199% compared to 526% for the business as usual scenario. The "Towards sustainable transport" scenario would have the greatest impact on mortality in London (7 439 disability-adjusted life-years) and Delhi (12 995 disability-adjusted life-years). Increasing the mode share of AT would have a

greater impact than relying on technology (e.g. lower-emissions motor vehicles).

of childhood asthma acute care events were also reported.

implemented at a larger scale and for longer periods of time.

mostly localized.

**3.2 Simulations** 

Combining data from the Neighborhood Quality of Life Study and the King County Land Use, Transportation, Air Quality, and Health Study, Frank and colleagues (2006) examined the interactions between land use, pollution and health. Their analyses indicate that a 5% increase in the index of walkability3 was associated with decreases in vehicle miles travelled (6.5%), NOx emissions (5.6%) and volatile organic compounds (5.5%). Such an increase in walkability was also associated with lower body mass indices (-0.228 kg/m2) and with 32.1% more minutes of weekly AT. Similar analyses based on the SMARTRAQ study in the Atlanta area have shown that increased residential density, street connectivity and transit accessibility were associated with increased walking and decreased driving (Frank et al., 2010). In contrast, increasing the land-use mix was associated with reductions in both driving and walking, probably because individuals travel shorter distances in this type of environment (Frank et al., 2010).

Further, a study from the US Environmental Protection Agency (1999) compared the impact of new residences in already built-up areas (e.g. infill development) vs. new residences built

<sup>2</sup> This model can be found at www.epa.gov/otaq/mobile.htm

<sup>3</sup> Walkability refers to the ability to walk to certain destinations within the neighborhood. This concept is further described in section 5.1 of this book chapter.

The Environmental and Population Health Benefits of Active Transport: A Review 419

improvements were also noted for HDL cholesterol, blood lactate concentration during exercise and cardiovascular endurance in the AT group (Oja et al., 1991). In the second randomized controlled trial, 122 Dutch adults were assigned either to a PT or a cycling group for a 6 months period (Hendriksen et al., 2000). Maximal power output increased by 13% in both men and women, and VO2max increased by 6% in men while, in the PT group, VO2max decreased by 5-10%. In addition, despite no changes in body mass, there was a

In a Belgian non-randomized controlled trial, 65 middle-aged men and women were asked to cycle to work at least three times a week for a one year period, while the 15 participants from the control group were asked not to change their travel patterns (de Geus et al., 2009). Participants from the experimental group cycled 2.5 days per week for 45 minutes per day. Their maximal power output increased by 5%, but their VO2max decreased by 1%, while corresponding figures in control participants were 2% and -7%. Albeit small, differences between groups were significant for both maximal power output and VO2max, and individuals who cycled the most had greater improvements in fitness. As hypothesized by Oja et al. (2011), the smaller effect observed in this study compared to the 2 randomized controlled trials could be explained by lower cycling frequency and lower daily cycling

Andersen and colleagues (2000) assessed the relationship between cycling to work and allcause mortality among a cohort of 6 954 Danish adults. The relative risk of mortality was 0.72 (95% CI 0.57-0.91) among the cyclists, even after adjusting for physical activity, cholesterol and triglyceride levels, body mass index, blood pressure, education level, smoking status, gender, and age at the beginning of the study. Similarly, a prospective study of 75 221 women aged 40-70 from seven communities in Shanghai has shown lower rates of all-cause mortality after 5.7 years of follow-up. The relative risk of mortality was 0.66 (95% CI 0.40-1.07) among the women who cycled the most when adjusted for a similar range of

Gordon-Larsen and colleagues (2009) examined the prospective impact of AT to work on body composition, cardiovascular disease risk factors and cardiovascular fitness among 2364 participants from the Coronary Artery Risk Development in Young Adults that were followed for 20 years. When controlling for age, race, income, education, smoking, examination center and physical activity index, men who used AT to work had a reduced risk of obesity (OR = 0.50; 95% CI = 0.33-0.76), lower triglyceride levels (OR = 0.88; 99% CI = 0.80-0.98), fasting insulin (OR = 0.86; 95% CI = 0.78-0.93) and diastolic blood pressure (OR = -1.67; 95% CI = -3.20;-0.15). Furthermore, both men and women who used AT had greater

Hu and colleagues (2005) prospectively followed a cohort of 47 721 Finnish adults aged 25- 64 and without history of coronary heart disease, stroke or cancer at baseline. They found that compared to participants who did not used AT, those who used AT for 1-29 minutes per day (RR = 0.92; 95% CI = 0.84-1.01) and those who engaged in 30 minutes of more of daily AT (RR = 0.89; 95% CI = 0.80-0.98) were less likely to have suffered a stroke during the 19 years of follow-up. In the same cohort, AT to work was also associated with a reduced

cardiovascular fitness as measured by time to exhaustion on a treadmill test.

risk of coronary heart disease in women, but not in men (Hu et al., 2007).

reduction in the percentage of body fat in the cycling group (Hendriksen et al., 2000).

time.

**4.2 Prospective studies** 

potential confounders (Matthews et al., 2007).

in the suburbs and exurbs (e.g. greenfield development) in San Diego (CA), Montgomery County (MD) and West Palm Beach (FL). In the denser "infill" developments, vehicle miles travelled were 40-50% lower, emissions of CO2 and NOx were about 50% lower and public infrastructure costs were also markedly lower.

However, the concentrations of NOx can nevertheless be higher in more "walkable" areas, which are characterized by higher population density and land-use mix (Marshall et al., 2009), emphasizing the need to distinguish emissions and concentrations. For example, in Vancouver (Canada), a strong correlation was noted between neighborhoods walkability and NOx concentrations. High walkability areas were mostly located in the more densely populated city center through which residents of the suburbs are likely to commute to work. However, ozone is a secondary pollutant formed in the atmosphere rather than emitted directly from traffic exhaust, so high concentrations generally occur downwind from the high density areas; hence ozone concentrations in Vancouver where higher in the suburbs (Marshall et al., 2009).

Altogether, these studies suggest that the design of more walkable neighborhoods could lead to a decrease in the mode share of driving that should, in turn, lead to reductions of GHG emissions. However, because these studies have used a cross-sectional design, their findings must be interpreted with caution. There may be an underlying self-selection issue such that individuals choose to live in a neighborhood that meets certain preferences (Frank et al., 2007). To address this issue, Handy and colleagues (2005) compared the travel behaviors of residents who had moved during a 1-year period to others that did not while controlling for neighborhood preferences. They found that changes in neighborhood characteristics were associated with changes in walking and driving. In their model, accessibility of destinations was the strongest predictor of changes in driving, emphasizing the importance of land-use mix.

Another study explored the relationship between urban form and the ecological footprint associated with commuting patterns in the 163 municipalities of the Barcelona metropolitan region, Spain (Muniz & Galindo, 2005). Between 1986 and 1996, per capita ecological footprint increased by 95% driven by increases in the average trip distance (from 4.6 km to 6.7 km), and the mode share of car trips (22 to 35%). The mean trip distance and the ecological footprint of commuting increased systematically as distance from the city center increased reflecting higher car use. In addition, population density and accessibility of destination were negatively associated with ecological footprint.

## **4. Health-related outcomes of active transport**

## **4.1 Controlled trials**

A recent systematic review identified 2 randomized controlled trials in which individuals were randomly assigned either to an AT or a passive transport (PT) group (Oja et al., 2011). Specifically, in the first study, 68 Finnish adults who were classified as insufficiently active before the study were assigned either to a PT or an AT (walking or cycling) group. Individuals from the latter group were asked to engage in AT to work on every workday for 10 consecutive weeks (Oja et al., 1991). Participants accumulated an average of 1 hour of daily AT. Cardiovascular fitness (VO2max) increased by 4.5% in the AT group, while cycling led to a greater improvement than walking (7% vs. 2%). Statistically significant improvements were also noted for HDL cholesterol, blood lactate concentration during exercise and cardiovascular endurance in the AT group (Oja et al., 1991). In the second randomized controlled trial, 122 Dutch adults were assigned either to a PT or a cycling group for a 6 months period (Hendriksen et al., 2000). Maximal power output increased by 13% in both men and women, and VO2max increased by 6% in men while, in the PT group, VO2max decreased by 5-10%. In addition, despite no changes in body mass, there was a reduction in the percentage of body fat in the cycling group (Hendriksen et al., 2000).

In a Belgian non-randomized controlled trial, 65 middle-aged men and women were asked to cycle to work at least three times a week for a one year period, while the 15 participants from the control group were asked not to change their travel patterns (de Geus et al., 2009). Participants from the experimental group cycled 2.5 days per week for 45 minutes per day. Their maximal power output increased by 5%, but their VO2max decreased by 1%, while corresponding figures in control participants were 2% and -7%. Albeit small, differences between groups were significant for both maximal power output and VO2max, and individuals who cycled the most had greater improvements in fitness. As hypothesized by Oja et al. (2011), the smaller effect observed in this study compared to the 2 randomized controlled trials could be explained by lower cycling frequency and lower daily cycling time.

## **4.2 Prospective studies**

418 Greenhouse Gases – Emission, Measurement and Management

in the suburbs and exurbs (e.g. greenfield development) in San Diego (CA), Montgomery County (MD) and West Palm Beach (FL). In the denser "infill" developments, vehicle miles travelled were 40-50% lower, emissions of CO2 and NOx were about 50% lower and public

However, the concentrations of NOx can nevertheless be higher in more "walkable" areas, which are characterized by higher population density and land-use mix (Marshall et al., 2009), emphasizing the need to distinguish emissions and concentrations. For example, in Vancouver (Canada), a strong correlation was noted between neighborhoods walkability and NOx concentrations. High walkability areas were mostly located in the more densely populated city center through which residents of the suburbs are likely to commute to work. However, ozone is a secondary pollutant formed in the atmosphere rather than emitted directly from traffic exhaust, so high concentrations generally occur downwind from the high density areas; hence ozone concentrations in Vancouver where higher in the suburbs

Altogether, these studies suggest that the design of more walkable neighborhoods could lead to a decrease in the mode share of driving that should, in turn, lead to reductions of GHG emissions. However, because these studies have used a cross-sectional design, their findings must be interpreted with caution. There may be an underlying self-selection issue such that individuals choose to live in a neighborhood that meets certain preferences (Frank et al., 2007). To address this issue, Handy and colleagues (2005) compared the travel behaviors of residents who had moved during a 1-year period to others that did not while controlling for neighborhood preferences. They found that changes in neighborhood characteristics were associated with changes in walking and driving. In their model, accessibility of destinations was the strongest predictor of changes in driving, emphasizing

Another study explored the relationship between urban form and the ecological footprint associated with commuting patterns in the 163 municipalities of the Barcelona metropolitan region, Spain (Muniz & Galindo, 2005). Between 1986 and 1996, per capita ecological footprint increased by 95% driven by increases in the average trip distance (from 4.6 km to 6.7 km), and the mode share of car trips (22 to 35%). The mean trip distance and the ecological footprint of commuting increased systematically as distance from the city center increased reflecting higher car use. In addition, population density and accessibility of

A recent systematic review identified 2 randomized controlled trials in which individuals were randomly assigned either to an AT or a passive transport (PT) group (Oja et al., 2011). Specifically, in the first study, 68 Finnish adults who were classified as insufficiently active before the study were assigned either to a PT or an AT (walking or cycling) group. Individuals from the latter group were asked to engage in AT to work on every workday for 10 consecutive weeks (Oja et al., 1991). Participants accumulated an average of 1 hour of daily AT. Cardiovascular fitness (VO2max) increased by 4.5% in the AT group, while cycling led to a greater improvement than walking (7% vs. 2%). Statistically significant

destination were negatively associated with ecological footprint.

**4. Health-related outcomes of active transport** 

infrastructure costs were also markedly lower.

(Marshall et al., 2009).

the importance of land-use mix.

**4.1 Controlled trials** 

Andersen and colleagues (2000) assessed the relationship between cycling to work and allcause mortality among a cohort of 6 954 Danish adults. The relative risk of mortality was 0.72 (95% CI 0.57-0.91) among the cyclists, even after adjusting for physical activity, cholesterol and triglyceride levels, body mass index, blood pressure, education level, smoking status, gender, and age at the beginning of the study. Similarly, a prospective study of 75 221 women aged 40-70 from seven communities in Shanghai has shown lower rates of all-cause mortality after 5.7 years of follow-up. The relative risk of mortality was 0.66 (95% CI 0.40-1.07) among the women who cycled the most when adjusted for a similar range of potential confounders (Matthews et al., 2007).

Gordon-Larsen and colleagues (2009) examined the prospective impact of AT to work on body composition, cardiovascular disease risk factors and cardiovascular fitness among 2364 participants from the Coronary Artery Risk Development in Young Adults that were followed for 20 years. When controlling for age, race, income, education, smoking, examination center and physical activity index, men who used AT to work had a reduced risk of obesity (OR = 0.50; 95% CI = 0.33-0.76), lower triglyceride levels (OR = 0.88; 99% CI = 0.80-0.98), fasting insulin (OR = 0.86; 95% CI = 0.78-0.93) and diastolic blood pressure (OR = -1.67; 95% CI = -3.20;-0.15). Furthermore, both men and women who used AT had greater cardiovascular fitness as measured by time to exhaustion on a treadmill test.

Hu and colleagues (2005) prospectively followed a cohort of 47 721 Finnish adults aged 25- 64 and without history of coronary heart disease, stroke or cancer at baseline. They found that compared to participants who did not used AT, those who used AT for 1-29 minutes per day (RR = 0.92; 95% CI = 0.84-1.01) and those who engaged in 30 minutes of more of daily AT (RR = 0.89; 95% CI = 0.80-0.98) were less likely to have suffered a stroke during the 19 years of follow-up. In the same cohort, AT to work was also associated with a reduced risk of coronary heart disease in women, but not in men (Hu et al., 2007).

The Environmental and Population Health Benefits of Active Transport: A Review 421

substituted by walking. Their findings indicate that 31.2% of the daily trips were < 1 km and that 33.0%of these trips were motorized. Overall, these trips would require an average of 2238 steps per child, which translates into 16.6% of the recommended volume of daily

Faulkner and colleagues' (2009) systematic review has shown that AT to and from school was associated with a higher level of objectively measured physical activity level among children and adolescents in 11 out of 13 studies. A more recent systematic review has evaluated the associations between AT to and from school, PA, body composition and cardiovascular fitness (Larouche et al., submitted). Currently, 78% of studies have shown that active commuters accumulate significantly more daily PA than passive commuters. A dose-response relationship such that active commuters who lived further away from school accumulated more PA than those who lived closer has been found in two of the reviewed

However, there remains inconclusive evidence regarding the association between AT to and from school and body composition outcomes (Larouche et al., submitted). This is not very surprising because body weight regulation is a very complex phenomenon (Jéquier & Tappy, 1999; Keith et al., 2006) and even the evidence on the association between overall physical activity levels and body weight remains conflicting (Harris et al., 2009). Further, the energy expenditure associated with AT can be compensated by changes in other energybalance related behaviors during the day (i.e. by eating more or by spending more time in sedentary behaviors). However, all reviewed studies have shown that cycling was associated with greater physical fitness, while associations were less consistent for walking

Despite the potential of AT to reduce GHG emissions while improving many indicators of public health, the proportion of individuals using active modes of transportation remains very low in several countries. While several initiatives have been implemented to promote utilitarian walking and cycling, few of these have been evaluated rigorously. Systematic reviews of interventions to promote walking and cycling as an alternative to driving have indicated modest positive changes (Ogilvie et al., 2004; Yang et al., 2010). Furthermore, few studies have sought to determine the factors associated with changes in travel mode (Hume

Altogether, these limitations in current knowledge highlight the importance of examining more carefully the correlates of AT. While a detailed review of the correlates of AT is beyond the scope of this book, a brief description of the main correlates is provided for five different levels of influence according to the ecological model: the built and physical environment4, public policies, communities and organizations, interpersonal factors, and

4 The built environment refers to "the physical features of the urban landscape (i.e. alterations to the natural landscape) that collectively define the public realm, which might be as modest as a sidewalk or

an in-neighborhood retail shop or as large as a new town." (Cervero & Kockelman, 1997).

or AT in general (e.g. combined walking and cycling) (Larouche et al., submitted).

physical activity.

studies (Panter et al., 2011; van Sluijs et al., 2009).

**5. Correlates of active transportation** 

et al., 2009; Owen et al., 2004).

personal factors.

Using a prospective case-control study design, Wennberg et al. (2010) found that car commuting was associated with a markedly increased risk of myocardial infarction (OR = 1.77; 95% CI = 1.05-2.99), even when adjusting for smoking, education level, leisure time physical activity, occupational physical activity, hypertension and diabetes. The effect of AT was partly mediated by variables such as BMI, the ratio of apolipoprotein B/apolipoprotein A-1 and other inflammatory and hemostatic biomarkers.

In line with these findings, a meta-analysis of prospective epidemiological studies has indicated that individuals who use AT have an 11% lower risk of cardiovascular events including mortality, coronary heart disease, stroke, hypertension and diabetes (Hamer & Chida, 2008). However, differences were more pronounced in women (RR = 0.87; 95% CI = 0.77-0.98) than in men (RR = 0.91; 95% CI = 0.80-1.04) and direct comparison of studies is complicated by the inclusion of different endpoints.

Wen and Rissel (2008) examined the association between different modes of transport to work and the risk of overweight and obesity in a representative sample of 6810 Australian adults. The prevalence of overweight (39.8% vs. 60.8%; OR=0.49; 95% CI 0.31-0.76) and obesity (5.4% vs. 15.5%; OR = 0.34; 95% CI = 0.13-0.87) was significantly lower among men who cycled to work. No such differences were seen in women, of whom only 0.33% cycled to work. There were also no difference between individuals who walked to work and those who drove.

Among children, data from the 322 children that participated to the 6-year follow-up of the Danish arm of the European Youth Heart Study indicated a positive relationship between cycling to school and cardiovascular fitness (Cooper et al., 2008). Children who cycled at both time points and those who changed from non-cycling to cycling had a 9% greater maximal aerobic power than those who did not cycle at any time point and those who stopped cycling. However, no difference in BMI was observed in this study, and children who walked to school did not have greater fitness than those who were driven. A more recent analysis of the same dataset revealed longitudinal associations between cycling to school and lower waist circumference, serum glucose and insulin (Andersen et al., 2011). Cyclists were less resistant to insulin according to the homeostasis model assessment (e.g. HOMA), and their ratio of HDL to total cholesterol was also lower. Differences in other risk factors were not significant, but overall, the clustered risk of cardiovascular disease was markedly lower among cyclists.

#### **4.3 Cross-sectional studies and literature reviews**

Pucher and colleagues (2010) assessed whether the mode share of AST was associated with the prevalence of obesity and diabetes and with the proportion of individuals meeting the PA guidelines in 14 different countries, all 50 states in the US, and 47 of the 50 largest American cities. They found large negative correlations (*r* = -0.45 to -0.8) between the mode share of AT and obesity rates at the country, state and city levels. They also found that the prevalence of diabetes was markedly lower and that more individuals were sufficiently active in cities and states with higher mode share of AT.

Using data from the Montréal Origin-Destination survey, Morency and Demers (2009) determined the number of trips of less than 1 km made by children aged 5-14 (N = 16 837) and estimated the number of steps per day that these trips would involve if they were

Using a prospective case-control study design, Wennberg et al. (2010) found that car commuting was associated with a markedly increased risk of myocardial infarction (OR = 1.77; 95% CI = 1.05-2.99), even when adjusting for smoking, education level, leisure time physical activity, occupational physical activity, hypertension and diabetes. The effect of AT was partly mediated by variables such as BMI, the ratio of apolipoprotein B/apolipoprotein

In line with these findings, a meta-analysis of prospective epidemiological studies has indicated that individuals who use AT have an 11% lower risk of cardiovascular events including mortality, coronary heart disease, stroke, hypertension and diabetes (Hamer & Chida, 2008). However, differences were more pronounced in women (RR = 0.87; 95% CI = 0.77-0.98) than in men (RR = 0.91; 95% CI = 0.80-1.04) and direct comparison of studies is

Wen and Rissel (2008) examined the association between different modes of transport to work and the risk of overweight and obesity in a representative sample of 6810 Australian adults. The prevalence of overweight (39.8% vs. 60.8%; OR=0.49; 95% CI 0.31-0.76) and obesity (5.4% vs. 15.5%; OR = 0.34; 95% CI = 0.13-0.87) was significantly lower among men who cycled to work. No such differences were seen in women, of whom only 0.33% cycled to work. There were also no difference between individuals who walked to work and those

Among children, data from the 322 children that participated to the 6-year follow-up of the Danish arm of the European Youth Heart Study indicated a positive relationship between cycling to school and cardiovascular fitness (Cooper et al., 2008). Children who cycled at both time points and those who changed from non-cycling to cycling had a 9% greater maximal aerobic power than those who did not cycle at any time point and those who stopped cycling. However, no difference in BMI was observed in this study, and children who walked to school did not have greater fitness than those who were driven. A more recent analysis of the same dataset revealed longitudinal associations between cycling to school and lower waist circumference, serum glucose and insulin (Andersen et al., 2011). Cyclists were less resistant to insulin according to the homeostasis model assessment (e.g. HOMA), and their ratio of HDL to total cholesterol was also lower. Differences in other risk factors were not significant, but overall, the clustered risk of cardiovascular disease was

Pucher and colleagues (2010) assessed whether the mode share of AST was associated with the prevalence of obesity and diabetes and with the proportion of individuals meeting the PA guidelines in 14 different countries, all 50 states in the US, and 47 of the 50 largest American cities. They found large negative correlations (*r* = -0.45 to -0.8) between the mode share of AT and obesity rates at the country, state and city levels. They also found that the prevalence of diabetes was markedly lower and that more individuals were sufficiently

Using data from the Montréal Origin-Destination survey, Morency and Demers (2009) determined the number of trips of less than 1 km made by children aged 5-14 (N = 16 837) and estimated the number of steps per day that these trips would involve if they were

A-1 and other inflammatory and hemostatic biomarkers.

complicated by the inclusion of different endpoints.

who drove.

markedly lower among cyclists.

**4.3 Cross-sectional studies and literature reviews** 

active in cities and states with higher mode share of AT.

substituted by walking. Their findings indicate that 31.2% of the daily trips were < 1 km and that 33.0%of these trips were motorized. Overall, these trips would require an average of 2238 steps per child, which translates into 16.6% of the recommended volume of daily physical activity.

Faulkner and colleagues' (2009) systematic review has shown that AT to and from school was associated with a higher level of objectively measured physical activity level among children and adolescents in 11 out of 13 studies. A more recent systematic review has evaluated the associations between AT to and from school, PA, body composition and cardiovascular fitness (Larouche et al., submitted). Currently, 78% of studies have shown that active commuters accumulate significantly more daily PA than passive commuters. A dose-response relationship such that active commuters who lived further away from school accumulated more PA than those who lived closer has been found in two of the reviewed studies (Panter et al., 2011; van Sluijs et al., 2009).

However, there remains inconclusive evidence regarding the association between AT to and from school and body composition outcomes (Larouche et al., submitted). This is not very surprising because body weight regulation is a very complex phenomenon (Jéquier & Tappy, 1999; Keith et al., 2006) and even the evidence on the association between overall physical activity levels and body weight remains conflicting (Harris et al., 2009). Further, the energy expenditure associated with AT can be compensated by changes in other energybalance related behaviors during the day (i.e. by eating more or by spending more time in sedentary behaviors). However, all reviewed studies have shown that cycling was associated with greater physical fitness, while associations were less consistent for walking or AT in general (e.g. combined walking and cycling) (Larouche et al., submitted).

## **5. Correlates of active transportation**

Despite the potential of AT to reduce GHG emissions while improving many indicators of public health, the proportion of individuals using active modes of transportation remains very low in several countries. While several initiatives have been implemented to promote utilitarian walking and cycling, few of these have been evaluated rigorously. Systematic reviews of interventions to promote walking and cycling as an alternative to driving have indicated modest positive changes (Ogilvie et al., 2004; Yang et al., 2010). Furthermore, few studies have sought to determine the factors associated with changes in travel mode (Hume et al., 2009; Owen et al., 2004).

Altogether, these limitations in current knowledge highlight the importance of examining more carefully the correlates of AT. While a detailed review of the correlates of AT is beyond the scope of this book, a brief description of the main correlates is provided for five different levels of influence according to the ecological model: the built and physical environment4, public policies, communities and organizations, interpersonal factors, and personal factors.

<sup>4</sup> The built environment refers to "the physical features of the urban landscape (i.e. alterations to the natural landscape) that collectively define the public realm, which might be as modest as a sidewalk or an in-neighborhood retail shop or as large as a new town." (Cervero & Kockelman, 1997).

The Environmental and Population Health Benefits of Active Transport: A Review 423

in which individuals were exposed to high concentrations of coal and other pollutants (Frumkin et al., 2004). To address this public health issue, urban planning policies were implemented to segregate factories from households. Nevertheless, urban areas of the early 20th century were still generally favorable to AT: density was moderate to high, land-use mix was the norm and streets were built in a grid-like fashion which facilitates navigation. Then, after World War II, urban planning policies primarily focused on car-oriented development which has led to increasing urban sprawl and decreasing density, land-use mix and street connectivity. Hence, Turcotte (2008) reported that newer homes where located further away from the city center based on findings from the 2006 Canadian census. This type of development has led to major reductions in the mode share of AT (Frumkin et

In contrast, to address the environmental and public health issues associated with caroriented development, Northern European countries adopted a series of policies to promote AT and reduce car travel in the middle of the 1970s, as described by Pucher and Buehler (2008). These policies have led to 20-43% increases in the mode share of AST in cities such as Amsterdam and Groningen (Netherlands), Copenhagen and Odense (Denmark) and Berlin and Muenster (Germany), and large reductions in the relative risk of traffic injuries were

Local communities can promote AT through different strategies, including the provision of walking and cycling infrastructure (Dill & Carr, 2003; Pucher et al., 2010), bike-sharing schemes (DeMaio, 2009), parking restrictions (Pucher & Buehler, 2008) and Safe Routes to School and/or walking school buses schemes (Heelan et al., 2009; Hinckson & Badland, 2011). Moreover, Giles-Corti (2006) argues that the inclusive or exclusive nature of the social environment fosters the social norm that may increase or decrease the likelihood of PA in general or AT specifically – and that car-oriented development favors social exclusion. It has also been proposed that increasing social cohesion could lead to more "eyes on the street"

Workplaces can also play a role in the promotion of AT. For example, a workplace-based social marketing campaign in a large Finnish factory (N = 1256 employees) achieved a 7% increase in the proportion of workers using AT (Oja et al., 1998). Australian researchers examined the influence of workplaces on travel mode among 888 adults (Wen et al., 2010). Respondents whose workplace encouraged AT were significantly less likely to drive to work than those without such encouragement (49% vs. 73%; OR = 0.41; 95% CI = 0.23-0.73). In contrast, participants who had access to convenient parking near their workplace were almost 5 times more likely to drive. In Belgium, a similar survey revealed that individuals who reported that their workplace offered safe bicycle parking spots and showers were

According to Bandura's (1986) social cognitive theory, human behavior is strongly influenced by modeling and by social interactions with significant others. In support of this theory, researchers have found that individuals whose friends and/or family members

(Jacobs, 1961), which could favor children's independent mobility and AT.

al., 2004; King et al., 2002).

**5.3 Communities and organizations** 

more likely to cycle to work (de Geus et al., 2008).

**5.4 Interpersonal factors** 

observed.

## **5.1 Built and physical environment correlates**

Several reviews highlight strong associations between elements of the built environment, the mode share of active transport and the number of vehicle miles travelled (Larouche & Trudeau, 2010; Saelens & Handy, 2008; Sallis et al., 2006). The term "walkability" is often used by urban planners and researchers to refer to the ability to walk to certain destinations within the neighborhood (Sallis et al., 2006). The 3D model proposed by Cervero and Kockelman (1997) is a central component in most indices of walkability (Sallis et al., 2006). The three components of this model are: 1) Density (e.g. the amount of individuals or households within a given area); 2) Diversity (often referred to as land-use mix or the variety of destination within a given area); and 3) Design (which includes various attributes such as street connectivity, sidewalks and cycling infrastructure, aesthetics, etc.).

There is consistent evidence supporting the associations between density and land-use mix with adults' travel mode (Ewing et al., 2003; Frank et al., 2006; Saelens & Handy, 2008). Pooling data from the 206 992 adult participants to the 1998, 1999 and 2000 Behavioral Risk Factor Surveillance System, Ewing et al. (2003) assessed the relationship between urban sprawl and physical activity, obesity and morbidity in 448 counties and 83 metropolitan areas in the US. The county sprawl index was positively associated to body mass index and hypertension, and negatively associated with time spent walking. Another American study reported that individual who lived in neighborhoods characterized by high walkability accumulated 70 more minutes of weekly PA than those who lived in less walkable neighborhoods (Saelens et al., 2003). However, studies that examined the impact of design on travel mode have found conflicting results (Saelens & Handy, 2008; Sallis et al., 2006). Another caveat of the walkability construct is neighborhood self-selection; i.e. active individuals may decide to live in walkable areas (Handy et al., 2006).

Amongst children and adolescents, a systematic review of 14 studies which measured environmental correlates of AT to/from school with geographic information systems has shown that distance was the only consistent correlate (Wong et al., 2011). There was conflicting evidence regarding the impact of density and diversity, suggesting that these constructs may not be as relevant as for adults. The mode share of AT to/from school also tends to be lower in secondary schools students who generally need to travel greater distances (Pabayo et al., 2008, 2011).

#### **5.2 Public policies**

Differences between countries in the mode share of cycling are particularly striking. For example, the proportion of trips made by bicycle varies from about 1% in countries such as USA, UK, Australia and Canada to 18% in Denmark and 27% in the Netherlands (Pucher & Buehler, 2008). Further the mode shares of cycling for distances < 2.5 km, 2.5-4.4 km and 4.5- 6.4 km in the Netherlands are respectively 37%, 37% and 24%, while corresponding figures in the US are 2%, 1% and 0.4%. These large variations can be partly explained by differences between countries in factors such as the provision of cycling infrastructure, the quality of public transit networks, gas prices and other expenses associated with car ownership, and urban planning practices (Banister, 2008; Chapman, 2007).

From an historical perspective, the industrial revolution has led to a major increase in the population of cities and infectious diseases spread rapidly in overcrowded neighborhoods in which individuals were exposed to high concentrations of coal and other pollutants (Frumkin et al., 2004). To address this public health issue, urban planning policies were implemented to segregate factories from households. Nevertheless, urban areas of the early 20th century were still generally favorable to AT: density was moderate to high, land-use mix was the norm and streets were built in a grid-like fashion which facilitates navigation. Then, after World War II, urban planning policies primarily focused on car-oriented development which has led to increasing urban sprawl and decreasing density, land-use mix and street connectivity. Hence, Turcotte (2008) reported that newer homes where located further away from the city center based on findings from the 2006 Canadian census. This type of development has led to major reductions in the mode share of AT (Frumkin et al., 2004; King et al., 2002).

In contrast, to address the environmental and public health issues associated with caroriented development, Northern European countries adopted a series of policies to promote AT and reduce car travel in the middle of the 1970s, as described by Pucher and Buehler (2008). These policies have led to 20-43% increases in the mode share of AST in cities such as Amsterdam and Groningen (Netherlands), Copenhagen and Odense (Denmark) and Berlin and Muenster (Germany), and large reductions in the relative risk of traffic injuries were observed.

## **5.3 Communities and organizations**

422 Greenhouse Gases – Emission, Measurement and Management

Several reviews highlight strong associations between elements of the built environment, the mode share of active transport and the number of vehicle miles travelled (Larouche & Trudeau, 2010; Saelens & Handy, 2008; Sallis et al., 2006). The term "walkability" is often used by urban planners and researchers to refer to the ability to walk to certain destinations within the neighborhood (Sallis et al., 2006). The 3D model proposed by Cervero and Kockelman (1997) is a central component in most indices of walkability (Sallis et al., 2006). The three components of this model are: 1) Density (e.g. the amount of individuals or households within a given area); 2) Diversity (often referred to as land-use mix or the variety of destination within a given area); and 3) Design (which includes various attributes

There is consistent evidence supporting the associations between density and land-use mix with adults' travel mode (Ewing et al., 2003; Frank et al., 2006; Saelens & Handy, 2008). Pooling data from the 206 992 adult participants to the 1998, 1999 and 2000 Behavioral Risk Factor Surveillance System, Ewing et al. (2003) assessed the relationship between urban sprawl and physical activity, obesity and morbidity in 448 counties and 83 metropolitan areas in the US. The county sprawl index was positively associated to body mass index and hypertension, and negatively associated with time spent walking. Another American study reported that individual who lived in neighborhoods characterized by high walkability accumulated 70 more minutes of weekly PA than those who lived in less walkable neighborhoods (Saelens et al., 2003). However, studies that examined the impact of design on travel mode have found conflicting results (Saelens & Handy, 2008; Sallis et al., 2006). Another caveat of the walkability construct is neighborhood self-selection; i.e. active

Amongst children and adolescents, a systematic review of 14 studies which measured environmental correlates of AT to/from school with geographic information systems has shown that distance was the only consistent correlate (Wong et al., 2011). There was conflicting evidence regarding the impact of density and diversity, suggesting that these constructs may not be as relevant as for adults. The mode share of AT to/from school also tends to be lower in secondary schools students who generally need to travel greater

Differences between countries in the mode share of cycling are particularly striking. For example, the proportion of trips made by bicycle varies from about 1% in countries such as USA, UK, Australia and Canada to 18% in Denmark and 27% in the Netherlands (Pucher & Buehler, 2008). Further the mode shares of cycling for distances < 2.5 km, 2.5-4.4 km and 4.5- 6.4 km in the Netherlands are respectively 37%, 37% and 24%, while corresponding figures in the US are 2%, 1% and 0.4%. These large variations can be partly explained by differences between countries in factors such as the provision of cycling infrastructure, the quality of public transit networks, gas prices and other expenses associated with car ownership, and

From an historical perspective, the industrial revolution has led to a major increase in the population of cities and infectious diseases spread rapidly in overcrowded neighborhoods

such as street connectivity, sidewalks and cycling infrastructure, aesthetics, etc.).

individuals may decide to live in walkable areas (Handy et al., 2006).

urban planning practices (Banister, 2008; Chapman, 2007).

distances (Pabayo et al., 2008, 2011).

**5.2 Public policies** 

**5.1 Built and physical environment correlates** 

Local communities can promote AT through different strategies, including the provision of walking and cycling infrastructure (Dill & Carr, 2003; Pucher et al., 2010), bike-sharing schemes (DeMaio, 2009), parking restrictions (Pucher & Buehler, 2008) and Safe Routes to School and/or walking school buses schemes (Heelan et al., 2009; Hinckson & Badland, 2011). Moreover, Giles-Corti (2006) argues that the inclusive or exclusive nature of the social environment fosters the social norm that may increase or decrease the likelihood of PA in general or AT specifically – and that car-oriented development favors social exclusion. It has also been proposed that increasing social cohesion could lead to more "eyes on the street" (Jacobs, 1961), which could favor children's independent mobility and AT.

Workplaces can also play a role in the promotion of AT. For example, a workplace-based social marketing campaign in a large Finnish factory (N = 1256 employees) achieved a 7% increase in the proportion of workers using AT (Oja et al., 1998). Australian researchers examined the influence of workplaces on travel mode among 888 adults (Wen et al., 2010). Respondents whose workplace encouraged AT were significantly less likely to drive to work than those without such encouragement (49% vs. 73%; OR = 0.41; 95% CI = 0.23-0.73). In contrast, participants who had access to convenient parking near their workplace were almost 5 times more likely to drive. In Belgium, a similar survey revealed that individuals who reported that their workplace offered safe bicycle parking spots and showers were more likely to cycle to work (de Geus et al., 2008).

#### **5.4 Interpersonal factors**

According to Bandura's (1986) social cognitive theory, human behavior is strongly influenced by modeling and by social interactions with significant others. In support of this theory, researchers have found that individuals whose friends and/or family members

The Environmental and Population Health Benefits of Active Transport: A Review 425

like it has happened several times in history, albeit at a much smaller scale (Weiss &

As the transport sector is currently one of the most important contributors of climate change (Unger et al., 2010) while PT represents a significant health risk, it is fair to consider that transport practices represent a major nexus between the environment and public health. In other words, interventions promoting AT as a source of physical activity also have important environmental implications and, inversely, GHG emissions policies aiming to

One factor that may play a determinant role with respect to both GHG emissions and indicators of public health is the likely occurrence of "peak oil" which is defined as "*the maximum rate of the production of oil in any area under consideration, recognizing that it is a finite natural resource subject to depletion*"5 While some authors perceive peak oil as a great opportunity to promote healthier modes of transport (Demers, 2008; Frumkin et al., 2004; Younger et al., 2008), others are less optimistic (Hanlon & McCartney, 2008; Kaza et al., 2011). Nevertheless, the impacts of peak oil are difficult to forecast given the uncertainty associated with its occurrence, its impact on gas price, the rapidly increasing oil demand from developing countries, the availability of alternative fuels and their actual impact on pollutant emissions, the public policies that will be implemented by different countries, etc. The gas embargo of the 1970s show that countries can use different strategies in order to adapt to the rising price of gas (Mapes, 2009). In the US, sales of bicycles literally doubled in the 2 years following the embargo, but once gas prices dropped, bicycle sales plummeted, and the mode share of cycling remained very low thereafter. In contrast, this crisis made Dutch policy makers realize that they did not have control over their gas supplies so, after the embargo was over, they maintained high fuel taxes and invested massively in the

development of infrastructure for alternative modes of transportation (Mapes, 2009).

Another insightful example in this regard is the fuel shortage experienced during Cuba's "special period" following the collapse of the Soviet Union in the early 1990s (Franco et al., 2008) which has led to a major increase in the mode share of AT. The proportion of physically active individuals increased from 30 to 67%, while the prevalence of obesity declined from 14 to 7% and large reductions in mortality from diabetes (-51%), coronary heart disease (-35%), stroke (-20%) and all-cause mortality (-18%) were observed. However, a neuropathy outbreak affected 50 000 individuals in 1992-1993, but its impact amongst children, pregnant women and the elderly was minimized by a special rationing system. These historical examples suggest that different countries will implement different strategies in response to peak oil, and these strategies may have markedly different environmental

Health impact assessment studies represent a promising strategy to simultaneously assess the environmental and health-related impacts of AT policies and interventions (de Nazelle et al., 2011). As illustrated in Figure 2, a comprehensive framework has been proposed to integrate the impact of policies/interventions on: 1) the behavioral changes (including

reduce car use would have major public health implications.

Bradley, 2001).

and public health impacts.

**6.1 The case for health impact assessment studies** 

5 For more details, see: http://www.peakoil.net/about-peak-oil

engaged in AT were more likely to walk or cycle for utilitarian purposes (de Geus et al., 2007; Titze et al., 2007). Similarly, social support has been shown to be positively related to AT (Ball et al., 2007; Titze et al., 2008).

In addition, current theoretical frameworks assume that parents are the key decision makers for their children's mode of transportation to school (McMillan, 2005; Panter et al., 2008). To further explore this decision making process, Faulkner et al. (2010) conducted semistructured interviews with 37 parents. Their findings suggest that, after deciding if their children needed to be escorted, parents typically chose the mode of transport that they perceived to be the quickest and most convenient. Other studies have shown that parental concerns about road safety are inversely related with children's likelihood of walking or cycling to school (Davison et al., 2008; Panter et al., 2008).

## **5.5 Personal factors**

Several studies have shown that men were more likely than women to use AT and this gender difference tends to be much greater for cycling (Bell et al., 2006; Davison et al., 2008; Garrard et al., 2008). Interestingly, in countries with higher mode shares of cycling, no gender differences were observed whereas in Australia, US, UK and Canada, less than 30% of bike trips are made by women (Pucher & Buehler, 2008). Australian researchers suggested that this situation may reflect higher risk aversion among women; underscoring the importance of developing safe cycling infrastructure to promote commuter cycling among women (Garrard et al., 2008). Perceptions of lack of time (Oja et al., 1998; Shannon et al., 2006), car ownership (Ogilvie et al., 2008; Wen et al., 2010), road safety concerns (Garrard et al., 2008; Panter et al., 2008; Titze et al., 2007) and hilly routes (Troped et al., 2003) have also been associated with lower rates of AT. While weather conditions are often cited as a barrier to AT (Humpel et al., 2004), it is noteworthy that the Scandinavian countries all have cycling mode shares above 10%, while cycling is much less common in Mediterranean countries, US and Australia (Pucher & Buehler, 2008).

In contrast, studies have consistently shown a positive association between self-efficacy perceptions and AT (Ball et al., 2007; de Geus et al., 2008; Sallis et al., 2006; Troped et al., 2003). Other personal factors that may favor AT include the desire to be physically active or to increase fitness (Oja et al., 1998; Shannon et al., 2006; Titze et al., 2007), intentions and habits to walk or cycle (Lemieux & Godin, 2009), and environment-friendly attitudes (Bopp et al., 2011; de Geus et al., 2008).

## **6. The nexus between the environment and public health**

As highlighted by McMichael and Butler (2011), there is currently a paradox between the increasing disruptions of the global environments (i.e. rising GHG emissions, oversized ecological footprint, acidification of the oceans, destruction of stratospheric ozone, degradation of arable land, depletion of freshwater supplies, loss of biodiversity, etc.) and the improvement in indices of population health (especially life expectancy). They further explain: "*The factors that have historically underpinned population health gains are now, by dint of their much increased scale, scope, and intensity undermining sustainable good health as we exceed Earth's capacity to renew, replenish, provide and restore*" (McMichael and Butler, 2011). Hence, profound changes in lifestyle will likely be required to avoid widespread societal collapse,

engaged in AT were more likely to walk or cycle for utilitarian purposes (de Geus et al., 2007; Titze et al., 2007). Similarly, social support has been shown to be positively related to

In addition, current theoretical frameworks assume that parents are the key decision makers for their children's mode of transportation to school (McMillan, 2005; Panter et al., 2008). To further explore this decision making process, Faulkner et al. (2010) conducted semistructured interviews with 37 parents. Their findings suggest that, after deciding if their children needed to be escorted, parents typically chose the mode of transport that they perceived to be the quickest and most convenient. Other studies have shown that parental concerns about road safety are inversely related with children's likelihood of walking or

Several studies have shown that men were more likely than women to use AT and this gender difference tends to be much greater for cycling (Bell et al., 2006; Davison et al., 2008; Garrard et al., 2008). Interestingly, in countries with higher mode shares of cycling, no gender differences were observed whereas in Australia, US, UK and Canada, less than 30% of bike trips are made by women (Pucher & Buehler, 2008). Australian researchers suggested that this situation may reflect higher risk aversion among women; underscoring the importance of developing safe cycling infrastructure to promote commuter cycling among women (Garrard et al., 2008). Perceptions of lack of time (Oja et al., 1998; Shannon et al., 2006), car ownership (Ogilvie et al., 2008; Wen et al., 2010), road safety concerns (Garrard et al., 2008; Panter et al., 2008; Titze et al., 2007) and hilly routes (Troped et al., 2003) have also been associated with lower rates of AT. While weather conditions are often cited as a barrier to AT (Humpel et al., 2004), it is noteworthy that the Scandinavian countries all have cycling mode shares above 10%, while cycling is much less common in Mediterranean

In contrast, studies have consistently shown a positive association between self-efficacy perceptions and AT (Ball et al., 2007; de Geus et al., 2008; Sallis et al., 2006; Troped et al., 2003). Other personal factors that may favor AT include the desire to be physically active or to increase fitness (Oja et al., 1998; Shannon et al., 2006; Titze et al., 2007), intentions and habits to walk or cycle (Lemieux & Godin, 2009), and environment-friendly attitudes (Bopp

As highlighted by McMichael and Butler (2011), there is currently a paradox between the increasing disruptions of the global environments (i.e. rising GHG emissions, oversized ecological footprint, acidification of the oceans, destruction of stratospheric ozone, degradation of arable land, depletion of freshwater supplies, loss of biodiversity, etc.) and the improvement in indices of population health (especially life expectancy). They further explain: "*The factors that have historically underpinned population health gains are now, by dint of their much increased scale, scope, and intensity undermining sustainable good health as we exceed Earth's capacity to renew, replenish, provide and restore*" (McMichael and Butler, 2011). Hence, profound changes in lifestyle will likely be required to avoid widespread societal collapse,

AT (Ball et al., 2007; Titze et al., 2008).

**5.5 Personal factors** 

cycling to school (Davison et al., 2008; Panter et al., 2008).

countries, US and Australia (Pucher & Buehler, 2008).

**6. The nexus between the environment and public health** 

et al., 2011; de Geus et al., 2008).

like it has happened several times in history, albeit at a much smaller scale (Weiss & Bradley, 2001).

As the transport sector is currently one of the most important contributors of climate change (Unger et al., 2010) while PT represents a significant health risk, it is fair to consider that transport practices represent a major nexus between the environment and public health. In other words, interventions promoting AT as a source of physical activity also have important environmental implications and, inversely, GHG emissions policies aiming to reduce car use would have major public health implications.

One factor that may play a determinant role with respect to both GHG emissions and indicators of public health is the likely occurrence of "peak oil" which is defined as "*the maximum rate of the production of oil in any area under consideration, recognizing that it is a finite natural resource subject to depletion*"5 While some authors perceive peak oil as a great opportunity to promote healthier modes of transport (Demers, 2008; Frumkin et al., 2004; Younger et al., 2008), others are less optimistic (Hanlon & McCartney, 2008; Kaza et al., 2011). Nevertheless, the impacts of peak oil are difficult to forecast given the uncertainty associated with its occurrence, its impact on gas price, the rapidly increasing oil demand from developing countries, the availability of alternative fuels and their actual impact on pollutant emissions, the public policies that will be implemented by different countries, etc.

The gas embargo of the 1970s show that countries can use different strategies in order to adapt to the rising price of gas (Mapes, 2009). In the US, sales of bicycles literally doubled in the 2 years following the embargo, but once gas prices dropped, bicycle sales plummeted, and the mode share of cycling remained very low thereafter. In contrast, this crisis made Dutch policy makers realize that they did not have control over their gas supplies so, after the embargo was over, they maintained high fuel taxes and invested massively in the development of infrastructure for alternative modes of transportation (Mapes, 2009).

Another insightful example in this regard is the fuel shortage experienced during Cuba's "special period" following the collapse of the Soviet Union in the early 1990s (Franco et al., 2008) which has led to a major increase in the mode share of AT. The proportion of physically active individuals increased from 30 to 67%, while the prevalence of obesity declined from 14 to 7% and large reductions in mortality from diabetes (-51%), coronary heart disease (-35%), stroke (-20%) and all-cause mortality (-18%) were observed. However, a neuropathy outbreak affected 50 000 individuals in 1992-1993, but its impact amongst children, pregnant women and the elderly was minimized by a special rationing system. These historical examples suggest that different countries will implement different strategies in response to peak oil, and these strategies may have markedly different environmental and public health impacts.

#### **6.1 The case for health impact assessment studies**

Health impact assessment studies represent a promising strategy to simultaneously assess the environmental and health-related impacts of AT policies and interventions (de Nazelle et al., 2011). As illustrated in Figure 2, a comprehensive framework has been proposed to integrate the impact of policies/interventions on: 1) the behavioral changes (including

<sup>5</sup> For more details, see: http://www.peakoil.net/about-peak-oil

The Environmental and Population Health Benefits of Active Transport: A Review 427

program was implemented at the national level. Moreover, studies have repetitively shown that more than 60% of Danish adolescents cycle to school (Andersen et al., 2011). School travel planning was also very successful in the UK where pedestrian and cyclist casualties

Fig. 2. Conceptual model of health impacts of active travel policies. In bold are shown behavioral and environmental quality variables recognized as having strongest exposurehealth quantifications available, while variables in italics are the most uncertain to quantify.

A recent systematic review of interventions promoting AT to school reported that most were effective in increasing the mode share of AT and/or decreasing driving (Chillón et al., 2011). However, the majority of these studies reported small effect sizes and most did not assess the broader impacts of the interventions on environmental and health-related outcomes.

These studies provide preliminary evidence to suggest that a transition from PT to AT could have rather positive environmental and health impacts. The magnitude of these impacts will likely depend on the degree of behavior change. For example, larger reduction in driving would achieve greater reductions in exhaust emissions and could decrease walkers' and cyclists' risk of accident to a greater extent. However, further studies will need to take into account socio-economic status because individuals living in deprived neighborhoods are typically exposed to higher concentration of pollutants (Marshall et al., 2009; Tonne et al.,

(Reproduced with permission from de Nazelle et al., 2011).

felt by 77% and 28% respectively (Appleyard, 2003).

changes in modes of transport); 2) the environment (i.e. air pollution, greenhouse gases); and 3) the health risks and benefits resulting from the interaction between behavioral changes and environmental exposures. It is anticipated that such an approach could allow researchers and policy-makers to examine if a given intervention can achieve improvements in the health status of individuals while providing environmental co-benefits.

However, only a few such assessments have been conducted to date (de Nazelle et al., 2011). First, Dutch researchers simulated the impact on all-cause mortality if 500 000 individuals would switch from car travel to cycling for short trips on a daily basis (de Hartog et al., 2010). Their findings indicate that the benefits from increased PA (3-14 months gained) would be much greater than the mortality resulting from greater doses of inhaled air pollution (0.8-40 days lost) and the potential increase in traffic accidents (5-9 days lost). Furthermore, such a change in mode of transportation would result in air quality improvements that would benefit the whole population.

A second health impact assessment study was conducted for the Bicing bike-sharing scheme in Barcelona, Spain which was implemented in 2007 as a strategy to reduce traffic congestion (Rojas-Rueda et al., 2011). By August 2009, 182 062 residents had subscribed (about 11% of the population), of which 28 251 were considered regular users (Rojas-Rueda et al., 2011). It was estimated that the Bicing initiative led to an annual reduction of 9 062 344 kg of CO2. The health impact assessment predicts that Bicing would cause 0.16 deaths/year because of traffic accidents and inhalation of air pollution, but it would prevent 12.46 deaths/year due to increased physical activity (benefit-risk ratio = 77). Considering that bike-sharing schemes have been implemented in several cities in Europe, Asia and North America (DeMaio, 2009), these initiatives could have positive impacts both for population health and climate change mitigation.

The findings of these health impact assessment studies are coherent with the "safety in number" principle which suggests that a mode of transport becomes safer as more people use it (Jacobsen, 2003; Tin Tin et al., 2011). This principle also has its corollary: when less people use a mode of transport, it tends to become more dangerous. In this regard, the number of cycling injuries has increased while per capita time spent cycling decreased during the last decade in New Zealand (Tin Tin et al., 2011).

Jacobsen's (2003) power law model predicts that, if a community was to double its mode share of AT, the relative risk of injury for each walker or cyclist would drop to 0.66. At the community level, the rate of injury would increase by 32% for a 100% increase in the mode share of AT. It is noteworthy that the relationship between the mode share of AT and rates of injuries is non-linear and tends to vary between communities. Furthermore, Elvik (2009) argued that due to the non-linearity of risk, very large transfers of trips from motorized modes to AT could lead to a reduction in the absolute number of reported accidents.

Among children and adolescents, various schemes such as "Safe Routes to School" and walking school buses have been implemented to promote AT while reducing the risk of injuries (Davison et al., 2008). In the mid-1970s, Denmark had the highest child pedestrian accident rate in Europe (Appleyard, 2003). To address this issue, the city of Odense invested in the creation of an extensive network of pedestrian and bicycle paths, narrowed roads, added traffic islands, and mandated all of its 45 schools to identify specific road-safety issues. After ten years, child pedestrian and cyclist casualties had decreased by 80% and this

changes in modes of transport); 2) the environment (i.e. air pollution, greenhouse gases); and 3) the health risks and benefits resulting from the interaction between behavioral changes and environmental exposures. It is anticipated that such an approach could allow researchers and policy-makers to examine if a given intervention can achieve improvements

However, only a few such assessments have been conducted to date (de Nazelle et al., 2011). First, Dutch researchers simulated the impact on all-cause mortality if 500 000 individuals would switch from car travel to cycling for short trips on a daily basis (de Hartog et al., 2010). Their findings indicate that the benefits from increased PA (3-14 months gained) would be much greater than the mortality resulting from greater doses of inhaled air pollution (0.8-40 days lost) and the potential increase in traffic accidents (5-9 days lost). Furthermore, such a change in mode of transportation would result in air quality

A second health impact assessment study was conducted for the Bicing bike-sharing scheme in Barcelona, Spain which was implemented in 2007 as a strategy to reduce traffic congestion (Rojas-Rueda et al., 2011). By August 2009, 182 062 residents had subscribed (about 11% of the population), of which 28 251 were considered regular users (Rojas-Rueda et al., 2011). It was estimated that the Bicing initiative led to an annual reduction of 9 062 344 kg of CO2. The health impact assessment predicts that Bicing would cause 0.16 deaths/year because of traffic accidents and inhalation of air pollution, but it would prevent 12.46 deaths/year due to increased physical activity (benefit-risk ratio = 77). Considering that bike-sharing schemes have been implemented in several cities in Europe, Asia and North America (DeMaio, 2009), these initiatives could have positive impacts both for population

The findings of these health impact assessment studies are coherent with the "safety in number" principle which suggests that a mode of transport becomes safer as more people use it (Jacobsen, 2003; Tin Tin et al., 2011). This principle also has its corollary: when less people use a mode of transport, it tends to become more dangerous. In this regard, the number of cycling injuries has increased while per capita time spent cycling decreased

Jacobsen's (2003) power law model predicts that, if a community was to double its mode share of AT, the relative risk of injury for each walker or cyclist would drop to 0.66. At the community level, the rate of injury would increase by 32% for a 100% increase in the mode share of AT. It is noteworthy that the relationship between the mode share of AT and rates of injuries is non-linear and tends to vary between communities. Furthermore, Elvik (2009) argued that due to the non-linearity of risk, very large transfers of trips from motorized

Among children and adolescents, various schemes such as "Safe Routes to School" and walking school buses have been implemented to promote AT while reducing the risk of injuries (Davison et al., 2008). In the mid-1970s, Denmark had the highest child pedestrian accident rate in Europe (Appleyard, 2003). To address this issue, the city of Odense invested in the creation of an extensive network of pedestrian and bicycle paths, narrowed roads, added traffic islands, and mandated all of its 45 schools to identify specific road-safety issues. After ten years, child pedestrian and cyclist casualties had decreased by 80% and this

modes to AT could lead to a reduction in the absolute number of reported accidents.

in the health status of individuals while providing environmental co-benefits.

improvements that would benefit the whole population.

during the last decade in New Zealand (Tin Tin et al., 2011).

health and climate change mitigation.

program was implemented at the national level. Moreover, studies have repetitively shown that more than 60% of Danish adolescents cycle to school (Andersen et al., 2011). School travel planning was also very successful in the UK where pedestrian and cyclist casualties felt by 77% and 28% respectively (Appleyard, 2003).

Fig. 2. Conceptual model of health impacts of active travel policies. In bold are shown behavioral and environmental quality variables recognized as having strongest exposurehealth quantifications available, while variables in italics are the most uncertain to quantify. (Reproduced with permission from de Nazelle et al., 2011).

A recent systematic review of interventions promoting AT to school reported that most were effective in increasing the mode share of AT and/or decreasing driving (Chillón et al., 2011). However, the majority of these studies reported small effect sizes and most did not assess the broader impacts of the interventions on environmental and health-related outcomes.

These studies provide preliminary evidence to suggest that a transition from PT to AT could have rather positive environmental and health impacts. The magnitude of these impacts will likely depend on the degree of behavior change. For example, larger reduction in driving would achieve greater reductions in exhaust emissions and could decrease walkers' and cyclists' risk of accident to a greater extent. However, further studies will need to take into account socio-economic status because individuals living in deprived neighborhoods are typically exposed to higher concentration of pollutants (Marshall et al., 2009; Tonne et al.,

The Environmental and Population Health Benefits of Active Transport: A Review 429

studies provide preliminary evidence suggesting that transport policies could achieve

The second objective of this chapter was to summarize the literature on the health benefits of AT. Altogether, the reviewed studies suggest that increasing the mode share of AT may lead to multiple health benefits (i.e. declines in the risk of obesity, cardiovascular diseases, diabetes, hypertension, cancer, etc.). The benefits vary across studies, but importantly, all available health impact assessment studies have concluded that the benefits of increasing the mode share of AT largely outweighed the risks (de Hartog et al., 2010; de Nazelle et al., 2011; Rojas-Rueda et al., 2011). Moreover, controlled trials have consistently shown that commuter cycling led to improved cardiovascular fitness. The findings from another randomized controlled trial suggest that lifestyle-oriented physical activity (such as AT) can result in greater compliance than structured physical activity programs (Dunn et al., 1999). Despite this wide range of potential benefits, the mode share of AT is exceedingly low in most industrialised countries except The Netherlands, Germany and the Scandinavian countries. This highlights the necessity of developing a better understanding of the

correlates of AT in order to increase the effectiveness of interventions to promote AT.

acceptability in order to drive social and political action (Banister, 2008).

modes to inform the design of more effective interventions.

Current models that predict the choice of travel mode (i.e. Panter et al., 2008; Sallis et al., 2006) are rooted in the social ecological theory. Based on this theory, intervening only by trying to convince individuals to change their transport behaviors would likely be effective only among individuals who are highly receptive to this message (i.e. those who have a high degree of concerns about environmental issues). In this context, Banister (2008) argues that a paradigm shift from the "conventional approach" to transport planning (or car-oriented development) towards a sustainable mobility paradigm (or people-oriented development) is urgently needed. However, to achieve such a paradigm shift, it is necessary to gain public

The example of Northern European countries suggests that it is feasible to achieve public acceptability. Furthermore, the cities that had the most success in increasing the mode share of AT are those that implemented comprehensive multi-level strategies (Pucher et al., 2008, 2010). Such strategies can include a combination of incentives to individuals who adopt active transportation (i.e. tax credits), an increased availability of sidewalks, cycle paths, improved public transport systems, an increased community/organizational support for active transportation (i.e. provision of safe places to store bicycles, showers in workplaces), traffic calming strategies, policies to increase the cost of car use (e.g. increased fuel taxes or reduced availability of car parking), and well-designed social marketing campaigns (Oja &

As noted in the 'limitations and strengths' section, there is a clear need for more prospective intervention studies in order to increase the overall quality of evidence. In this regard, future studies should: 1) assess in a more comprehensive fashion the effectiveness of AT interventions and their impact on environmental variables (including GHG emissions) and health-related outcomes; and 2) identify key variables associated with changes in travel

Although stronger proof of effectiveness from experimental studies and natural experiments would increase the quality of evidence and may potentially increase public acceptability, it

considerable reductions in GHG emissions.

Vuori, 2000; Pucher et al., 2010).

2008). Moreover, low income households and families with young children are likely to be more affected by a potential increase in the cost of food transportation caused by fuel scarcity (Kaza et al., 2011).

## **7. Limitation and strengths of present review**

Overall, this review is limited by the lack of prospective intervention studies that examined the impact of policies aimed at reducing greenhouse gases emissions. In the absence of such studies, simulations provide an interesting alternative to examine the impact of large-scale changes in transportation practices. However, simulations are limited by their reliance on highly uncertain assumptions about factors such as the potential for behavioral change, the occurrence of peak oil, the confounding variables not accounted for, and the magnitude of technological innovations.

While there has been a lot of studies on the correlates of AT, few of these have rigorously examined the factors associated with changes in travel mode. This may partly explain the modest success of interventions that sought to increase the mode share of AT. The findings of the current review nevertheless suggest that more comprehensive interventions tend to be more effective, highlighting the need for better integrated policies to promote AT and reduce the mode share of PT.

In addition the use of the social-ecological approach for the description of the correlates of AT can be perceived both as a strength and as a limitation. This model allows the consideration of a wide range of factors located at different levels of influence emphasizing the interactions between different variables (McLeroy et al., 1988; Sallis et al., 2006). However, ecological models can be criticized for being too complicated; rendering the identification of intervention targets more difficult (Stokols, 1996).

The main strength of the current review is the broadness of its scope which allowed a simultaneous examination of both the environmental and the public health impacts of AT. Further, this review summarizes the increasing evidence supporting the positive association between AT and several indicators of public health including increased PA and physical fitness, and decreased morbidity and all-cause mortality. This body of evidence should therefore be useful for policy-makers to justify investments to promote AT by the perspective of combined environmental and public health benefits. Haines et al. (2010) argue that such policies should be more cost-effective and socially attractive than strategies addressing climate change or public health separately.

## **8. Conclusion**

The share of the global CO2 emissions attributable to the transport sector is increasing steadily while the level of motorization is rising rapidly in developing countries (International Transport Forum, 2010). In this context, the present review sought to examine the impact of transportation strategies to reduce emissions of GHG and other pollutants. Few studies have used a rigorous experimental or natural experiment design and current investigations assessed localized interventions (i.e. the congestion schemes in central London and Stockholm) or exceptional measures taken for reducing traffic congestion during the Olympic Games in Atlanta and Beijing. However, simulations and observational

2008). Moreover, low income households and families with young children are likely to be more affected by a potential increase in the cost of food transportation caused by fuel

Overall, this review is limited by the lack of prospective intervention studies that examined the impact of policies aimed at reducing greenhouse gases emissions. In the absence of such studies, simulations provide an interesting alternative to examine the impact of large-scale changes in transportation practices. However, simulations are limited by their reliance on highly uncertain assumptions about factors such as the potential for behavioral change, the occurrence of peak oil, the confounding variables not accounted for, and the magnitude of

While there has been a lot of studies on the correlates of AT, few of these have rigorously examined the factors associated with changes in travel mode. This may partly explain the modest success of interventions that sought to increase the mode share of AT. The findings of the current review nevertheless suggest that more comprehensive interventions tend to be more effective, highlighting the need for better integrated policies to promote AT and

In addition the use of the social-ecological approach for the description of the correlates of AT can be perceived both as a strength and as a limitation. This model allows the consideration of a wide range of factors located at different levels of influence emphasizing the interactions between different variables (McLeroy et al., 1988; Sallis et al., 2006). However, ecological models can be criticized for being too complicated; rendering the

The main strength of the current review is the broadness of its scope which allowed a simultaneous examination of both the environmental and the public health impacts of AT. Further, this review summarizes the increasing evidence supporting the positive association between AT and several indicators of public health including increased PA and physical fitness, and decreased morbidity and all-cause mortality. This body of evidence should therefore be useful for policy-makers to justify investments to promote AT by the perspective of combined environmental and public health benefits. Haines et al. (2010) argue that such policies should be more cost-effective and socially attractive than strategies

The share of the global CO2 emissions attributable to the transport sector is increasing steadily while the level of motorization is rising rapidly in developing countries (International Transport Forum, 2010). In this context, the present review sought to examine the impact of transportation strategies to reduce emissions of GHG and other pollutants. Few studies have used a rigorous experimental or natural experiment design and current investigations assessed localized interventions (i.e. the congestion schemes in central London and Stockholm) or exceptional measures taken for reducing traffic congestion during the Olympic Games in Atlanta and Beijing. However, simulations and observational

identification of intervention targets more difficult (Stokols, 1996).

addressing climate change or public health separately.

scarcity (Kaza et al., 2011).

technological innovations.

reduce the mode share of PT.

**8. Conclusion** 

**7. Limitation and strengths of present review** 

studies provide preliminary evidence suggesting that transport policies could achieve considerable reductions in GHG emissions.

The second objective of this chapter was to summarize the literature on the health benefits of AT. Altogether, the reviewed studies suggest that increasing the mode share of AT may lead to multiple health benefits (i.e. declines in the risk of obesity, cardiovascular diseases, diabetes, hypertension, cancer, etc.). The benefits vary across studies, but importantly, all available health impact assessment studies have concluded that the benefits of increasing the mode share of AT largely outweighed the risks (de Hartog et al., 2010; de Nazelle et al., 2011; Rojas-Rueda et al., 2011). Moreover, controlled trials have consistently shown that commuter cycling led to improved cardiovascular fitness. The findings from another randomized controlled trial suggest that lifestyle-oriented physical activity (such as AT) can result in greater compliance than structured physical activity programs (Dunn et al., 1999).

Despite this wide range of potential benefits, the mode share of AT is exceedingly low in most industrialised countries except The Netherlands, Germany and the Scandinavian countries. This highlights the necessity of developing a better understanding of the correlates of AT in order to increase the effectiveness of interventions to promote AT.

Current models that predict the choice of travel mode (i.e. Panter et al., 2008; Sallis et al., 2006) are rooted in the social ecological theory. Based on this theory, intervening only by trying to convince individuals to change their transport behaviors would likely be effective only among individuals who are highly receptive to this message (i.e. those who have a high degree of concerns about environmental issues). In this context, Banister (2008) argues that a paradigm shift from the "conventional approach" to transport planning (or car-oriented development) towards a sustainable mobility paradigm (or people-oriented development) is urgently needed. However, to achieve such a paradigm shift, it is necessary to gain public acceptability in order to drive social and political action (Banister, 2008).

The example of Northern European countries suggests that it is feasible to achieve public acceptability. Furthermore, the cities that had the most success in increasing the mode share of AT are those that implemented comprehensive multi-level strategies (Pucher et al., 2008, 2010). Such strategies can include a combination of incentives to individuals who adopt active transportation (i.e. tax credits), an increased availability of sidewalks, cycle paths, improved public transport systems, an increased community/organizational support for active transportation (i.e. provision of safe places to store bicycles, showers in workplaces), traffic calming strategies, policies to increase the cost of car use (e.g. increased fuel taxes or reduced availability of car parking), and well-designed social marketing campaigns (Oja & Vuori, 2000; Pucher et al., 2010).

As noted in the 'limitations and strengths' section, there is a clear need for more prospective intervention studies in order to increase the overall quality of evidence. In this regard, future studies should: 1) assess in a more comprehensive fashion the effectiveness of AT interventions and their impact on environmental variables (including GHG emissions) and health-related outcomes; and 2) identify key variables associated with changes in travel modes to inform the design of more effective interventions.

Although stronger proof of effectiveness from experimental studies and natural experiments would increase the quality of evidence and may potentially increase public acceptability, it

The Environmental and Population Health Benefits of Active Transport: A Review 431

Bandura, A. (1986) *Social foundations of thought and action: a social cognitive theory*. Englewoods

Banister, D (2008). The sustainable mobility paradigm. *Transport Policy*, Vol.15, No.2 (March

Bauman, A., Sallis, J. F., Dzewaltowski, D. A. & Owen, N. (2002). Toward a better

Bell, C., Garrard, J., & Swinburn, B. (2006). Active transport to work in Australia: is it all

Bopp, M., Kaczynski, A.T., & Wittman, P. (2011). The relationship of eco-friendly attitudes

Brunekreef, B., & Holgate, S.T. (2002). Air pollution and health. *Lancet*, Vol.360, No.9341

Buliung, R.N., Mitra, R., & Faulkner, G. (2009). Active school transportation in the Greater

*Preventive Medicine*, Vol.48, No.6 (June 2009), pp. 507-512, ISSN 0091-7435 Cavill, N., Kahlmeier, S., Rutter, H., Racioppi, F. & Oja, P. (2008). Economic analyses of

Cervero, R. & Kockelman, K. (1997). Travel demands and the 3Ds: Density, diversity and

Chapman, L. (2007). Transport and climate change: A review. *Journal of Transport Geography*,

Chillón, P., Evenson, K.R., Vaughn, A. & Ward, D.S. (2011). A systematic review of

Colley, R.C, Garriguet, D., Janssen, I., Craig, C.L., Clarke, J. & Tremblay, M.S. (2011).

Colley, R.C, Garriguet, D., Janssen, I., Craig, C.L., Clarke, J. & Tremblay, M.S. (2011).

Vol.17, No.5 (September/October 2011), pp. E9-E17, ISSN 1078-4659 Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, Mass: Harvard University Press. ISBN 0674224566. Brook, R.D., Rajagopalan, S., Pope III, A., Brook, J.R., Bhatnagar, A. et al. (2010). Particulate

understanding of the influences on physical activity: the role of determinants, correlates, causal variables, mediators, moderators, and confounders. *American Journal of Preventive Medicine*, Vol.23 No.2 (August 2002), pp. S5-S14, ISSN 0749-

downhill from here? *Asia Pacific Journal of Public Health*, Vol.18, No.1 (March 2006),

with walking and biking to work*. Journal of Public Health Management and Practice*,

matter air pollution and cardiovascular disease: an update to the scientific statement from the American Heart Association. *Circulation*, Vol.121 (June 2010),

Toronto Area, Canada: An exploration of trends in space and time (1986-2006).

transport infrastructure and policies including health effects related to cycling and walking: A systematic review. *Transport Policy*, Vol.15, No.5 (September, 2008), pp.

design. *Transportation Research Part D: Transport and Environment*, Vol.2, No.3

interventions for promoting active transportation to school. *International Journal of Behavioral Nutrition and Physical Activity*, Vol.8, No.10, ISSN 1479-5868

Physical activity of Canadian children and youth: Accelerometer results from the 2007 to 2009 Canadian Health Measures Survey. *Health Reports*, Vol.22, No.1 (March

Physical activity of Canadian adults: Accelerometer results from the 2007 to 2009

Cliff, New Jersey: Prentice Hall. ISBN 013815614X

2008), pp. 73-80, ISSN 0967-070X

pp. 62-68, ISSN 1010-5395

pp. 2331-2378, ISSN 0009-7322

291-304, ISSN 0967-070X

2011), pp. 15-23, ISSN 0840-6529

(October 2002), pp. 1233-1242, ISSN 0140-6736

(September 1997), pp. 199-219, ISSN 1361-9209

Vol.15, No.5 (September 2007), pp. 354-367, ISSN 0966-6923

http://www.ijbnpa.org/content/pdf/1479-5868-8-10.pdf

3797

should not be viewed as a reason to maintain car-oriented development. First, large increases of motorization in developing countries such as China and India, coupled with the status quo in developed countries are expected to lead to major increases in GHG emissions (Woodcock et al., 2007, 2009). The transport sector already accounts for 23% of worldwide CO2 emissions. Second, exposure to vehicle exhaust gases is strongly associated with cardiovascular morbidity and mortality (Brook et al., 2010). Third, there is already strong evidence supporting that commuter cycling leads to significant increases in cardiovascular fitness, which is an important cardiovascular disease risk factor (Oja et al., 2011). Fourth, due to the safety in number principle, an increase in the mode share of AT typically leads to a decrease in the relative risk of traffic injury for each commuter (Jacobsen, 2003).

Further, there are economic reasons supporting a shift to the sustainable mobility paradigm. First, Stern's (2006) review illustrated that if we do not act now to reduce the magnitude of climate change, the price to pay would approximate 5 to 20% of the global GDP each year. In contrast, reducing greenhouse gas emissions to avoid the worst consequences of climate change would cost only 1% of the global GDP per year. Second, current evidence suggests that investments in AT-related infrastructure are highly cost-effective (Cavill et al., 2008; Gotschi, 2011). Therefore, by attacking several of today's most important environmental and health problems, AT is a powerful strategy that should not be overlooked by policy-makers. Hence, "*Levels of cycling can be seen as a measure of progress towards a healthier sustainable future in both the developed and the developing world*" (Woodcock et al., 2007).

## **9. Acknowledgements**

Richard Larouche is a member of the Healthy Active Living and Obesity Research Group at the Children's Hospital of Eastern Ontario and a doctoral student at the University of Ottawa. He holds a Frederick Banting and Charles Best Canada Graduate Scholarship from the Canadian Institutes of Health Research. Publication of this book chapter was supported by the University of Ottawa Author Fund.

## **10. References**


should not be viewed as a reason to maintain car-oriented development. First, large increases of motorization in developing countries such as China and India, coupled with the status quo in developed countries are expected to lead to major increases in GHG emissions (Woodcock et al., 2007, 2009). The transport sector already accounts for 23% of worldwide CO2 emissions. Second, exposure to vehicle exhaust gases is strongly associated with cardiovascular morbidity and mortality (Brook et al., 2010). Third, there is already strong evidence supporting that commuter cycling leads to significant increases in cardiovascular fitness, which is an important cardiovascular disease risk factor (Oja et al., 2011). Fourth, due to the safety in number principle, an increase in the mode share of AT typically leads to

Further, there are economic reasons supporting a shift to the sustainable mobility paradigm. First, Stern's (2006) review illustrated that if we do not act now to reduce the magnitude of climate change, the price to pay would approximate 5 to 20% of the global GDP each year. In contrast, reducing greenhouse gas emissions to avoid the worst consequences of climate change would cost only 1% of the global GDP per year. Second, current evidence suggests that investments in AT-related infrastructure are highly cost-effective (Cavill et al., 2008; Gotschi, 2011). Therefore, by attacking several of today's most important environmental and health problems, AT is a powerful strategy that should not be overlooked by policy-makers. Hence, "*Levels of cycling can be seen as a measure of progress towards a healthier sustainable future* 

Richard Larouche is a member of the Healthy Active Living and Obesity Research Group at the Children's Hospital of Eastern Ontario and a doctoral student at the University of Ottawa. He holds a Frederick Banting and Charles Best Canada Graduate Scholarship from the Canadian Institutes of Health Research. Publication of this book chapter was supported

Andersen, L.B., Schnohr, P., Schroll, M. & Hein, H.O. (2000). All-cause mortality associated

Andersen, L.B., Wedderkopp, N., Kristensen, P., Moller, N.C., Froberg, K. & Cooper, A.R.

Appleyard, B. (2003). Planning Safe Routes to School. *Planning*, Vol.69, No.5 (May 2003),

Ball, K., Timperio, A., Salmon, J., Giles-Corti, B., Roberts, R. & Crawford, D. (2007). Personal,

with physical activity during leisure time, work, sports, and cycling to work *Archives of Internal Medicine*, Vol.160, No.11 (June 2000), pp. 1621-1628, ISSN 0003-

(2011). Cycling to school and cardiovascular risk factors: a longitudinal study. *Journal of Physical Activity and Health*, Vol.8, No. (November 2011), pp. 1025-1033.

social and environmental determinants of educational inequalities in walking. *Journal of Epidemiology and Community Health*, Vol.61, No.2 (February 2007), pp. 108-

a decrease in the relative risk of traffic injury for each commuter (Jacobsen, 2003).

*in both the developed and the developing world*" (Woodcock et al., 2007).

**9. Acknowledgements** 

**10. References** 

9926

ISSN 1543-3080

pp.34-37. ISSN 0001-2610

114. ISSN 0143-005X

by the University of Ottawa Author Fund.


The Environmental and Population Health Benefits of Active Transport: A Review 433

Faulkner, G.E.J., Richichi, V., Buliung, R.N., Fusco, C., & Moola, F. (2010). What's "quickest

Franco, M., Ordunez, P., Caballero, B., & Cooper, R.S. (2008). Obesity reduction and its

Frank, L.D., Stone, B.J., & Bachman, W. (2000). Linking land use with household vehicle

Frank, L.D., Saelens, B.E., Powell, K.E. & Chapman, J.E. (2007). Stepping towards causation:

Frank, L.D., Greenwald, M.J., Winkelman, S., Chapman, J. & Kavage, S. (2010). Carbonless

Friedenreich, C. M., & Orenstein, M. R. (2002). Physical activity and cancer prevention:

Friedman, M.S., Powell, K.E., Hutwagner, L., Graham, L.M. & Teague, W.G. (2001). Impact

Frumkin, H., Frank, L. & Jackson, R. *Urban sprawl and public health: Designing, planning, and building for healthy communities*. Washington, DC: Island Press. ISBN 1-55963-305-0 Garrard, J., Rose, G., Lo, S.J. (2008). Promoting transportation cycling for women: the role of

Giles-Corti, B. (2006). People or places: what should be the target? *Journal of Science and Medicine in Sports*, Vol.9, No.5 (October 2006), pp. 357-366, ISSN 1440-2440 Gordon-Larsen, P., Boone-Heinonen, J., Sidney, S., Sternfeld, B., Jacobs, D.R. Jr. & Lewis,

Gotschi, T. (2009). Costs and benefits of bicycling investments in Portland, Oregon. *Journal of* 

Grize, L., Bringolf-Isler, B., Martin, E., Braun-Fahrländer, C. (2010). Trends in active

http://www.biomedcentral.com/content/pdf/1479-5868-7-62.pdf

4363

ISSN 0091-7435

ISSN 0091-7435

0003-9926

1543-3080

2007), pp.1898-1914. ISSN 0277-9536

(November 2002), pp. 3456S-3364S, ISSN 0022-3166

and easiest"?: parental decision making about school trip mode. *International Journal of Behavioral Nutrition and Physical Activity*, Vol.7, No.62, ISSN 1479-5868.

possible consequences : what can we learn from Cuba's special period? *Canadian Medical Association Journal*, Vol.178, No.8 (April 2008), pp.1032-1034, ISSN 0820-3946

emissions in the central Puget Sound: Methodological framework and findings. *Transportation Research Part D*, Vol.5, No.3 (May 2000), pp. 173-190, ISSN 1361-9209 Frank, L.D., Sallis, J.F., Conway, T.L., Chapman, J.E., Saelens, B.E. & Bachman, W. (2006).

Many pathways from land use to health – associations between neighborhood walkability and active transportation, body mass index, and air quality. *Journal of the American Planning Association*, Vol.72, No.1 (Winter 2006), pp. 75-87. ISSN 0194-

do built environments or neighborhood and travel preferences explain physical activity, driving, and obesity? *Social Science & Medicine*, Vol.65, No.9 (November

footprints: promoting health and climate stabilization through active transportation. *Preventive Medicine*, Vol.50, No.1 (Suppl.) (January 2010), pp.99-105,

Etiologic evidence and biological mechanisms. *Journal of Nutrition*, Vol.132, No.11

of changes in transportation and commuting behaviors during the 1996 Summer Olympic Games in Atlanta on air quality and childhood asthma. *Journal of the American Medical Association*, Vol.285, No.7 (July 2001), pp. 897-905. ISSN 0098-7484

bicycle infrastructure. *Preventive Medicine*, Vol.46, No.1 (January 2008), pp.55-59,

C.E. (2009). Active commuting and cardiovascular disease risk: The CARDIA study. *Archives of Internal Medicine*, Vol.169, No.13 (July 2009), pp. 1216-1223, ISSN

*Physical Activity and Health*, Vol.8, No.1 (Suppl.) (January 2010), pp. S49-S58, ISSN

transport to school among Swiss school children and its associated factors: three

Canadian Health Measures Survey. *Health Reports*, Vol.22, No.1 (March 2011), pp. 7- 14, ISSN 0840-6529


Cooper, A.R., Wedderkopp, N., Jago, R., Kristensen, P.L., Moller, N.C., et al. (2008).

de Geus, B., de Bourdeaudhuij, I., Jannes, C., & Meeusen, R. (2008). Psychosocial and

de Geus, B., Joncheere, J., & Meeusen, R. (2009).Commuter cycling: effect on physical

de Hartog, J.J., Boogaard, H., Nijland, H. & Hoek, G. (2010). Do the health benefits of cycling

de Nazelle, A., Morton, B.J., Jerret, M., & Crawford-Brown, D. (2010). Short trips: an

de Nazelle, A., Nieuwenhuijsen, M.J., Anto, J.M., Brauer, M., Briggs, D., et al. (2011).

DeMaio, P. (2009). Bike-sharing: History, impacts, models of provision and future. *Journal of* 

Demers, M., (2008) Pour une ville qui marche: Aménagement urbain et santé. Montréal,

Dill, J., & Carr, T. (2003). Bicycle commuting and facilities in major U.S. cities: if you build

Dunn, A.L., Marcus, B.H., Kampert, J.B., Garcia, M.E., Kohl III, H.W. & Blair, S.N. (1999).

*Medical Association*, Vol.281, No.4 (January 1999), 327-334. ISSN 1538-3598 Elvik, R. (2009). The non-linearity of risk and the promotion of environmentally sustainable

Ewing, R., Schmid, T., Killingsworth, R., Zlot, A., & Raudenbush, S. (2003). Relationship

Vol.15, No.8 (December 2010), pp.451-457, ISSN 1361-9209

*Public Transport*, Vol.12, No.4, pp. 41-56, ISSN 1077-291X

*Medicine*, Vol.48, No.3 (September 2008), pp. 324-328, ISSN 0091-7435 Davison, K.K., Werder, J.L, & Lawson, C.T. (2008). Children's active commuting to school:

14, ISSN 0840-6529

0268-1153

0001-4575

(July 2008), pp. 1-11, ISSN 1545-1151

(April 2009), 179-187, ISSN 1600-838

pp. 1109-1116, ISSN 0091-6765

2011), pp. 766-777, ISSN 0160-4120

2003), pp. 116-123, ISSN 0361-1981

Québec : Écosociété ISBN 978-2-923165-35-6

Canadian Health Measures Survey. *Health Reports*, Vol.22, No.1 (March 2011), pp. 7-

Longitudinal associations of cycling to school with adolescent fitness. *Preventive* 

current knowledge and future directions. *Preventing Chronic Diseases*, Vol.5, No.3

environmental factors associated with cycling for transport among a working population. *Health Education Research*, Vol.23, No.4 (August 2008), pp. 697-708, ISSN

performance in untrained men and women in Flanders: minimum dose to improve indexes of fitness. *Scandinavian Journal of Medicine & Science in Sports*, Vol. 19, No.2

outweigh the risks? *Environmental Health Perspectives*, Vol.118, No.8 (August 2010),

opportunity for reducing mobile-source emissions? *Transportation Research Part D*,

Improving health through policies that promote active travel: A review of evidence to support health impact assessment. *Environment International*, Vol.37, No.4 (May

them, commuters will use them. *Transportation Research Record*, Vol.1828 (January,

Comparison of lifestyle and structured interventions to increase physical activity and cardiorespiratory fitness: a randomized controlled trial. *Journal of the American* 

transport. *Accident Analysis & Prevention*, Vol.41, No.4 (July 2009), pp. 849-855, ISSN

between urban sprawl and physical activity, obesity and morbidity. *American Journal of Health Promotion*, Vol.18, No.1 (January 2003), pp. 47-57 ISSN 0890-1171 Faulkner, G.E.J., Buliung, R.N., Flora, P.K. & Fusco, C. (2009). Active school transport,

physical activity levels and body weight of children and youth: A systematic Review. *Preventive Medicine*, Vol.48, No.1 (January 2009), pp. 3-8, ISSN 0091-7435


The Environmental and Population Health Benefits of Active Transport: A Review 435

Humpel, N., Owen, N, Iverson, D., Leslie, E. & Bauman, A. (2004). Perceived environmental

Jacobs, J. (1961). The death and life of great American cities. New York: Random House.

Jacobson, P.L. (2003). Safety in numbers: more walkers and bicyclists, safer walking and

Jéquier, E. & Tappy, L. (1999). Regulation of body weight in humans. *Physiological Reviews*,

Johanssen, C., Burman, L. & Forsberg, B. (2009). The effects of congestions tax on air quality

Kaza, N., Knaap, G.J., Knaap, I., & Lewis, R. (2011). Peak oil, urban form, and public health:

Keith, S.W., Redden, D.T., Katzmarzyk, P.T., Boggiano, M.M., Hanlon, E.C. et al. (2006).

King, A.C., Stokols, D., Tahen, E., Brassington, G.S. & Killingsworth, R. (2002). Theoretical

International Transport Forum (2010). *Reducing transport greenhouse gas emissions: Trends and* 

Larouche, R., Saunders, T.J., Faulkner, G.E.J., Colley, R.C. & Tremblay, M.S. (submitted).

Lemieux, M., & Godin, G. (2009). How well do cognitive and environment variables predict

Levy, J.I., Buonocore, J.J., & von Stackelberg, K. (2010). Evaluation of the public health

No.65, ISSN 1476-069X http://www.ehjournal.net/content/9/1/65 Lobstein, T., Baur L. & Uauy, R. (2004). Obesity in children and young people: a crisis in public health. *Obesity Reviews,* Vol.5 No.1 Suppl. (May 2004), 4-85, ISSN 1467-7881 Manson, J.E., Greenland, P.E., LaCroix, A.Z., Stefanick, M.L., Mouton, C.P., et al. (2002).

 http://www.internationaltransportforum.org/Pub/pdf/10GHGTrends.pdf Larouche, R. & Trudeau, F. (2010). Étude des impacts du transport actif sur la pratique

Vol.25, No.5 (November 2010), pp. 227-237. ISSN 0765-1597

*Physical Activity and Health*. ISSN 1543-3080

Vol.79, No.2 (April 1999), pp. 451-480, ISSN 0031-9333

(September 2011), pp.1598-1606, ISSN 1541-0048

3797

8047

ISBN 9780679741954

ISSN 1352-2310

ISSN 0307-0365

5868-6-12.pdf

S15-S25, ISSN 0749-3797

*data 2010*. Leipzig, Germany.

attributes, residential location, and walking for particular purposes. *American Journal of Preventive Medicine*, Vol.26, No.2 (February 2004), pp. 119-125, ISSN 0749-

bicycling. *Injury Prevention*, Vol.9, No.3 (September 2003), pp. 205-209, ISSN 1353-

and health. *Atmospheric Environment*, Vol.43, No.31 (October 2009), pp. 4843-4854,

exploring the connections. *American Journal of Public Health*, Vol.101, No.9

Putative contributors to the secular increase in obesity: exploring the roads less travelled. *International Journal of Obesity*, Vol.30, No.11 (June 2006), pp. 1585-1594,

approaches to the promotion of physical activity: forging a transdisciplinary paradigm. *American Journal of Preventive Medicine*, Vol.22, No.2 (February 2002), pp.

d'activités physiques et la santé et de ses principaux déterminants. *Science & Sports*,

Associations between active school transport and physical activity, body composition and cardiovascular fitness: a systematic review of 57 studies. *Journal of* 

active commuting? *International Journal of Behavioral Nutrition and Physical Activity*, Vol.6, No.12, ISSN 1479-5868 http://www.biomedcentral.com/content/pdf/1479-

impacts of traffic congestion: A health risk assessment. *Environmental Health*, Vol.9,

Walking compared with vigorous exercise for the prevention of cardiovascular

cross-sectional surveys 1994, 2000 and 2005. *International Journal of Behavioral Nutrition and Physical Activity*, Vol.7, No.28, ISSN 1479-5868. http://www.ijbnpa.org/content/7/1/28.


Haines, A., McMichael, A.J., Smith, K.R., Roberts, I., Woodcock, J. et al. (2010). Public health

Hamer, M. & Chida, Y. (2008). Active commuting and cardiovascular risk: a meta-analytic

Handy, S.L., Cao, X. & Mokhtarian, P.L. (2005). Correlation or causality between the built

Handy, S.L., Cao, X., & Mokhtarian, P.L. (2006). Self-selection in the relationship between

Hanlon, P., & McCartney, G. (2008). Peak oil: will it be public health's greatest challenge. *Public Health*, Vol.122, No.7 (July 2008), pp. 647-652, ISSN 0033-3506 Harris, K.C., Kuramoto, L.K., Schulzer, M, & Retallack, J.E. (2009). Effect of school-based

Heelan, K.A., Abbey, B.M., Donnely, JE., Mayo, M.S. & Welk, G.J. (2009). Evaluation of a

Hinckson, E.A. & Badland, H.M. (2011). School travel plans: Preliminary evidence for

Hu, G., Sarti, C., Jousilahti, P., Silventoinen, K., Barengo, N.C. & Tuomilehto, J. (2005).

*Stroke*, Vol.36, No.9 (September 2005), pp. 1994-1999, ISSN 0039-2499 Hu, G., Jousilahti, P., Borodulin, K., Barengo, N.C., Lakka, T.A., Nissinen, A. & Tuomilehto,

*Atherosclerosis*, Vol.194, No.2 (October 2007), pp. 490-497, ISSN 0021-9150 Hume, C., Timperio, A., Salmon, J., Carver, A., Giles-Corti, B. & Crawford, D. (2009).

Vol.72, No.1 (Winter 2006), pp.55-74. ISSN 0194-4363

(September 2007), pp. 921, ISSN 0140-6736

http://www.ijbnpa.org/content/7/1/28.

2114, ISSN 0140-6736

1361-9209.

0820-3946

ISSN 0195-9131

195-200, ISSN 0749-3797

cross-sectional surveys 1994, 2000 and 2005. *International Journal of Behavioral Nutrition and Physical Activity*, Vol.7, No.28, ISSN 1479-5868.

benefits of strategies to reduce greenhouse-gas emissions: Overview and implications for policy makers. *Lancet*, Vol.374, No.9707 (January 2010), pp. 2104-

review. *Preventive Medicine*, Vol.46, No.1 (January 2008), pp. S9-S13, ISSN 0091-7435

environment and travel behavior? Evidence from Northern California. *Transportation Research Part D*, Vol.10, No.6 (November 2005), pp. 427-444, ISSN

the built environment and walking. *Journal of the American Planning Association*,

physical activity interventions on body mass index in children: a meta-analysis. *Canadian Medical Association Journal*, Vol.180, No.7 (July 2008), pp. 719-726, ISSN

walking school bus for promoting physical activity in youth. *Journal of Physical Activity and Health*, Vol.6, No.5 (September 2009), pp. 560-567, ISSN 1543-3080 Hendriksen, I.J.M., Zuiderveld, B., Kemper, H.C.G. & Bezemer, P.D. (2000). Effect of

commuter cycling on physical performance of male and female employees. *Medicine & Science in Sports and Exercise*, Vol.32, No.2 (February 2000), pp. 504-510.

changing school-related travel patterns in elementary school children. *American Journal of Health Promotion*, Vol.25, No.6 (July 2011), pp.368-371, ISSN 0890-1171 Horton, R. (2007). Righting the balance: energy for health. *Lancet*, Vol.370, No.9591

Leisure time, occupational, and commuting physical activity and the risk of stroke.

J. (2007). Leisure time, occupational, and commuting physical activity in relation to coronary heart disease among middle-aged Finnish men and women.

Walking and cycling to school: Predictors of increase among children and adolescents. *American Journal of Preventive Medicine*, Vol.36, No.3 (March 2009), pp.


http://www.internationaltransportforum.org/Pub/pdf/10GHGTrends.pdf


The Environmental and Population Health Benefits of Active Transport: A Review 437

Oja, P. & Vuori, I., & Paronen, O. (1998). Daily walking and cycling to work: their utility as

Oja, P. & Vuori, I. (2000). *Promotion of transport walking and cycling in Europe: strategy* 

Oja, P., Titze, S., Bauman, A., de Geus, A., Krenn, P., Reger-Nash, B., & Kohlberger, T.

Owen, N., Humpel, N., Leslie, E., Bauman, A. & Sallis, J.F. (2004). Understanding

Pabayo, R. & Gauvin, L. (2008). Proportion of students who use various modes of

Pabayo, R., Gauvin, L. & Barnett, T.A. (2011). Longitudinal changes in active transportation

Panter, J.R., Jones, A.P., van Sluijs & E.M.F. (2008). Environmental determinants of active

Panter, J.R., Jones, A.P., van Sluijs, E.M.F. & Griffin, S.J. (2011). The influence of distance to

Pucher, J., Dill, J. & Handy, S. (2010). Infrastructures, programs and policies to increase

Pucher, J., Buehler, R., Bassett, D.R. Jr. & Dannenberg, A.L. (2010). Walking and cycling to

Rojas-Rueda, D., de Nazelle, A., Tainio, M. & Nieuwenhuijsen, M.J. (2011). The health risks

*Behavioral Nutrition and Physical Activity*, Vol.5, No.34, ISSN 1479-5868

*Exercise Science*, Vol.23, No.1 (March 2011), pp. 72-86, ISSN 1541-0048 Pucher, J. & Buehler, R. (2008). Making cycling irresistible: Lessons from The Netherlands,

(April 1998), pp. S87-S94, ISSN 0738-3991

(February 1993), pp.538-545, ISSN 0028-4793

ng%20Strategy%20.pdf

63-66, ISSN 0091-7435

http://www.ijbnpa.org/content/5/1/34/

pp. S106-S125, ISSN 0091-7435

ISSN 0031-4005

ISSN 0899-8493

1992, ISSN 1541-0048

2011, ISSN 0959-8138

health enhancing physical activity. *Patient Education and Counselling*, Vol.33, No.1

*directions*. Tampere, Finland: UKK Institute for Health Promotion Research. http://www.panh.ch/hepaeurope/materials/HEPA%20Walking%20and%20Cycli

(2011). Health benefits of cycling: Systematic review. *Scandinavian Journal of Medicine & Science in Sports*, Vol.21, No.4 (August 2011), pp.496-509, ISSN 1600-838

environmental influences on walking: Review and research agenda. *American Journal of Preventive Medicine*, Vol.27, No.1 (July 2004), pp. 67-76, ISSN 0749-3797 Paffenbarger, R.S. Jr., Hyde, R.T., Wing, A.L., Lee, I., Jung, D.L. & Kampert, J.B. (1993). The

association of changes in physical activity level and other lifestyle characteristics with mortality among men. *New England Journal of Medicine*, Vol.328, No.8

transportation to and from school in a representative population-based sample of children and adolescents, 1999. *Preventive Medicine*, Vol.46, No.1 (January 2008), pp.

to school in Canadian youth aged 6 through 16 years. *Pediatrics*, Epub 2011 Jul 4.

travel in youth: A review and framework for future research. *International Journal of* 

school on the association between active commuting and physical activity. *Pediatric* 

Denmark and Germany. *Transport Reviews*, Vol.28, No.4 (July 2008), pp. 495-528,

bicycling: an international review. *Preventive Medicine*, Vol.50, No.1 (January 2010),

health: A comparison of recent evidence from cities, states, and international studies. *American Journal of Public Health*. Vol.100, No.10 (October 2010), pp. 1986-

and benefits of cycling in urban environments compared with car use: health impact assessment study. *British Medical Journal*. Epub ahead of print, August 4,

events in women. *New England Journal of Medicine*, Vol.347, No.10 (September 2002), pp. 716-725. ISSN 0028-4793


Mapes, J. (2009). Pedaling revolution: how cyclists are changing American cities. Corvallis,

Marshall, J.D., Brauer, M. & Frank, L.D. (2009). Healthy neighborhoods: walkability and air

Marshall, J.D., Wilson, R.D., Meyer, K.L., Rajangam, S.K., McDonald, N.C. & Wilson, E.J.

Matthews, C.E., Juri, A.L., Shu, X.O., Li, H.L., Yang, G., Li, Q., Gao, Y.T. & Zheng, W. (2007).

McDonald, N.C. (2007). Active transportation to school: Trends among U.S. schoolchildren,

McLeroy, K.R., Bibeau, D., Steckler, A. & Glanz, K. (1988). An ecological perspective on

McMichael, A.J., Butler, C.D. (2011). Promoting global population health while constraining

McMillan, T.E. (2005). Urban form and a child's trip to school: The current literature and a

Morency, P. & Demers, M. (2009). Active transportation as a way to increase physical

Muniz, I., & Galindo, A. (2005). Urban form and the ecological footprint of commuting. The

Ogilvie, D., Egan, M., Hamilton, V. & Petticrew, M. (2004). Promoting walking and cycling

Ogilvie, D., Mitchell, R., Mutrie, N., Petticrew, M., & Platt, S. (2008). Personal and

Oja, P., Mänttäri, A., Heinonen, A., Kukkonen-Harjula, K., Laukanen, R., Pasanen, M. &

No.43, ISSN 1479-5868 http://www.ijbnpa.org/content/5/1/43

OR: Oregon State University Press, ISBN 978-0-87071-419-1

pp. 716-725. ISSN 0028-4793

pp.1752-1759, ISSN 0091-6765

2007), pp.1343-1350, ISSN 0002-9262

1543, ISSN 0013-936X

516, ISSN 0749-3797

ISSN 0921-8009

ISSN 1600-838

pp. 351-377, ISSN 0195-8402

pp. 179-197, ISSN 0136-9529

2005), pp. 440-456, ISSN 0885-4122

2010), pp. 421-427, ISSN 1365-2214

No.7469 (October 2004), pp.763-766, ISSN 0959-8146

events in women. *New England Journal of Medicine*, Vol.347, No.10 (September 2002),

pollution. *Environmental Health Perspectives*, Vol.117, No.11 (November 2009),

(2010). Vehicle emissions during children's school commuting: Impact of education policy. *Environmental Science & Technology*, Vol.44, No.5 (January 2010), pp. 1537-

Influence f exercise, walking, cycling, and overall nonexercise physical activity on mortality in Chinese women. *American Journal of Epidemiology*, Vol.165, No.12 (June

1969-2001. *American Journal of Preventive Medicine*, Vol.32, No.6 (June 2007), pp. 509-

health promotion programs. *Health Education Quarterly*, Vol.15, No.4 (Winter 1988),

the environmental footprint. *Annual Reviews of Public Health*, Vol.32, (April 2011),

framework for future research. *Journal of Planning Literature*, Vol.19, No.4 (May

activity among children. *Child: Care, Health and Development*, Vol. 36, No.3 (May

case of Barcelona. *Ecological Economics*, Vol.55, No.4 (December 2005), pp. 499-514,

as an alternative to using cars: systematic review. *British Medical Journal*, Vol.329,

environmental correlates of active travel and physical activity in a deprived urban population. *International Journal of Behavioral Nutrition and Physical Activity*, Vol.5,

Vuori, I. (1991). Physiological effects of walking and cycling to work. *Scandinavian Journal of Medicine & Science in Sports*, Vol.1, No.3 (September 1991), pp. 151-157,


The Environmental and Population Health Benefits of Active Transport: A Review 439

Turcotte, M. (2008). Dépendance à l'automobile dans les quartiers urbains. Tendances

Unger, N., Bond, T.C., Wang, J.S., Koch, D.M., Menon, S. et al. (2010). Attribution of climate

US Environmental Protection Agency. (1999). *The transportation and environmental impacts of* 

van Sluijs, E.M.F., Fearne, V.A., Mattocks, C., Riddoch, C., Griffin, S.J., Ness, A. (2009). The

Wang, Y., Hao, J., McElroy, M.B., Munger, J.W., Ma, H., Chen, D., Nielsen, C. P. (2009).

Wang, Y. & Lobstein, T. (2006). Worldwide trends in childhood overweight and obesity. *International Journal of Pediatric Obesity*, Vol.1, No.1, pp. 11-25, ISSN 1747-7166 Weiss, H. & Bradley, R.S. (2001). What drives societal collapse? *Science*, Vol.291, No.5504

Wen, L.M. & Rissel, C. (2008). Inverse associations between cycling to work, public

Wen, L.M., Kite, J., & Rissel, C. (2010). Is there a role for workplaces for reducing employees'

Woodcock, J., Banister, D., Edwards, P., Prentice, A.M. & Roberts, I. (2007). Energy and

Woodcock, J., Edwards, P., Tonne, C., Armstrong, B.G., Ashiru, O. et al. (2009). Public health

*Lancet*, Vol.374, No.9705, pp.1930-1943 (December 2009), ISSN 0140-6736 Wong, B.Y.M., Faulkner, G. & Buliung, R. (2011). GIS measured environmental correlates of

Australia. *BMC Public Health*, Vol.10, No.50, ISSN 1471-2458

http://www.biomedcentral.com/content/pdf/1471-2458-10-50.pdf Wennberg, P., Wensley, F., Johansson, L., Boman, K., Di Angelantonio, E., et al. (2010).

Agency. http://www.epa.gov/smartgrowth/pdf/infill\_greenfield.pdf van der Ploeg, H.P., Merom, D., Corpuz, G., & Bauman, A.E. (2008). Trends in Australian

*Medicine*, Vol.46, No.1 (January 2008), pp. S60-S62, ISSN 0091-7435

forcing to economic sectors. *Proceedings of the National Academy of Science*, Vol.107,

*infill vs. greenfield development: a comparative case study analysis*. Washington, DC: Urban and Economic Development Division, U.S. Environmental Protection

children travelling to school 1971-2003: Burning petrol or carbohydrates? *Preventive* 

contribution of active travel to children's physical activity levels: Cross-sectional results from the ALSPAC study. *Preventive Medicine*, Vol.48, No.6 (June 2009), pp.

Ozone air quality during the 2008 Beijing Olympics: effectiveness of emission restrictions. *Atmospheric Chemistry and Physics*, Vol.9, No.3 (January 2007), pp.5237-

transport, and overweight and obesity: findings from a population based study in Australia. *Preventive Medicine*, Vol.46, No.1 (January 2008), pp.29-32, ISSN 0091-

driving to work? Findings from a cross-sectional survey from inner-west Sydney,

Reduced risk of myocardial infarction related to active commuting: inflammatory and haemostatic effects are potential major mediating mechanisms. *European Journal of Cardiovascular Prevention and Rehabilitation*, Vol.17, No. 1 (February 2010),

transport. *Lancet*, Vol.370 No.9592 (September 2007), pp. 1078-1088, ISSN 0140-6736

benefits of strategies to reduce greenhouse-gas emissions: urban land transport.

active school transport: A systematic review of 14 studies. *International Journal of* 

Sociales Canadiennes, Vol.1, No.1, p.21-32, ISSN 1481-1642 http://www.statcan.gc.ca/pub/11-008-x/2008001/article/10503-fra.pdf

No.8 (February 2010), pp. 3382-3387, ISSN 1091-6490

519-524, ISSN 0091-7435

5251, ISSN 1352-2310

pp.56-62, ISSN 1741-8267

7435

(January 2001), pp.609-610, ISSN 0036-8075


Saelens, B.E., Sallis, J.F., Black, J.B., & Chen, D. (2003). Neighborhood-based differences in

Saelens, B.E. & Handy, S.L. (2008). Built environment correlates of walking: A review.

Sallis, J.F., Cervero, R.B., Ascher, W., Henderson, K.A., Kraft, M.K. & Kerr, J. (2006). An

Shannon, T., Giles-Corti, B., Pikora, T., Bulsara, M., Shilton, T., & Bull, F. (2006). Active

Sigal, R.J., Kenny, G.P., Wasserman, D.H., Castaneda-Sceppa, D. & White, R.D. (2006).

Stern, N. (2006). *Stern review on the economics of climate change*. London: HM Treasury. http://mudancasclimaticas.cptec.inpe.br/~rmclima/pdfs/destaques/sternreview

Stokols, D. (1996). Translating social ecological theory into guidelines for community health

Tin Tin, S., Woodward, A., Thornley, S. & Ameratunga, S. (2011). Regional variations in

Titze, S., Stronneger, W.J., Janschitz, S., & Oja, P. (2007). Environmental, social and personal

Tonne, C., Beevers, S., Armstrong, B., Kelly, F. & Wilkinson, P. (2008). Air pollution and

Troiano, R.P., Berrigan, D., Dodd, K.W., Mâsse, L.C., Tilert, T. & McDowell, M. (2008).

Troped, P., Saunders, R.P., Pate, R.R., Reininger, B., & Addy, C.L. (2003). Correlates of

*Activity and Health*, Vol.4, No.1 (January 2007), pp. 66-79, ISSN 1543-3080 Titze, S., Stronneger, W.J., Janschitz, S., & Oja, P. (2008). Association of built-environment,

Vol.93, No.9 (September 2003), pp. 1552-1558, ISSN 1541-0048

*Health*, Vol.27, (April 2006), pp. 297-322, ISSN 0163-7525

pp. S550-S566, ISSN 0195-9131

070X

1438, ISSN 0149-5992

\_report\_complete.pdf

pp. 282-298, ISSN 0890-1171

pp.357-363. ISSN 1753-6405

pp. 252-259, ISSN 0091-7435

pp. 620-627, ISSN 1470-7926

0091-7435

physical activity: an environment scale evaluation. *American Journal of Public Health*,

*Medicine and Science in Sports & Exercise*, Vol.40, No.7 (Supplement 7) (July 2008),

ecological approach to creating active living communities. *Annual Reviews of Public* 

commuting in a university setting: assessing commuting habits and potential for modal change. *Transport Policy*, Vol.13, No.3 (May 2006), pp. 240-253, ISSN 0967-

Physical activity / Exercise and type 2 diabetes: A consensus statement from the American Diabetes Association. *Diabetes Care*, Vol.29, No.6 (June 2006), pp. 1433-

promotion. *American Journal of Health Promotion*, Vol.10, No.4 (March-April 1996),

pedal cyclist injuries in New Zealand: safety in numbers or risk in scarcity? *Australian and New Zealand Journal of Public Health*, Vol.35, No.4 (August 2011),

correlates of cycling for transportation in a student population. *Journal of Physical* 

social-environment and personal factors with bicycling as a mode of transportation among Austrian city dwellers. *Preventive Medicine*, Vol.47, No.3 (September 2008),

mortality benefits of the London Congestion Charge: spatial and socioeconomic inequalities. *Occupational Environmental Medicine*, Vol.65, No.9 (September 2008),

Physical activity in the United States measured by accelerometers. *Medicine & Science in Sports & Exercise*. Vol.40, No.1 (January 2008), pp. 181-188, ISSN 0195-9131

recreational and transportation physical activity among adults in a New England community. *Preventive Medicine*, Vol.37, No.4 (October 2003), pp.304-310, ISSN

	- http://www.biomedcentral.com/content/pdf/1471-2458-10-50.pdf

**20**

*USA* 

**Arctic Sea Ice Decline** 

Julienne C. Stroeve and Walter Meier

*Sciences, University of Colorado, Boulder, CO,* 

*National Snow and Ice Data Center, Cooperative Institute for Research in Environmental* 

The Arctic is currently undergoing rapid environmental changes (ACIA, 2005; IPCC, 2007). One of the most striking of these changes is the decline in the floating sea ice cover. Sea ice is frozen sea water that forms in winter as temperatures drop and darkness sets over the northern high latitudes. At its winter maximum, roughly 14 to 16 million km2 of the Northern Hemisphere ocean surface is covered by sea ice, extending as far south as Newfoundland, Canada (at 50o latitude) along the eastern coast of Canada and as far south as Bohai Bay, China (at 38o latitude) on the Eurasian side. During summer as temperatures warm, the ice cover shrinks to about half its winter size by September, covering on average 6

Arctic sea ice is an important regulator of the exchange of heat and moisture between the atmosphere and the ocean. The presence of the sea ice cover insulates the relatively warm ocean from the colder atmosphere and because of its high reflectivity, its presence helps to keep the northern high latitudes cool by reflecting a large portion of the sun's energy back to space. Sea ice also exerts a strong influence on global atmospheric and oceanic circulation. Because of the Earth's orientation relative to the sun, the equator receives more incoming solar radiation than the poles. The inequality in the amount of solar radiation received gives rise to a temperature gradient that drives the circulation of air in the atmosphere that

During winter, there is little solar energy and sea ice acts as a very effective insulator, preventing heat in the Arctic Ocean from warming the lower atmosphere. During early spring, as the sun rises in the Arctic, but temperatures remain cold, the snow-covered sea ice remains frozen and may reflect more than 80 percent of the sun's energy back to space. As temperatures rise through the spring and summer, the snow melts, bare ice is exposed and melt ponds form, allowing for more of the sun's energy to be absorbed by the surface. However, even under these melt conditions the ice surface usually reflects 50% or more of the sun's energy. Additionally, since most of the solar energy absorbed by the ice in summer is used to melt ice, near surface air temperatures stay fixed near the freezing point. As the ice melts out completely in summer, the exposed darker ocean surface absorbs roughly 90 percent of the sun's incoming energy. This causes the oceans to warm up. Before the ice can once again reform in autumn, the ocean must first release the heat gained in summer back to

**1. Introduction** 

million km2 (Figure 1).

transports heat from the tropics to the poles.

the atmosphere, warming the lower troposphere.

*Behavioral Nutrition and Physical Activity*, Vol.8, No.39, ISSN 1479-5868 http://www.ijbnpa.org/content/pdf/1479-5868-8-39.pdf


## **Arctic Sea Ice Decline**

Julienne C. Stroeve and Walter Meier

*National Snow and Ice Data Center, Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, CO, USA* 

## **1. Introduction**

440 Greenhouse Gases – Emission, Measurement and Management

Yang, L., Sahlqvist, S., McMinn, A., Griffin, S.J. & Ogilvie, D. (2010) Interventions to

Younger, M., Morrow-Almeida, H.R., Vindigni, S.M., & Dannenberg, A.L. (2008). The built

Zhou, W. & Szyliowicz, J.S. (2006). The development and current status of China's

http://www.ijbnpa.org/content/pdf/1479-5868-8-39.pdf

(October 2010), ISSN 0959-8146

2006), pp. 10-16, ISSN 1352-7614

3797

*Behavioral Nutrition and Physical Activity*, Vol.8, No.39, ISSN 1479-5868

promote cycling: systematic review. *British Medical Journal*, Vol.341, No.5293

environment, climate change, and health: opportunities for co-benefits. *American Journal of Preventive Medicine*, Vol.35, No.5 (November 2008), 517-526, ISSN 0749-

transportation system. *World Transport Policy & Practice*, Vol.12 No.4 (Setember

The Arctic is currently undergoing rapid environmental changes (ACIA, 2005; IPCC, 2007). One of the most striking of these changes is the decline in the floating sea ice cover. Sea ice is frozen sea water that forms in winter as temperatures drop and darkness sets over the northern high latitudes. At its winter maximum, roughly 14 to 16 million km2 of the Northern Hemisphere ocean surface is covered by sea ice, extending as far south as Newfoundland, Canada (at 50o latitude) along the eastern coast of Canada and as far south as Bohai Bay, China (at 38o latitude) on the Eurasian side. During summer as temperatures warm, the ice cover shrinks to about half its winter size by September, covering on average 6 million km2 (Figure 1).

Arctic sea ice is an important regulator of the exchange of heat and moisture between the atmosphere and the ocean. The presence of the sea ice cover insulates the relatively warm ocean from the colder atmosphere and because of its high reflectivity, its presence helps to keep the northern high latitudes cool by reflecting a large portion of the sun's energy back to space. Sea ice also exerts a strong influence on global atmospheric and oceanic circulation. Because of the Earth's orientation relative to the sun, the equator receives more incoming solar radiation than the poles. The inequality in the amount of solar radiation received gives rise to a temperature gradient that drives the circulation of air in the atmosphere that transports heat from the tropics to the poles.

During winter, there is little solar energy and sea ice acts as a very effective insulator, preventing heat in the Arctic Ocean from warming the lower atmosphere. During early spring, as the sun rises in the Arctic, but temperatures remain cold, the snow-covered sea ice remains frozen and may reflect more than 80 percent of the sun's energy back to space. As temperatures rise through the spring and summer, the snow melts, bare ice is exposed and melt ponds form, allowing for more of the sun's energy to be absorbed by the surface. However, even under these melt conditions the ice surface usually reflects 50% or more of the sun's energy. Additionally, since most of the solar energy absorbed by the ice in summer is used to melt ice, near surface air temperatures stay fixed near the freezing point. As the ice melts out completely in summer, the exposed darker ocean surface absorbs roughly 90 percent of the sun's incoming energy. This causes the oceans to warm up. Before the ice can once again reform in autumn, the ocean must first release the heat gained in summer back to the atmosphere, warming the lower troposphere.

Arctic Sea Ice Decline 443

Before the satellite era, it was not easy to obtain reliable measurements of Arctic sea ice. The Arctic is remote and can be an inhospitable place. Early ship, submarine and aircraft observations provide some local information, but it wasn't until the late 1960s/early 1970s that satellites were able to provide the first broad look at changes in the sea ice cover. The modern passive microwave satellite data record further allowed for monitoring of ice conditions regardless of sunlight or cloud cover. Starting with the launch of NASA's Scanning Multichannel Microwave Radiometer (SMMR) satellite in October 1978, and continuing on with a series of Special Sensor Microwave/Imager (SSM/I) instruments on the Defense Meteorological Satellite Program (DMSP) satellites, scientists now have a

Analysis of this data record shows that since the late 1970s, the amount of sea ice covering the ocean has declined during all calendar months (Serreze et al., 2007). The trends are smallest in winter and strongest in late summer and early autumn (Figure 2). However, in recent years, the trend in the end-of-summer ice extent has accelerated. Every year since 2002 has been characterized by extreme September ice extent minima (Stroeve et al., 2008; Comiso et al., 2008; Stroeve et al., 2011a). Through 2001, the linear trend in September ice extent over the satellite record stood at -7.0% per decade. By 2010, it increased to -12.4% per decade. Yet another record or near-record low is expected for September 2011. This represents a ~40% reduction in the surface area of the Arctic Ocean covered by sea ice.

**2. Recent changes in the Arctic sea ice cover** 

consistent time-series of sea ice changes that spans nearly 34 years.

**2.1 Ice extent** 

The strong seasonal cycle of the Arctic sea ice cover not only affects our weather patterns, it also affects human activities and biological habitats. For example, hunters in the Arctic depend on the ice cover as a platform to hunt for Arctic mammals, such as seals, whales, walruses and polar bears. These species themselves depend on sea ice for their habitat, using the ice as places to breed, feed and hunt for food. The timing of the phytoplankton bloom, which supplies energy to the entire Arctic ecosystem, is also regulated by the timing of the ice retreat. Shipping companies are starting to take advantage of less sea ice in the summer to ship materials through the Arctic and the potential for extraction of natural resources, such as oil and natural gas, is also drawing increasing attention.

Over the past few decades, the Arctic has warmed at about twice the rate as the rest of the planet. As a result, significant changes are happening in the Arctic sea ice cover, with potentially large implications not only regionally but also for the global climate. This chapter discusses recent Arctic sea ice changes, the factors responsible for these changes, what climate models project for the future, and our current understanding of the climatic impacts of continued sea ice loss.

Fig. 1. March (left) and September (right) mean seasonal ice cover. The area in grey near the pole is the area not seen by the satellite sensor.

### **2. Recent changes in the Arctic sea ice cover**

#### **2.1 Ice extent**

442 Greenhouse Gases – Emission, Measurement and Management

The strong seasonal cycle of the Arctic sea ice cover not only affects our weather patterns, it also affects human activities and biological habitats. For example, hunters in the Arctic depend on the ice cover as a platform to hunt for Arctic mammals, such as seals, whales, walruses and polar bears. These species themselves depend on sea ice for their habitat, using the ice as places to breed, feed and hunt for food. The timing of the phytoplankton bloom, which supplies energy to the entire Arctic ecosystem, is also regulated by the timing of the ice retreat. Shipping companies are starting to take advantage of less sea ice in the summer to ship materials through the Arctic and the potential for extraction of natural

Over the past few decades, the Arctic has warmed at about twice the rate as the rest of the planet. As a result, significant changes are happening in the Arctic sea ice cover, with potentially large implications not only regionally but also for the global climate. This chapter discusses recent Arctic sea ice changes, the factors responsible for these changes, what climate models project for the future, and our current understanding of the climatic

Fig. 1. March (left) and September (right) mean seasonal ice cover. The area in grey near the

pole is the area not seen by the satellite sensor.

resources, such as oil and natural gas, is also drawing increasing attention.

impacts of continued sea ice loss.

Before the satellite era, it was not easy to obtain reliable measurements of Arctic sea ice. The Arctic is remote and can be an inhospitable place. Early ship, submarine and aircraft observations provide some local information, but it wasn't until the late 1960s/early 1970s that satellites were able to provide the first broad look at changes in the sea ice cover. The modern passive microwave satellite data record further allowed for monitoring of ice conditions regardless of sunlight or cloud cover. Starting with the launch of NASA's Scanning Multichannel Microwave Radiometer (SMMR) satellite in October 1978, and continuing on with a series of Special Sensor Microwave/Imager (SSM/I) instruments on the Defense Meteorological Satellite Program (DMSP) satellites, scientists now have a consistent time-series of sea ice changes that spans nearly 34 years.

Analysis of this data record shows that since the late 1970s, the amount of sea ice covering the ocean has declined during all calendar months (Serreze et al., 2007). The trends are smallest in winter and strongest in late summer and early autumn (Figure 2). However, in recent years, the trend in the end-of-summer ice extent has accelerated. Every year since 2002 has been characterized by extreme September ice extent minima (Stroeve et al., 2008; Comiso et al., 2008; Stroeve et al., 2011a). Through 2001, the linear trend in September ice extent over the satellite record stood at -7.0% per decade. By 2010, it increased to -12.4% per decade. Yet another record or near-record low is expected for September 2011. This represents a ~40% reduction in the surface area of the Arctic Ocean covered by sea ice.

Arctic Sea Ice Decline 445

Fig. 3. Sea ice extent anomaly maps, 2002-2010: Sea ice conditions for the month of September, 2002 through 2010, derived from the NSIDC Sea Ice Index. Each image shows the concentration anomaly (see color key) and the 1979-2000 mean September ice edge (pink line). Nearly every year, the ice edge is well north of its mean position off the coasts of Alaska and Siberia. Image provided by National Snow and Ice Data Center, University of

Colorado, Boulder.

Fig. 2. Monthly mean March (top) and September (bottom) sea ice extent from 1979 through present derived from the passive microwave satellite data record and using the NASA Team sea ice algorithm.

During the 1980s and the 1990s, a low summer sea ice year would often be followed the next year by higher summer extents. In addition, the Arctic-wide low sea ice anomalies tended to be characterized by less ice the Eurasian side (i.e. Laptev or E. Siberian seas) with average or anomalously more sea ice on the Alaskan side (i.e. Chukchi and Beaufort seas) or vice versa. However, ice losses during the past decade have been characterized by pronounced sea ice retreat from both the Alaska and Siberia shores (Figure 3). The last several years have also seen opening up of the northern shipping routes. The last four summers saw opening of the Northern Sea Route, a shipping route that runs along the Russian Arctic coast from Murmansk on the Barents Sea to the Bering Strait and the Far East. Additionally, the Northwest Passage southern route, a waterway amidst the Canadian Arctic Archipelago was open the past five summers.

Fig. 2. Monthly mean March (top) and September (bottom) sea ice extent from 1979 through present derived from the passive microwave satellite data record and using the NASA Team

During the 1980s and the 1990s, a low summer sea ice year would often be followed the next year by higher summer extents. In addition, the Arctic-wide low sea ice anomalies tended to be characterized by less ice the Eurasian side (i.e. Laptev or E. Siberian seas) with average or anomalously more sea ice on the Alaskan side (i.e. Chukchi and Beaufort seas) or vice versa. However, ice losses during the past decade have been characterized by pronounced sea ice retreat from both the Alaska and Siberia shores (Figure 3). The last several years have also seen opening up of the northern shipping routes. The last four summers saw opening of the Northern Sea Route, a shipping route that runs along the Russian Arctic coast from Murmansk on the Barents Sea to the Bering Strait and the Far East. Additionally, the Northwest Passage southern route, a waterway amidst the Canadian Arctic Archipelago

sea ice algorithm.

was open the past five summers.

Fig. 3. Sea ice extent anomaly maps, 2002-2010: Sea ice conditions for the month of September, 2002 through 2010, derived from the NSIDC Sea Ice Index. Each image shows the concentration anomaly (see color key) and the 1979-2000 mean September ice edge (pink line). Nearly every year, the ice edge is well north of its mean position off the coasts of Alaska and Siberia. Image provided by National Snow and Ice Data Center, University of Colorado, Boulder.

Arctic Sea Ice Decline 447

relatively stable prior to 2007 and where long-term survival of sea ice through summer is

Fig. 4. Spatial distribution of ice of different age classes during March (week 11) in 1985 (top, left) and 2011 (top, right). Bottom figure shows the total extent of ice of different age classes for the Arctic Ocean domain. Data from C. Fowler and J. Maslanik, CCAR, University of

considered to be most likely.

Colorado.

#### **2.2 Ice thickness**

At the same time that the spatial coverage of the summer ice cover is shrinking, the ice cover is also thinning. Comparisons between early upward looking sonar aboard submarines (1958-1976) and those from 1993 to 1997 suggested a reduction of 1.3 m in the mean late summer ice thickness over much of the central Arctic Ocean [(Rothrock et al., 1999)]. However, sparse sampling has complicated the interpretation of wide-spread changes in the Arctic ice thickness. Rothrock et al. [(2003)] further analyzed the submarine data in conjunction with model simulations and found that there was a modest recovery in ice thickness after 1996. *In situ* measurements of thickness have been collected during short field experiments at scattered locations. These measurements have proven valuable for validation of submarine and satellite data, but do not provide useful information on the Arctic-wide sea ice thickness.

In recent years, satellites have allowed for further assessment of sea ice thickness changes. Giles et al. [(2008)] and Laxon et al. [(2003)] demonstrate the ability to derive ice thickness estimates from satellite radar altimetry (i.e., from the European Space Agency's (ESA) ERS1/2, Envisat) to an accuracy of 4 to 6 cm, excluding regions thin ice (below 0.5 m). These satellites have a latitudinal limit of 81.5°N but have proven valuable in assessing changes in ice thickness, with high correlations between derived ice thickness fields and submarine estimates of ice thickness (r=0.99) [(Laxon et al., 2003)]. Thickness estimates based on measurements taken by NASA's Ice, Cloud and land Elevation Satellite (ICESat) laser altimeter launched in January 2003 provide a further update on thickness changes during the time when large changes in the summer ice cover have occurred (satellite mission ended In 2009). Using this data in conjunction with the submarine thickness record, Kwok and Rothrock [(2009)] determined that the Arctic mean ice thickness declined from 3.64 meters in 1980 to 1.89 meters in 2008, or a total decline of 1.75 meters. CryoSat-2 is another radar altimeter instrument that was launched in April 2010 by ESA, but covers higher latitudes than the earlier radar altimeter satellites and has a 250 m resolution along the track. ICESat-2 is planned for launch early 2016. Together CryoSat-2 and ICESat-2 are expected to achieve improved accuracy in sea ice thickness retrievals over a broad spatial scale.

#### **2.3 Ice age**

The large decreases in summer sea ice extent in recent years [(e.g., Comiso et al., 2008; Stroeve et al., 2008; Stroeve et al., 2005)] further manifest a fundamental transition from a largely perennial ice cover where ice persists from year to year, to a seasonal coverage with substantial open water during summer [(e.g., Kwok, 2007; Maslanik et al., 2007a; Kwok and Cunningham, 2010)]. Using satellite-derived ice motion fields, the age of an individual ice parcel can be tracked by treating each grid cell that contains ice as an independent Lagrangian particle and advecting the particles at weekly time steps [(Fowler et al., 2004)]. In cases where particles of different ages fall within a single grid cell, the age of the grid cell is assigned to the oldest particle. These fields reveal a dramatic decline during the past 30 years in the Arctic's multiyear ice cover [(Figure 4]). The fraction of total ice extent made up of multiyear ice in March decreased from about 75% in the mid-1980s to 45% in 2011, while the proportion of the oldest ice (i.e. ice 5 years or older) declined from 50% of the multiyear ice pack to 10% [(Maslanik et al., 2011)]. Reductions in old ice now extend into the central Arctic Ocean and adjacent to the Canadian Archipelago, areas where the ice cover was

At the same time that the spatial coverage of the summer ice cover is shrinking, the ice cover is also thinning. Comparisons between early upward looking sonar aboard submarines (1958-1976) and those from 1993 to 1997 suggested a reduction of 1.3 m in the mean late summer ice thickness over much of the central Arctic Ocean [(Rothrock et al., 1999)]. However, sparse sampling has complicated the interpretation of wide-spread changes in the Arctic ice thickness. Rothrock et al. [(2003)] further analyzed the submarine data in conjunction with model simulations and found that there was a modest recovery in ice thickness after 1996. *In situ* measurements of thickness have been collected during short field experiments at scattered locations. These measurements have proven valuable for validation of submarine and satellite data, but do not provide useful information on the

In recent years, satellites have allowed for further assessment of sea ice thickness changes. Giles et al. [(2008)] and Laxon et al. [(2003)] demonstrate the ability to derive ice thickness estimates from satellite radar altimetry (i.e., from the European Space Agency's (ESA) ERS1/2, Envisat) to an accuracy of 4 to 6 cm, excluding regions thin ice (below 0.5 m). These satellites have a latitudinal limit of 81.5°N but have proven valuable in assessing changes in ice thickness, with high correlations between derived ice thickness fields and submarine estimates of ice thickness (r=0.99) [(Laxon et al., 2003)]. Thickness estimates based on measurements taken by NASA's Ice, Cloud and land Elevation Satellite (ICESat) laser altimeter launched in January 2003 provide a further update on thickness changes during the time when large changes in the summer ice cover have occurred (satellite mission ended In 2009). Using this data in conjunction with the submarine thickness record, Kwok and Rothrock [(2009)] determined that the Arctic mean ice thickness declined from 3.64 meters in 1980 to 1.89 meters in 2008, or a total decline of 1.75 meters. CryoSat-2 is another radar altimeter instrument that was launched in April 2010 by ESA, but covers higher latitudes than the earlier radar altimeter satellites and has a 250 m resolution along the track. ICESat-2 is planned for launch early 2016. Together CryoSat-2 and ICESat-2 are expected to achieve

improved accuracy in sea ice thickness retrievals over a broad spatial scale.

The large decreases in summer sea ice extent in recent years [(e.g., Comiso et al., 2008; Stroeve et al., 2008; Stroeve et al., 2005)] further manifest a fundamental transition from a largely perennial ice cover where ice persists from year to year, to a seasonal coverage with substantial open water during summer [(e.g., Kwok, 2007; Maslanik et al., 2007a; Kwok and Cunningham, 2010)]. Using satellite-derived ice motion fields, the age of an individual ice parcel can be tracked by treating each grid cell that contains ice as an independent Lagrangian particle and advecting the particles at weekly time steps [(Fowler et al., 2004)]. In cases where particles of different ages fall within a single grid cell, the age of the grid cell is assigned to the oldest particle. These fields reveal a dramatic decline during the past 30 years in the Arctic's multiyear ice cover [(Figure 4]). The fraction of total ice extent made up of multiyear ice in March decreased from about 75% in the mid-1980s to 45% in 2011, while the proportion of the oldest ice (i.e. ice 5 years or older) declined from 50% of the multiyear ice pack to 10% [(Maslanik et al., 2011)]. Reductions in old ice now extend into the central Arctic Ocean and adjacent to the Canadian Archipelago, areas where the ice cover was

**2.2 Ice thickness** 

Arctic-wide sea ice thickness.

**2.3 Ice age** 

relatively stable prior to 2007 and where long-term survival of sea ice through summer is considered to be most likely.

Fig. 4. Spatial distribution of ice of different age classes during March (week 11) in 1985 (top, left) and 2011 (top, right). Bottom figure shows the total extent of ice of different age classes for the Arctic Ocean domain. Data from C. Fowler and J. Maslanik, CCAR, University of Colorado.

Arctic Sea Ice Decline 449

leaving open water areas that foster new ice formation [(Rigor et al., 2002)]. By promoting more thin ice in spring, the positive AO sets the stage for negative anomalies in summer ice extent in the Siberian Arctic. It is widely believed that the late 1980s through the mid 1990s positive winter AO state led to significant export of old, thick ice out of the Arctic via Fram Strait, leaving behind thinner ice that is more vulnerable to summer melt [(e.g. Rigor and

Another key driver for recent record low summer ice extents has been persistence of a summer atmospheric circulation pattern defined by anomalously high SLP over the Beaufort Sea and Canada Basin and unusually low SLP over eastern Siberia, termed the Arctic Dipole Anomaly (DA) [(Wu et al., 2006; Overland et al., 2008; Wang et al., 2009)]. This circulation pattern leads to warm southerly winds in the Chukchi and East Siberian seas, favoring melt and transporting ice northwards towards the pole. In 2007 the DA was particularly well developed throughout the entire summer [(Figure 5]), resulting in very strong southerly winds that transported ice away from the coasts of Siberia and Alaska, as well as enhanced transport of ice out of the Arctic Ocean and into the North Atlantic through Fram Strait [(Wang et al., 2009)]. Another feature of the anomalously high pressure over the Beaufort Sea in 2007 was unusually clear skies that enhanced surface and basal melt [(Kay et al*.*, 2008)]. While the DA has not been as strong or persistent in the summers following 2007, it has been a common feature of the summer circulation pattern [(Stroeve et al., 2011a)].

Fig. 5. Composite mean sea level pressure from May through August for 2007 through 2010 (top) and corresponding 925 hPa air temperature anomaly (bottom). Data are from the

NOAA/ESRL Physical Sciences Division. Figure from Stroeve et al*.* (2011a).

Wallace, 2004; Lindsay and Zhang, 2005; Maslanik et al., 2007; Nghiem et al., 2007)].

Mapping the age of the sea ice not only provides a measure of how much the ice cover has transitioned from a largely perennial ice cover, where ice persists from year to year, to more of a seasonal coverage, it also provides an additional indication of sea ice thickness. Using February and March ICESat-derived thickness fields for the years 2003 and 2006, Maslanik et al. [(2007)] calculated mean ice thickness as a function of ice age. This comparison suggested a nearly linear relationship between age and thickness, with a 0.19 m yr-1 increase in thickness with age up to thickness values about 3 m, the approximate limit for thermodynamic ice growth. Thus, substituting these mean thicknesses for age at each grid cell provides an approximation of thickness, with the advantage of offering extended time coverage versus the ICESat data alone. Maslanik et al*.* [(2007)] suggested that using the ice age as a proxy for thickness shows that the Arctic has experienced significant thinning between the 1980s and 2006, corresponding to a ~40% reduction in ice volume.

## **3. Factors behind the sea ice decline**

The decline in Arctic sea ice can be best explained from a combination of natural climate variability, such as variability in air temperature, atmospheric and oceanic circulation, and from external forcing due to rising concentrations of atmospheric greenhouse gases (GHGs) [(e.g. Serreze et al., 2007; Kay et al., 2011)]. Global climate models have long predicted the rise in atmospheric air temperatures as GHG concentrations in the atmosphere increase [(e.g. IPCC, 2007 and references therein)]. In the 2007 Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4), hindcast simulations from the coupled GCMs that incorporate the observed record of GHGs show that the increase in global temperature results in a decline in the Arctic sea ice cover. However, while these model simulations were found to reasonably simulate the seasonal cycle in the sea ice cover [(e.g. Zhang and Walsh, 2006)], Stroeve et al. [(2007)] found that the observed winter (March) and summer (September) rates of decline were triple the respective model ensemble means, and for September the observed rate of decline was larger than all of the individual model runs. While the qualitative agreement between the GCM simulations and the observations provides evidence for a role of GHG forcing on the observed decline, the disagreement on the rate of decline could indicate a substantial natural variability component or an underestimation of the sea ice sensitivity in the models.

#### **3.1 Changes in atmospheric circulation**

An important contribution towards the recent decline in extent and thickness of the sea ice cover came from an atmospheric shift in the late 1980s, early 1990s, when the Arctic Oscillation (AO), also referred to as the Northern Annual Mode (NAM), was predominately in a positive phase [(Rigor and Wallace, 2004)]. The AO is defined as the leading stationary mode of variability in the northern hemisphere sea level pressure (SLP) field based on Empirical Orthogonal Function (EOF) analysis. As such, the AO can be interpreted as an exchange of atmospheric mass between the Arctic and the mid-latitudes [(Thompson and Wallace, 1998)]. When the AO is in its positive phase, SLP over the Icelandic region and extending into the Arctic is anomalously low, promoting a cyclonic (counter-clockwise) sea ice circulation anomaly. This results in decreased ice transport from the Beaufort Sea westward across the date line into the Chukchi Sea, increased ice transport out of the Arctic Ocean through Fram Strait, and increased transport of ice away from the Siberian coast,

Mapping the age of the sea ice not only provides a measure of how much the ice cover has transitioned from a largely perennial ice cover, where ice persists from year to year, to more of a seasonal coverage, it also provides an additional indication of sea ice thickness. Using February and March ICESat-derived thickness fields for the years 2003 and 2006, Maslanik et al. [(2007)] calculated mean ice thickness as a function of ice age. This comparison suggested a nearly linear relationship between age and thickness, with a 0.19 m yr-1 increase in thickness with age up to thickness values about 3 m, the approximate limit for thermodynamic ice growth. Thus, substituting these mean thicknesses for age at each grid cell provides an approximation of thickness, with the advantage of offering extended time coverage versus the ICESat data alone. Maslanik et al*.* [(2007)] suggested that using the ice age as a proxy for thickness shows that the Arctic has experienced significant thinning

The decline in Arctic sea ice can be best explained from a combination of natural climate variability, such as variability in air temperature, atmospheric and oceanic circulation, and from external forcing due to rising concentrations of atmospheric greenhouse gases (GHGs) [(e.g. Serreze et al., 2007; Kay et al., 2011)]. Global climate models have long predicted the rise in atmospheric air temperatures as GHG concentrations in the atmosphere increase [(e.g. IPCC, 2007 and references therein)]. In the 2007 Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4), hindcast simulations from the coupled GCMs that incorporate the observed record of GHGs show that the increase in global temperature results in a decline in the Arctic sea ice cover. However, while these model simulations were found to reasonably simulate the seasonal cycle in the sea ice cover [(e.g. Zhang and Walsh, 2006)], Stroeve et al. [(2007)] found that the observed winter (March) and summer (September) rates of decline were triple the respective model ensemble means, and for September the observed rate of decline was larger than all of the individual model runs. While the qualitative agreement between the GCM simulations and the observations provides evidence for a role of GHG forcing on the observed decline, the disagreement on the rate of decline could indicate a substantial natural variability component or an

An important contribution towards the recent decline in extent and thickness of the sea ice cover came from an atmospheric shift in the late 1980s, early 1990s, when the Arctic Oscillation (AO), also referred to as the Northern Annual Mode (NAM), was predominately in a positive phase [(Rigor and Wallace, 2004)]. The AO is defined as the leading stationary mode of variability in the northern hemisphere sea level pressure (SLP) field based on Empirical Orthogonal Function (EOF) analysis. As such, the AO can be interpreted as an exchange of atmospheric mass between the Arctic and the mid-latitudes [(Thompson and Wallace, 1998)]. When the AO is in its positive phase, SLP over the Icelandic region and extending into the Arctic is anomalously low, promoting a cyclonic (counter-clockwise) sea ice circulation anomaly. This results in decreased ice transport from the Beaufort Sea westward across the date line into the Chukchi Sea, increased ice transport out of the Arctic Ocean through Fram Strait, and increased transport of ice away from the Siberian coast,

between the 1980s and 2006, corresponding to a ~40% reduction in ice volume.

**3. Factors behind the sea ice decline** 

underestimation of the sea ice sensitivity in the models.

**3.1 Changes in atmospheric circulation** 

leaving open water areas that foster new ice formation [(Rigor et al., 2002)]. By promoting more thin ice in spring, the positive AO sets the stage for negative anomalies in summer ice extent in the Siberian Arctic. It is widely believed that the late 1980s through the mid 1990s positive winter AO state led to significant export of old, thick ice out of the Arctic via Fram Strait, leaving behind thinner ice that is more vulnerable to summer melt [(e.g. Rigor and Wallace, 2004; Lindsay and Zhang, 2005; Maslanik et al., 2007; Nghiem et al., 2007)].

Another key driver for recent record low summer ice extents has been persistence of a summer atmospheric circulation pattern defined by anomalously high SLP over the Beaufort Sea and Canada Basin and unusually low SLP over eastern Siberia, termed the Arctic Dipole Anomaly (DA) [(Wu et al., 2006; Overland et al., 2008; Wang et al., 2009)]. This circulation pattern leads to warm southerly winds in the Chukchi and East Siberian seas, favoring melt and transporting ice northwards towards the pole. In 2007 the DA was particularly well developed throughout the entire summer [(Figure 5]), resulting in very strong southerly winds that transported ice away from the coasts of Siberia and Alaska, as well as enhanced transport of ice out of the Arctic Ocean and into the North Atlantic through Fram Strait [(Wang et al., 2009)]. Another feature of the anomalously high pressure over the Beaufort Sea in 2007 was unusually clear skies that enhanced surface and basal melt [(Kay et al*.*, 2008)]. While the DA has not been as strong or persistent in the summers following 2007, it has been a common feature of the summer circulation pattern [(Stroeve et al., 2011a)].

Fig. 5. Composite mean sea level pressure from May through August for 2007 through 2010 (top) and corresponding 925 hPa air temperature anomaly (bottom). Data are from the NOAA/ESRL Physical Sciences Division. Figure from Stroeve et al*.* (2011a).

Arctic Sea Ice Decline 451

Warmer autumn temperatures have correspondingly led to later autumn freeze-up and a lengthening of the melt season by 20 days over the last three decades [(Markus et al., 2009]). Temperature anomalies are especially strong in October, a month after the seasonal sea ice minimum. This has been linked to more open water areas in September, leading to strong transfer of heat from the ocean mixed layer (the top 50 meters of the ocean) to the atmosphere as the ocean refreezes [(e.g. Serreze et al., 2009; Screen and Simmonds, 2010a, 2010b; Serreze et al., 2011]). Because it takes time for the ocean to lose its mixed layer sensible heat, the ice that grows will not be as thick as it used to be in spring. This represents a feedback in that the thinner ice will more readily melt out the next summer. Other factors towards warmer autumn temperatures include increased horizontal atmospheric heat flux convergence [(e.g. Yang et al*.,* 2010]), increased ocean heat flux convergence [(Chylek et al., 2009]) and changes in cloud cover and atmospheric water vapor that augment the

downwelling longwave radiation flux to the surface [(Francis and Hunter, 2008]).

et al., 2011b; Maslanik et al., 2011, Kwok and Cunnigham, 2010]).

As mentioned in the introduction, snow-covered sea ice has a high albedo, reflecting much of the incoming sun's radiation back out to space. As the snow begins to melt, the bare ice is exposed which absorbs more of the sun's energy. During advanced melt, melt ponds and areas of dark open water (low albedo) form, which readily absorb solar radiation, fostering further ice melt. Without this "ice-albedo" feedback, the amplitude of the seasonal cycle in ice extent (the change from March through September) would be smaller than is observed. The continued decline in the summer ice cover has in part been attributed to a growing importance of the ice-albedo feedback [(Perovich et al., 2007; Lindsay et al. 2005]). With thinner first-year ice replacing thick multiyear ice, dark open water areas form earlier in spring, leading to increased sensible heat content of the ocean mixed layer, fostering further melt [(Stroeve et al., 2011a]). The shift towards more first-year ice also has large implications for the amount of solar energy absorbed by the sea ice itself since multiyear ice tends to have a higher albedo than first-year ice [(Perovich et al., 2000]). This in turn affects the strength of the ice-albedo feedback and hence the sea ice and upper ocean energy balance. The growing albedo feedback appears to be especially strong in the Beaufort and Chukchi seas, such that it is now warm enough that multiyear ice drifting into the region tends to melt out [(Stroeve

It has been thought that this feedback could lead to a "tipping point" where the Arctic reaches a point of no return making summer ice-free conditions inevitable and to some extent irreversible [(e.g. Holland et al., 2006; Lindsay and Zhang, 2005]). However, more recent model results indicate that such a tipping point is unlikely [(Amstrup et al., 2010; Tietsche et al., 2011]). After the additional heat in the ocean is dissipated in the autumn, the sea ice can grow rapidly under the cold atmosphere, and because ice growth is fastest when ice is thin and then slows as the ice thickens through the winter, the final spring thickness of seasonal ice is only slightly thinner than normal. This mitigates some of the effects of the icealbedo feedback. Thus, if temperatures stabilize or decrease, the summer sea ice will quickly recover to an equilibrium state determined by the temperature. Another implication of this is that the sea ice decline may be interrupted by several-year periods of little decline or even short-term increases in extent due to natural variability in the climate system [(Kay et al.,

**3.3 Enhanced ice-albedo feedback** 

2011]).

#### **3.2 A warming climate**

Another important factor behind the decline in the sea ice cover is that Arctic air temperatures are rising [(e.g. ACIA*,* 2005 and references therein)]. Figure 6 shows temperature anomalies by year and month at the 925 hPa level for the Arctic Ocean domain from JRA-25 atmospheric reanalysis, a product of the Japan Meteorological Agency [(Onogi et al., 2007]). Anomalies are computed with respect to means for the period 1979-2010. In the earlier part of the record, it was common for an anomalously warm summer, contributing to a negative September ice extent anomaly, to be followed by an anomalously cold winter or summer, helping to bring about recover in the ice cover. However, since about 2000, positive air temperature anomalies dominate all months, making it more difficult for the ice cover to rebound from an anomalously low sea ice year. The warmer air temperatures during the past decade have been linked to increased concentrations of black carbon aerosols [(e.g. Shindell and Faluvegi, 2009]), increased spring cloud cover that leads to increased downward longwave radiation at the surface [(Francis and Hunter, 2006]), changes in atmospheric circulation patterns that transports warm air into the Arctic [(*Yang et al*., 2010]), and changes in oceanic heat transport [(Polykov et al., 2005; Shimada et al., 2006]). Other factors include the loss of the ice cover itself as well as increases in atmospheric GHGs (discussed shortly).

Fig. 6. Near surface air temperature anomalies (925 hPa) for all regions north of 60oN by year (x-axis) and by month (y-axis). Temperature anomalies are derived relative to the 1979 to 2010 mean.

Warmer spring temperatures have led to earlier melt onset in recent years [(e.g. Markus et al., 2009; Belchansky et al., 2004]). The earlier the ice begins to melt, the earlier the albedo drops, leading to increased absorption of solar radiation that further melts the ice and helps to cause open water areas to develop earlier during summer. These expanding open water areas absorb even more of the sun's energy, leading to enhanced basal and lateral melt. Deposition of black carbon and other impurities on the ice cover further increases the absorption of solar radiation, leading to enhanced ice melt.

Another important factor behind the decline in the sea ice cover is that Arctic air temperatures are rising [(e.g. ACIA*,* 2005 and references therein)]. Figure 6 shows temperature anomalies by year and month at the 925 hPa level for the Arctic Ocean domain from JRA-25 atmospheric reanalysis, a product of the Japan Meteorological Agency [(Onogi et al., 2007]). Anomalies are computed with respect to means for the period 1979-2010. In the earlier part of the record, it was common for an anomalously warm summer, contributing to a negative September ice extent anomaly, to be followed by an anomalously cold winter or summer, helping to bring about recover in the ice cover. However, since about 2000, positive air temperature anomalies dominate all months, making it more difficult for the ice cover to rebound from an anomalously low sea ice year. The warmer air temperatures during the past decade have been linked to increased concentrations of black carbon aerosols [(e.g. Shindell and Faluvegi, 2009]), increased spring cloud cover that leads to increased downward longwave radiation at the surface [(Francis and Hunter, 2006]), changes in atmospheric circulation patterns that transports warm air into the Arctic [(*Yang et al*., 2010]), and changes in oceanic heat transport [(Polykov et al., 2005; Shimada et al., 2006]). Other factors include the loss of the ice cover itself as well as increases in atmospheric GHGs

Fig. 6. Near surface air temperature anomalies (925 hPa) for all regions north of 60oN by year (x-axis) and by month (y-axis). Temperature anomalies are derived relative to the 1979

absorption of solar radiation, leading to enhanced ice melt.

Warmer spring temperatures have led to earlier melt onset in recent years [(e.g. Markus et al., 2009; Belchansky et al., 2004]). The earlier the ice begins to melt, the earlier the albedo drops, leading to increased absorption of solar radiation that further melts the ice and helps to cause open water areas to develop earlier during summer. These expanding open water areas absorb even more of the sun's energy, leading to enhanced basal and lateral melt. Deposition of black carbon and other impurities on the ice cover further increases the

**3.2 A warming climate** 

(discussed shortly).

to 2010 mean.

Warmer autumn temperatures have correspondingly led to later autumn freeze-up and a lengthening of the melt season by 20 days over the last three decades [(Markus et al., 2009]). Temperature anomalies are especially strong in October, a month after the seasonal sea ice minimum. This has been linked to more open water areas in September, leading to strong transfer of heat from the ocean mixed layer (the top 50 meters of the ocean) to the atmosphere as the ocean refreezes [(e.g. Serreze et al., 2009; Screen and Simmonds, 2010a, 2010b; Serreze et al., 2011]). Because it takes time for the ocean to lose its mixed layer sensible heat, the ice that grows will not be as thick as it used to be in spring. This represents a feedback in that the thinner ice will more readily melt out the next summer. Other factors towards warmer autumn temperatures include increased horizontal atmospheric heat flux convergence [(e.g. Yang et al*.,* 2010]), increased ocean heat flux convergence [(Chylek et al., 2009]) and changes in cloud cover and atmospheric water vapor that augment the downwelling longwave radiation flux to the surface [(Francis and Hunter, 2008]).

### **3.3 Enhanced ice-albedo feedback**

As mentioned in the introduction, snow-covered sea ice has a high albedo, reflecting much of the incoming sun's radiation back out to space. As the snow begins to melt, the bare ice is exposed which absorbs more of the sun's energy. During advanced melt, melt ponds and areas of dark open water (low albedo) form, which readily absorb solar radiation, fostering further ice melt. Without this "ice-albedo" feedback, the amplitude of the seasonal cycle in ice extent (the change from March through September) would be smaller than is observed.

The continued decline in the summer ice cover has in part been attributed to a growing importance of the ice-albedo feedback [(Perovich et al., 2007; Lindsay et al. 2005]). With thinner first-year ice replacing thick multiyear ice, dark open water areas form earlier in spring, leading to increased sensible heat content of the ocean mixed layer, fostering further melt [(Stroeve et al., 2011a]). The shift towards more first-year ice also has large implications for the amount of solar energy absorbed by the sea ice itself since multiyear ice tends to have a higher albedo than first-year ice [(Perovich et al., 2000]). This in turn affects the strength of the ice-albedo feedback and hence the sea ice and upper ocean energy balance. The growing albedo feedback appears to be especially strong in the Beaufort and Chukchi seas, such that it is now warm enough that multiyear ice drifting into the region tends to melt out [(Stroeve et al., 2011b; Maslanik et al., 2011, Kwok and Cunnigham, 2010]).

It has been thought that this feedback could lead to a "tipping point" where the Arctic reaches a point of no return making summer ice-free conditions inevitable and to some extent irreversible [(e.g. Holland et al., 2006; Lindsay and Zhang, 2005]). However, more recent model results indicate that such a tipping point is unlikely [(Amstrup et al., 2010; Tietsche et al., 2011]). After the additional heat in the ocean is dissipated in the autumn, the sea ice can grow rapidly under the cold atmosphere, and because ice growth is fastest when ice is thin and then slows as the ice thickens through the winter, the final spring thickness of seasonal ice is only slightly thinner than normal. This mitigates some of the effects of the icealbedo feedback. Thus, if temperatures stabilize or decrease, the summer sea ice will quickly recover to an equilibrium state determined by the temperature. Another implication of this is that the sea ice decline may be interrupted by several-year periods of little decline or even short-term increases in extent due to natural variability in the climate system [(Kay et al., 2011]).

Arctic Sea Ice Decline 453

al., 2011a, 2011]), though to some degree mitigated by the more rapid autumn ice growth, as discussed above. During the record ice loss year of 2007, the early development of large open water areas in the Beaufort and Chukchi seas led to sea surface temperature anomalies in excess of 2.5oC [(**Figure 7**]), leading to more than 2 m of basal melt by the end of summer [(Perovich et al., 2007]). Third, the Arctic has warmed in all seasons [(Serreze et al., 2009; Stroeve et al., 2011a]), leading not only to earlier melt onset and later autumn freeze-up [(Markus et al., 2009]), but also a reduced likelihood of unusually cold conditions that could

However, even during decadal or multi-decadal periods of accelerated ice loss, climate model simulations such as those from CCSM3 and CCSM4 simulations [(Holland et al., 2007, Kay et al., 2011]), show there can be several-year periods of slow ice loss or even temporary increases in extent. Thus, while the observed trend is steeper over the past decade compared to the earlier part of the record, it is uncertain if this pattern will be sustained. The higher extent for September 2009 relative to 2007 and 2008 may have suggested a temporary recovery, yet September 2010 and 2011 saw less ice compared to 2009 despite a winter circulation pattern that should have helped to favor ice retention through the summer melt season in 2010 [(Stroeve et al., 2011b]), and generally less favorable

Fig. 7. Mean August sea surface temperatures and anomalies for 2007 through 2011.

Anomalies are computed relative to the 1982-2006 mean. Credit: Michael Steele and Wendy Ermold, Polar Science Center, Applied Physics Laboratory, University of Washington.

bring about temporary recovery through natural climate variability.

summer conditions for substantial ice loss in summer of 2011.

### **3.4 Warmer ocean temperatures**

There is growing evidence that the waters of the Arctic are warming. Since the 1990s, there has been an overall increase in the temperature and transport of warm Atlantic water through Fram Strait [(Schauer et al., 2004; Polyakov et al., 2005; Dmitrenko et al., 2008]). However it is unclear how much of this heat can be brought to the surface to melt the ice cover. This is because the warm Atlantic water exists below the colder, fresher (less dense) Arctic surface water. A more recent study found that the Atlantic water flowing north into the Arctic Ocean is the warmest that it has been in at least 2,000 years [(Spielhagen et al., 2011]). This large heat input into the Arctic appears to have caused an enhanced heat flux to the surface, concurrent with the decreasing sea ice cover.

On the Pacific side of the Arctic there is also indication of warm water inflow that is capable of reaching the surface and melting the ice from below. Shimada et al. [(2006]) find increases in Pacific Surface Water temperature in the Arctic Ocean beginning in the late 1990s that appear to have helped melt ice in the Chukchi and Beaufort seas. Jackson et al. [(2010]) finds that near surface (20-25 m depth) ocean temperatures in the Canada Basin have increased and moved further northward between 1993 and 2007.

#### **3.5 Atmospheric greenhouse gases**

All models participating in the IPCC 2007 AR4 report show declining sea ice extent over the period of observations when forced with the observed record of atmospheric GHGs [(e.g. Stroeve et al., 2007; Bo et al., 2009; Zhang and Walsh, 2006; Overpeck, 2005]). This is not the case when the models are run using pre-industrial levels of atmospheric greenhouse gasses and provides strong evidence that GHGs are in part responsible for the observed decline. Stroeve et al. [(2007]) estimated that 47-57% of the decline from 1979-2006 could be explained by GHG-forcing assuming that the multi-model ensemble mean was an accurate representation of the forced change by GHGs. Min et al. [(2008]) examined the observed changes from 1953-2006 together with the AR4 models and concluded that since the 1990s there has been a detectable anthropogenic forcing on the sea ice trends. A more recent paper by Kay et al. [(2011]) analyzed the influence of anthropogenic forcing on the late 20th and early 21st century Arctic sea ice extent trends using the latest version of the Community Climate System Model version 4 (CCSM4). They concluded that the observed ice losses could not be explained by natural climate variability alone.

Another feature of the GCM simulations is that some of the models show that the decline becomes steeper with time, but not until much later into the 21st century [(e.g. Wang and Overland, 2009]). However, it appears that accelerated ice loss is already happening today. Stroeve et al. [(2011a]) argue that the recent steepening in the rate of decline of September ice extent points to a growing non-linear response to external climate forcing. This growing non-linear response is related to the following. First, because there is less sea ice in September, there is more open water in which new ice can form in winter. Since this ice is thinner than ice that has survived more than one melt season, the following spring ice cover is increasingly dominated by thinner ice that is more vulnerable to melting out in summer, especially when atmospheric forcing is conducive to ice loss [(e.g. Lindsay et al., 2009]). Second, the presence of more thin ice in spring, allows open water areas to develop earlier in the melt season, leading to increased importance of the ice-albedo feedback [(e.g. Stroeve et

There is growing evidence that the waters of the Arctic are warming. Since the 1990s, there has been an overall increase in the temperature and transport of warm Atlantic water through Fram Strait [(Schauer et al., 2004; Polyakov et al., 2005; Dmitrenko et al., 2008]). However it is unclear how much of this heat can be brought to the surface to melt the ice cover. This is because the warm Atlantic water exists below the colder, fresher (less dense) Arctic surface water. A more recent study found that the Atlantic water flowing north into the Arctic Ocean is the warmest that it has been in at least 2,000 years [(Spielhagen et al., 2011]). This large heat input into the Arctic appears to have caused an enhanced heat flux to

On the Pacific side of the Arctic there is also indication of warm water inflow that is capable of reaching the surface and melting the ice from below. Shimada et al. [(2006]) find increases in Pacific Surface Water temperature in the Arctic Ocean beginning in the late 1990s that appear to have helped melt ice in the Chukchi and Beaufort seas. Jackson et al. [(2010]) finds that near surface (20-25 m depth) ocean temperatures in the Canada Basin have increased

All models participating in the IPCC 2007 AR4 report show declining sea ice extent over the period of observations when forced with the observed record of atmospheric GHGs [(e.g. Stroeve et al., 2007; Bo et al., 2009; Zhang and Walsh, 2006; Overpeck, 2005]). This is not the case when the models are run using pre-industrial levels of atmospheric greenhouse gasses and provides strong evidence that GHGs are in part responsible for the observed decline. Stroeve et al. [(2007]) estimated that 47-57% of the decline from 1979-2006 could be explained by GHG-forcing assuming that the multi-model ensemble mean was an accurate representation of the forced change by GHGs. Min et al. [(2008]) examined the observed changes from 1953-2006 together with the AR4 models and concluded that since the 1990s there has been a detectable anthropogenic forcing on the sea ice trends. A more recent paper by Kay et al. [(2011]) analyzed the influence of anthropogenic forcing on the late 20th and early 21st century Arctic sea ice extent trends using the latest version of the Community Climate System Model version 4 (CCSM4). They concluded that the observed ice losses

Another feature of the GCM simulations is that some of the models show that the decline becomes steeper with time, but not until much later into the 21st century [(e.g. Wang and Overland, 2009]). However, it appears that accelerated ice loss is already happening today. Stroeve et al. [(2011a]) argue that the recent steepening in the rate of decline of September ice extent points to a growing non-linear response to external climate forcing. This growing non-linear response is related to the following. First, because there is less sea ice in September, there is more open water in which new ice can form in winter. Since this ice is thinner than ice that has survived more than one melt season, the following spring ice cover is increasingly dominated by thinner ice that is more vulnerable to melting out in summer, especially when atmospheric forcing is conducive to ice loss [(e.g. Lindsay et al., 2009]). Second, the presence of more thin ice in spring, allows open water areas to develop earlier in the melt season, leading to increased importance of the ice-albedo feedback [(e.g. Stroeve et

**3.4 Warmer ocean temperatures** 

the surface, concurrent with the decreasing sea ice cover.

and moved further northward between 1993 and 2007.

could not be explained by natural climate variability alone.

**3.5 Atmospheric greenhouse gases** 

al., 2011a, 2011]), though to some degree mitigated by the more rapid autumn ice growth, as discussed above. During the record ice loss year of 2007, the early development of large open water areas in the Beaufort and Chukchi seas led to sea surface temperature anomalies in excess of 2.5oC [(**Figure 7**]), leading to more than 2 m of basal melt by the end of summer [(Perovich et al., 2007]). Third, the Arctic has warmed in all seasons [(Serreze et al., 2009; Stroeve et al., 2011a]), leading not only to earlier melt onset and later autumn freeze-up [(Markus et al., 2009]), but also a reduced likelihood of unusually cold conditions that could bring about temporary recovery through natural climate variability.

However, even during decadal or multi-decadal periods of accelerated ice loss, climate model simulations such as those from CCSM3 and CCSM4 simulations [(Holland et al., 2007, Kay et al., 2011]), show there can be several-year periods of slow ice loss or even temporary increases in extent. Thus, while the observed trend is steeper over the past decade compared to the earlier part of the record, it is uncertain if this pattern will be sustained. The higher extent for September 2009 relative to 2007 and 2008 may have suggested a temporary recovery, yet September 2010 and 2011 saw less ice compared to 2009 despite a winter circulation pattern that should have helped to favor ice retention through the summer melt season in 2010 [(Stroeve et al., 2011b]), and generally less favorable summer conditions for substantial ice loss in summer of 2011.

Fig. 7. Mean August sea surface temperatures and anomalies for 2007 through 2011. Anomalies are computed relative to the 1982-2006 mean. Credit: Michael Steele and Wendy Ermold, Polar Science Center, Applied Physics Laboratory, University of Washington.

Arctic Sea Ice Decline 455

While the model simulated September ice extents are in qualitative agreement with the observations that the sea ice has been in declining during the observational record, it is clear that the models are failing to capture the rate of decline. From 1953 to 2010, the observed rate of decline in the September ice extent is -8.8 + 0.6 percent per decade. The multi-model ensemble mean is only -2.8 + 0.2 percent per decade. Over the modern satellite data record (e.g. 1979-2010) the model ensemble mean increases to -4.4 + 0.2 percent per decade, but it remains three times less than the observed (-12.5 + 1.6 percent per decade) trend. The inability of the climate models to accurately simulate the observed rate of decline has raised speculation that the Arctic may become seasonally ice-free sooner than the climate models

As mentioned earlier, it appears that both anthropogenic forcing and natural climate variability are responsible for the observed decline. However it is clear from climate model simulations that as atmospheric GHGs continue to increase, the Arctic will transition towards a seasonally ice-free Arctic state. However, climate model simulations also suggest the natural climate model variability can exert a strong influence on the sea ice trends, especially on sub-20 year time-scales. Thus, while the Arctic ice cover appears to be on track to become ice-free in summers in the future, it is entirely possible to have periods of

Except during summer, the sea ice cover insulates a relatively warm Arctic Ocean from a much colder atmosphere. As mentioned earlier, with less sea ice at summer's end and more open water, the ocean absorbs the suns energy that would normally be reflected by the ice cover and the ocean mixed layer heats up. As temperatures drop in autumn and the sun disappears for winter, the ocean will eventually lose the extra mixed layer heat that it gained in summer and the ice will form. This results in large heat transfers from the ocean back to the atmosphere in autumn, most pronounced at the surface but also extending upwards in the lower troposphere. This "amplified warming" is a prominent feature of climate models. **Figure 9** shows how the pattern of cold-season warming over the Arctic Ocean grows with time in the model simulations. Additionally, the warming is focused near the surface, but extends up to considerable height within the troposphere. Analysis of the GCMs participating in the IPCC 2007 report reveal consistency in the seasonality and vertical structure of this warming, but with different timings, magnitudes and spatial patterns of change [(Serreze et al., 2009]). This is in part a result of inter-model scatter in the timing of the sea ice cover retreat through the coming century as well as differences in patterns of atmospheric heat transport, vertical mixing, cloud processes and water vapor. Atmospheric circulation and precipitation patterns are one of the least robust signals in climate model

One implication of amplified Arctic temperatures is the potential to hasten permafrost degradation [(e.g. Lawrence et al., 2008]). Permafrost is frozen land that remains below 0oC for at least two years. In the Northern Hemisphere, permafrost occupies approximately 22.79 million km2 (about 24% of the exposed land surface) [(Zhang et al. 2003]). This permafrost contains a significant store of the Earth's deposits of organic carbon. Observations indicate that permafrost in the Arctic is warming [(e.g. Isaksen et al., 2007;

predict.

temporary recovery [Kay et al., 2011].

**5.1 Reduced sea ice amplifies warming** 

simulations [(e.g. Deser et al., 2011]).

**5. Climate Implications** 

#### **4. Future ice conditions**

 Climate models have long suggested that as atmospheric greenhouse gas concentrations increase, the Arctic Ocean will likely become ice-free during. In some of the model projections, this seasonally ice-free Arctic state is expected to occur sometime after 2100 whereas in other models the estimate is as early as 2050 [(IPCC, 2007]). Figure 8 shows a comparison between the observed September ice extent from 1953 to 2010, based on a combination of early ship, aircraft, satellite and submarine records together with the modern satellite data record [(Stroeve et al., 2007; Stroeve et al., 2008]), and 15 models participating in the IPCC AR4 report using the "business as usual (A1B)" GHG scenario. It is clear that all models show declining September sea ice extent, with some models indicating that the Arctic will become ice-free sometime during the latter half of this century and others suggesting it will happen sometime after 2100. However, there is significant spread in the simulated September ice extents.

Fig. 8. Comparison between Global Climate Model (GCM) simulations of September sea ice extent (individual color lines) and observations (shown in red). The solid black line represents the ensemble mean and the dotted black lines are the +/- one standard deviation of the ensemble mean. Figure updated from Stroeve et al., 2007.

 Climate models have long suggested that as atmospheric greenhouse gas concentrations increase, the Arctic Ocean will likely become ice-free during. In some of the model projections, this seasonally ice-free Arctic state is expected to occur sometime after 2100 whereas in other models the estimate is as early as 2050 [(IPCC, 2007]). Figure 8 shows a comparison between the observed September ice extent from 1953 to 2010, based on a combination of early ship, aircraft, satellite and submarine records together with the modern satellite data record [(Stroeve et al., 2007; Stroeve et al., 2008]), and 15 models participating in the IPCC AR4 report using the "business as usual (A1B)" GHG scenario. It is clear that all models show declining September sea ice extent, with some models indicating that the Arctic will become ice-free sometime during the latter half of this century and others suggesting it will happen sometime after 2100. However, there is significant spread in the

Fig. 8. Comparison between Global Climate Model (GCM) simulations of September sea ice

represents the ensemble mean and the dotted black lines are the +/- one standard deviation

extent (individual color lines) and observations (shown in red). The solid black line

of the ensemble mean. Figure updated from Stroeve et al., 2007.

**4. Future ice conditions** 

simulated September ice extents.

While the model simulated September ice extents are in qualitative agreement with the observations that the sea ice has been in declining during the observational record, it is clear that the models are failing to capture the rate of decline. From 1953 to 2010, the observed rate of decline in the September ice extent is -8.8 + 0.6 percent per decade. The multi-model ensemble mean is only -2.8 + 0.2 percent per decade. Over the modern satellite data record (e.g. 1979-2010) the model ensemble mean increases to -4.4 + 0.2 percent per decade, but it remains three times less than the observed (-12.5 + 1.6 percent per decade) trend. The inability of the climate models to accurately simulate the observed rate of decline has raised speculation that the Arctic may become seasonally ice-free sooner than the climate models predict.

As mentioned earlier, it appears that both anthropogenic forcing and natural climate variability are responsible for the observed decline. However it is clear from climate model simulations that as atmospheric GHGs continue to increase, the Arctic will transition towards a seasonally ice-free Arctic state. However, climate model simulations also suggest the natural climate model variability can exert a strong influence on the sea ice trends, especially on sub-20 year time-scales. Thus, while the Arctic ice cover appears to be on track to become ice-free in summers in the future, it is entirely possible to have periods of temporary recovery [Kay et al., 2011].

## **5. Climate Implications**

#### **5.1 Reduced sea ice amplifies warming**

Except during summer, the sea ice cover insulates a relatively warm Arctic Ocean from a much colder atmosphere. As mentioned earlier, with less sea ice at summer's end and more open water, the ocean absorbs the suns energy that would normally be reflected by the ice cover and the ocean mixed layer heats up. As temperatures drop in autumn and the sun disappears for winter, the ocean will eventually lose the extra mixed layer heat that it gained in summer and the ice will form. This results in large heat transfers from the ocean back to the atmosphere in autumn, most pronounced at the surface but also extending upwards in the lower troposphere. This "amplified warming" is a prominent feature of climate models. **Figure 9** shows how the pattern of cold-season warming over the Arctic Ocean grows with time in the model simulations. Additionally, the warming is focused near the surface, but extends up to considerable height within the troposphere. Analysis of the GCMs participating in the IPCC 2007 report reveal consistency in the seasonality and vertical structure of this warming, but with different timings, magnitudes and spatial patterns of change [(Serreze et al., 2009]). This is in part a result of inter-model scatter in the timing of the sea ice cover retreat through the coming century as well as differences in patterns of atmospheric heat transport, vertical mixing, cloud processes and water vapor. Atmospheric circulation and precipitation patterns are one of the least robust signals in climate model simulations [(e.g. Deser et al., 2011]).

One implication of amplified Arctic temperatures is the potential to hasten permafrost degradation [(e.g. Lawrence et al., 2008]). Permafrost is frozen land that remains below 0oC for at least two years. In the Northern Hemisphere, permafrost occupies approximately 22.79 million km2 (about 24% of the exposed land surface) [(Zhang et al. 2003]). This permafrost contains a significant store of the Earth's deposits of organic carbon. Observations indicate that permafrost in the Arctic is warming [(e.g. Isaksen et al., 2007;

Arctic Sea Ice Decline 457

Osterkamp, 2007; Jorgenson et al, 2001]), and this warming is already leading to increased emissions of carbon dioxide and methane [(e.g. Schuur et al., 2009]). Climate models suggest that as much as 90% of the near-surface permafrost may disappear by the end of this century [(Lawrence et al., 2008]). This has the potential to release large amounts of carbon into the

Another potential impact of Arctic amplification is the potential to melt more of the land ice in the Arctic, including the ice on the Greenland ice sheet. More melting of land ice in the

Heating the atmosphere over the Arctic Ocean alters the temperature gradient from the equator to the poles. This impacts both the atmosphere's static stability and the atmospheric thickness gradient. A weaker thickness gradient toward the poles results in weaker vertical wind speed (or wind shear). This impacts on the development, tracks and strengths of storms and the precipitation they generate, that may well extend beyond the Arctic. For example, there is some evidence that autumn sea level pressure fields following summers with less sea ice exhibit higher pressures over much the Arctic Ocean and the North Atlantic, compensated by lower pressures in the mid-latitudes, otherwise known as the negative phase of the AO [(e.g. Francis et al., 2009]). This can have wide-spread impacts on

There is broad agreement in climate models that warming will lead to increased precipitation in the northern high latitudes [(Kattsov et al., 2007; Finnis et al., 2007; Deser et al, 2010]). This is linked to the influence of more open water on atmospheric temperatures and the ability of a warmer atmosphere to hold more water vapor. However, different modeling studies have come to vastly different conclusions as to the spatial distribution of these precipiation changes. While many models suggest a decrease in cyclone frequency, but more intense storms, other studies suggest increased frequency of intense storms [(Lambert and Fyfe, 2006]), while other studies suggest precipitation increases are a result of poleward

Simmonds and Kaey (2009) suggest the loss of the summer Arctic sea ice cover has led to increased strength of September cyclones. This link is associated with increased sensible and latent heat fluxes associated with more open water. Stroeve et al. [(2011c]) further note there has been a corresponding increase in autumn precipitation associated with more frequent and stronger cyclones in years that had anomalously low September sea ice extent [(**Figure 10**]). However cause and effect remain unclear. While there has been anomalously more sensible and latent heat fluxes, as well as more precipitable water vapor over the regions where the ice cover has retreated, the increase in precipitation is primarily linked to a shift in atmospheric circulation towards more frequent and intense cyclones in the Atlantic sector of the Arctic. Thus, the absence of a clear link between the spatial patterns of recent precipitation changes and sea ice extent anomalies may suggest the link between recent

However, it is entirely possible that the recent declines in the sea ice cover have in part forced the observed circulation changes [(Francis et al., 2009; Overland and Wang, 2010]). As changes in the temperature structure in the Arctic atmosphere become more pronounced in the coming decades as the summer sea ice cover continues to decline, a clearer link between

corresponding atmospheric circulation and precipitation changes should emerge.

temperatures and precipitation in the Northern Hemisphere [(e.g. Hurrell, 1995]).

atmosphere and contribute to further warming.

**5.2 Altered weather patterns** 

shifts in storm tracks (Yin, 2005).

changes in cyclone activity and sea ice loss is premature.

Arctic has the potential to raise global sea level by several meters.

Fig. 9. Air temperature simulations from the NCAR CCSM3 global climate model using the IPCC A1B emission scenario. The top figure shows the near surface (2 meter) air temperature anomalies by month and year over the Arctic Ocean while the bottom figure shows the latitude by height plot of the October through March temperature anomalies for 2050-2059 relative to 1979-2007. Figure from Serreze et al., 2009.

Fig. 9. Air temperature simulations from the NCAR CCSM3 global climate model using the

temperature anomalies by month and year over the Arctic Ocean while the bottom figure shows the latitude by height plot of the October through March temperature anomalies for

IPCC A1B emission scenario. The top figure shows the near surface (2 meter) air

2050-2059 relative to 1979-2007. Figure from Serreze et al., 2009.

Osterkamp, 2007; Jorgenson et al, 2001]), and this warming is already leading to increased emissions of carbon dioxide and methane [(e.g. Schuur et al., 2009]). Climate models suggest that as much as 90% of the near-surface permafrost may disappear by the end of this century [(Lawrence et al., 2008]). This has the potential to release large amounts of carbon into the atmosphere and contribute to further warming.

Another potential impact of Arctic amplification is the potential to melt more of the land ice in the Arctic, including the ice on the Greenland ice sheet. More melting of land ice in the Arctic has the potential to raise global sea level by several meters.

## **5.2 Altered weather patterns**

Heating the atmosphere over the Arctic Ocean alters the temperature gradient from the equator to the poles. This impacts both the atmosphere's static stability and the atmospheric thickness gradient. A weaker thickness gradient toward the poles results in weaker vertical wind speed (or wind shear). This impacts on the development, tracks and strengths of storms and the precipitation they generate, that may well extend beyond the Arctic. For example, there is some evidence that autumn sea level pressure fields following summers with less sea ice exhibit higher pressures over much the Arctic Ocean and the North Atlantic, compensated by lower pressures in the mid-latitudes, otherwise known as the negative phase of the AO [(e.g. Francis et al., 2009]). This can have wide-spread impacts on temperatures and precipitation in the Northern Hemisphere [(e.g. Hurrell, 1995]).

There is broad agreement in climate models that warming will lead to increased precipitation in the northern high latitudes [(Kattsov et al., 2007; Finnis et al., 2007; Deser et al, 2010]). This is linked to the influence of more open water on atmospheric temperatures and the ability of a warmer atmosphere to hold more water vapor. However, different modeling studies have come to vastly different conclusions as to the spatial distribution of these precipiation changes. While many models suggest a decrease in cyclone frequency, but more intense storms, other studies suggest increased frequency of intense storms [(Lambert and Fyfe, 2006]), while other studies suggest precipitation increases are a result of poleward shifts in storm tracks (Yin, 2005).

Simmonds and Kaey (2009) suggest the loss of the summer Arctic sea ice cover has led to increased strength of September cyclones. This link is associated with increased sensible and latent heat fluxes associated with more open water. Stroeve et al. [(2011c]) further note there has been a corresponding increase in autumn precipitation associated with more frequent and stronger cyclones in years that had anomalously low September sea ice extent [(**Figure 10**]). However cause and effect remain unclear. While there has been anomalously more sensible and latent heat fluxes, as well as more precipitable water vapor over the regions where the ice cover has retreated, the increase in precipitation is primarily linked to a shift in atmospheric circulation towards more frequent and intense cyclones in the Atlantic sector of the Arctic. Thus, the absence of a clear link between the spatial patterns of recent precipitation changes and sea ice extent anomalies may suggest the link between recent changes in cyclone activity and sea ice loss is premature.

However, it is entirely possible that the recent declines in the sea ice cover have in part forced the observed circulation changes [(Francis et al., 2009; Overland and Wang, 2010]). As changes in the temperature structure in the Arctic atmosphere become more pronounced in the coming decades as the summer sea ice cover continues to decline, a clearer link between corresponding atmospheric circulation and precipitation changes should emerge.

Arctic Sea Ice Decline 459

The Arctic sea ice is currently in a state of accelerated decline. This decline has been linked to a combination of natural climate variability and anthropogenic forcing linked to rising concentrations of atmospheric greenhouse gases. Global climate models suggest that as the amount of atmospheric CO2 continues to increase, the Arctic will eventually transition towards ice-free conditions in summer, though there could be periods of temporary recovery on that trajectory. Given the fact that the currently observed rate of decline exceeds that simulated by our most advanced climate models, it is entirely possible that a seasonally ice-free Arctic state may be achieved within the next 20 to 30 years. The loss of the Arctic summer sea ice cover will have profound implications for our climate. Today we are already seeing the impacts of amplified warming over the Arctic linked to the loss of the summer ice cover, with the potential to further amplify warming by releasing carbon stored in the permafrost and further melting of the Arctic's ice caps and glaciers. This warming will additionally invoke changes in atmospheric circulation. While there is at the moment no universal consensus regarding the changes in weather patterns expected, a common thread in the climate model simulations is that the changes will be significant and affect areas well

Amstrup, S.C., E.T. DeWeaver, D.C. Douglas, B.G. Marcot, G.M. Durner, C.M. Bitz, and D.A.

Arctic Climate Impact Assessment (ACIA) (2004), *Impacts of a Warming Arctic*. Cambridge

Belchansky, G.I., D.C. Douglas, and N.G. Platonov, (2004), Duration of the Arctic sea ice melt season: Regional and interannual variability, 1979-2001, *J. Clim., 17*, 67-80. Bo, J.; Hall, A.; Qu, X, (2009), September sea-ice cover in the Arctic Ocean projected to vanish

Chylek, P., C.K. Folland, G. Lesins, M.K. Dubey, and M-Y. Wang (2009), Arctic air

Comiso, J. C., C. L. Parkinson, R. Gersten, and L. Stock, (2008), Accelerated decline in the Arctic sea ice cover, *Geophys. Res. Lett*., 35, L01703,doi:10.1029/2007GL031972. Deser, C., A. S. Phillips, V. Bourdette, and H. Teng, (2011), Uncertainty in climate change

Deser, C., Tomas, R., Alexander, M. and Lawrence, D. (2010), The seasonal atmospheric

Dimitrenko, I.A., I.V., Polyakov, S.A. Krillov, L.A. Timokhov, I.E. Frolov, V.T. Sokolov, H.L.

Basin margins, *J. Geophys. Res.*, 113, C05023, doi:10.1029/2007JC004158. Finnis, J., Holland, M.M., Serreze, M.C. and Cassano, J.J. (2007), Response of Northern

temperature change amplification and the Atlantic Multidecadal Oscillation,

projections: The role of internal variability. Climate Dyn., DOI 10.1007/s00382-010-

response to projected Arctic sea ice loss in the late twenty-first century. *J. Climate*

Simmons, V.V. Ivnov, and D. Walsh (2008), Toward a warmer Arctic Ocean: Spreading the early 21st century Atlantic Water warm anomaly along the Eurasian

Hemisphere extratropical cyclone activity and associated precipitation to climate

Bailey (2010), Greenhouse gas mitigation can reduce sea-ice loss and increase polar

**6. Summary** 

beyond the boundaries of the Arctic.

bear persistence, *Nature*, 468, 955-958.

University Press, Cambridge, UK, pp139.

by 2100, *Nature Geoscience* 2 (5), 341, doi:10.1038/ngeo467

*Geophys. Res. Lett*. 36, L14801, doi:10.1029/2009GL038777.

**7. References** 

0977-x.

23, 333-351.

Fig. 10. Cyclone associated precipitation anomalies averaged for all regions north of 60oN (top) and corresponding sea ice extent anomalies (bottom). Anomalies are computed with respect to the 1979 to 2009 mean. Figure from Stroeve et al., 2011c.

#### **6. Summary**

458 Greenhouse Gases – Emission, Measurement and Management

Fig. 10. Cyclone associated precipitation anomalies averaged for all regions north of 60oN (top) and corresponding sea ice extent anomalies (bottom). Anomalies are computed with

respect to the 1979 to 2009 mean. Figure from Stroeve et al., 2011c.

The Arctic sea ice is currently in a state of accelerated decline. This decline has been linked to a combination of natural climate variability and anthropogenic forcing linked to rising concentrations of atmospheric greenhouse gases. Global climate models suggest that as the amount of atmospheric CO2 continues to increase, the Arctic will eventually transition towards ice-free conditions in summer, though there could be periods of temporary recovery on that trajectory. Given the fact that the currently observed rate of decline exceeds that simulated by our most advanced climate models, it is entirely possible that a seasonally ice-free Arctic state may be achieved within the next 20 to 30 years. The loss of the Arctic summer sea ice cover will have profound implications for our climate. Today we are already seeing the impacts of amplified warming over the Arctic linked to the loss of the summer ice cover, with the potential to further amplify warming by releasing carbon stored in the permafrost and further melting of the Arctic's ice caps and glaciers. This warming will additionally invoke changes in atmospheric circulation. While there is at the moment no universal consensus regarding the changes in weather patterns expected, a common thread in the climate model simulations is that the changes will be significant and affect areas well beyond the boundaries of the Arctic.

## **7. References**


Arctic Sea Ice Decline 461

Maslanik, J.A., C. Fowler, J. Stroeve and W. Emery (2011), Distribution and trends in Arctic

Maslanik, J.A., C. Fowler, J. Stroeve, S. Drobot, J. Zwally, D. Yi and W. Emery, (2007), A

Min, S.-K., X. Zhang, F. W. Zwiers, and T. Agnew (2008), Human influence on Arctic sea ice

Nghiem, S.V., I.G. Rigor, D.K. Perovich, P. Clemente-Colón, J. Richter-Menge, J. W.

Onogi, K., J. Tsutsui, H. Koide, M. Sakamoto, S. Kobayashi, H. Hatsushika, T. Matsumoto, S.

Overpeck, Jonathan T. (2005), Arctic System on Trajectory to New, Seasonally Ice-Free State,

Osterkamp, T.E. (2007), Characteristics of the recent warming of permafrost in Alaska. *J.* 

Perovich, D. K., J. A. Richter-Menge, K. F. Jones, and B. Light, (2008), Sunlight, water, and

Polyakov, I.V., et al. (2005), One more step toward a warmer Arctic, *Geophys. Res. Lett*., 32,

Rigor, I.G. and J.M. Wallace (2004), Variations in the age of Arctic sea ice and summer sea ice extent, Geophys. Res. Lett., 31, 109401, doi:10.1029/2004GL019492. Rigor, I.G., J.M. Wallace and R.L. Colony (2002), Response of the sea ice to the Arctic

Schauer, U., E. Fahrbach, S. Osterhus, and G. Rohardt (2004), Arctic warming through the

Schuur, E.A.G., J.G. Vogel, K.G. Crummer, H. Lee, J.O. Sickman and T.E. Osterkamp, (2009),

Screen, J.A., and I. Simmonds (2010a), The central role of diminishing sea ice in recent Arctic

Screen, J.A., and I. Simmonds (2010b), Increasing fall-winter energy loss from the Arctic

Serreze, M.C., A.P. Barrett and J.J. Cassano (2011), Circulation and surface controls on the

Serreze, M.C., A.P. Barrett, J.C. Stroeve, D.M. Kindig and M.M. Holland (2009), The emergence of surface-based Arctic amplification, *The Cryosphere*, 3, 1-9. Serreze, M.C., M.M. Holland and J. Stroeve, 2007. Perspectives on the Arctic's shrinking sea

Fram Strait – Oceanic heat transport from three years of measurements, *J. Geophys.* 

The effect of permafrost thaw on old carbon release and net carbon exchange from

Ocean and its role in Arctic temperature amplification, *Geophys. Res. Lett*. 37,

lower-tropospheric air temperature field of the Arctic, *J. Geophys. Res*., 116, D07104,

loss*. Geophys. Res. Lett*., 34, L24501, doi:10.1029/2007GL032043.

*Geophys. Res. Lett*., 24, L19504, doi:10.1029/2007GL031128.

*Eos, Transactions, American Geophysical Union,* 86 (34), 309–316.

*Geophys. Res*., 112, doi:10.1029/2006JF000578.

L11501, doi:10.1029/2008GL034007.

L17605, doi:10.1029/2005GL023740.

Oscillation, *J. Clim*., 15, 2648-2663.

*Res*., 109, C06026, doi:10.1029/203JC001823.

tundra, *Nature*, 459, doi:10.1038/nature0831.

L16707, doi: 10.1029/2010GL044136.

doi:10.1029/2010JD015127..

ice cover, *Science*, 315, 1533-1536.

temperature amplification, Nature, 464, 1334-1337.

doi:10.1029/2011GL047735.

doi:10.1029/2008GL035725.

369-432.

sea ice age through spring 2011, *Geophys. Res. Lett.*, 38, L13502,

younger, thinner Arctic ice cover: Increased potential for rapid, extensive sea-ice

detectable from early 1990s onwards, *Geophys. Res. Lett*., 35, L21701,

Weatherly, and G. Neumann, Rapid reduction of Arctic perennial sea ice, (2007),

Yamazaki, H. Kamahori, K. Takahashi, S. Kadokura, K. Wada, K. Kato, R. Oyama, T. Ose, N. Mannoji, and R. Taira (2007), The JRA-25 reanalysis, *J. Met. Soc. Japan*, 85,

ice: Extreme Arctic sea ice melt during the summer of 2007, *Geophys. Res. Lett.*, 35,

change, as represented by the Community Climate System Model. *J. Geophys. Res*. 112, G04S42, doi: 10.1029/2006JG000286.


Francis, J.A., W. Chan, D.J. Leathers, J.R. Miller and D.E. Veron, (2009), Winter Northern

Francis, J.A., and E. Hunter (2006), New insight into the disappearing Arctic sea ice, *EOS,* 

Gillett, Nathan P.; Stone, Dáithí A.; Stott, Peter A.; Nozawa, Toru; Karpechko, Alexey Yu.;

Intergovernmental Panel on Climate Change (IPCC) (2007), Climate Change 2007: The

Isaksen, K., R.E. Benestad, C. Harris, and J.L. Sollid, (2007), Recent extreme near-surface

Jackson, J.M., E.C. Carmack, F.A. Mc Laughlin, S.E. Allen, and R.G. Ingram (2010),

Jorgenson, M.T., C.H. Racine, J.C. Walters, and T.E. Osterkamp, (2001), Permafrost

Kattsov, V.M., Walsh, J.E., Chapman, W.L., Govorkova, V.A., Pavlova, T.V. and Zhang, X.

Kay, J.E., M. M. Holland, and A. Jahn (2011), *Inter-annual to multi-decadal Arctic sea ice extent* 

Kay, J.E., T. L'Ecuyer, A. Gettelman, G. Stephens, and C. O'Dell (2008), The contribution of

Kwok, R., and G. F. Cunningham, (2010), Contribution of melt in the Beaufort Sea to the

Lambert, S.J. and Fyfe, J.C. (2006), Changes in winter cyclone frequencies and strengths

Markus, T., J. C. Stroeve, and J. Miller, (2009), Recent changes in Arctic sea ice melt onset, freeze-up, and melt season length, *J. Geophys. Res.*, doi:10.1029/2009JC005436.

participating in the IPCC diagnostic exercise. *Clim. Dynam*., 26, 713-728. Lindsay, R.W. and J. Zhang (2005), The thinning of Arctic sea ice, 1988-2003: have we passed

112, G04S42, doi: 10.1029/2006JG000286.

*Trans. Am. Geophys. Union*, 87, 509-511.

and precipitation, *Science*, 269, 676-679.

doi:10.1029/2009JC005265.

doi:10.1175/JHM575.1.

Alaska, *Climatic Change*, 48, 551-579.

L20501, doi:10.1029/2010GL044678.

a tipping point*? J. Climate*, 18, 4879-4894.

*Res. Lett*., 35, L08503, doi:10.1029/2008GL033451.

al., Cambridge Univ. Press, Cambridge, U. K.

Geophysical Research Letters doi:10.1029/2007GL031002.

*Res. Lett*., 36, L07503, doi:10.1029/2009GL037274.

change, as represented by the Community Climate System Model. *J. Geophys. Res*.

Hemisphere weather patterns remember summer Arctic sea ice extent, *Geophys.* 

Hegerl, Gabriele C.; Wehner, Michael F.; Jones, Philip D. (2008), Attribution of polar warming to human influence", *Nature Geoscience*, 1, 750, doi:10.1038/ngeo33. Holland, M.M., C.M. Bitz, and B. Tremblay, (2006), Future abrupt reductions in the summer Arctic sea ice. *Geophys. Res. Lett., 33*, L23503, doi:10.1029/2006GL028024. Hurrel, J.W., (1995), Decadal trends in the North Atlantic Oscillation: Regional tempeatures

Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, edited by S. Solomon et

permafrost temperatures on Svalbard in relation to future climate scenarios.

Identification, characterization, and change of the near-surface temperature maximum in the Canada Basin, 1993-2008, *J. Geophys. Res*., 115, C05021,

degradation and ecological changes associated with a warming climate in central

2007. Simulation and projection of Arctic freshwater budget components by the IPCC AR4 Global Climate Models. *J. Hydrometeorol*. 8, 571-589,

*trends in a warming world*, *Geophys. Res. Lett.*, 38, L15708, doi:10.1029/2011GL048008.

cloud and radiation anomalies to the 2007 Arctic sea ice extent minimum, *Geophys.* 

decline in Arctic multiyear sea ice coverage: 1993–2009, *Geophys. Res. Lett.*, 37,

simulated in enhanced greenhouse warming experiments: Results from models


**21** 

*1Germany 2,3Spain* 

**Post-Combustion CO2 Capture with** 

*1Technische Universität Berlin, Institute for Energy Engineering,* 

**Environmental Assessment<sup>1</sup>**

Fontina Petrakopoulou1,2, George Tsatsaronis1,

Alicia Boyano1,3 and Tatiana Morosuk1

*2IMDEA Energy, Energy Systems Analysis Unit, 3Join Research Center-European commission, Institute for Prospective Technological Studies,* 

**Monoethanolamine in a Combined-Cycle** 

Post-combustion capture through chemical absorption is the most technologically-mature CO2 capture method, developed about 70 years ago to remove acid gases from natural gas streams (Herzog, 2001). The prominent disadvantage of the method is the significant amount of thermal energy required for the regeneration of the chemicals used, resulting in a considerable efficiency penalty. In recent years, many studies have reviewed different postcombustion technologies and compared the effectiveness of different types of absorbents for chemical absorption (e.g., Kothandaraman et al., 2009; Rubin & Rao, 2002). Analyses have yet to reveal any significant breakthrough, leaving chemical absorption as one of the most energy intensive methodologies for CO2 capture. Nevertheless, the significant advantage of chemical absorption is that existing plants can be retrofitted with a capture unit without further rearrangements. This straightforward application of the technology makes it

This study evaluates the performance of a combined-cycle power plant with postcombustion CO2 capture using monoethanolamine (MEA plant). The structure and operating conditions of the plant are based on a reference power plant that does not include CO2 capture. The methods used in the evaluation process are exergy-based. In an exergetic analysis the physical and chemical exergies of process streams are calculated and the performance of the plant components is assessed using exergy destruction and exergetic efficiency. The combination of an exergetic analysis with an economic analysis and a life cycle assessment (LCA) constitutes an exergoeconomic and an exergoenvironmental

1 The views expressed are purely those of the authors and may not in any circumstances be regarded as

interesting from both a practical and an economic point of view.

stating an official position of the European Commission

**1. Introduction** 

**Power Plant: Exergetic, Economic and** 


## **Post-Combustion CO2 Capture with Monoethanolamine in a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment<sup>1</sup>**

Fontina Petrakopoulou1,2, George Tsatsaronis1, Alicia Boyano1,3 and Tatiana Morosuk1 *1Technische Universität Berlin, Institute for Energy Engineering, 2IMDEA Energy, Energy Systems Analysis Unit, 3Join Research Center-European commission, Institute for Prospective Technological Studies, 1Germany 2,3Spain* 

## **1. Introduction**

462 Greenhouse Gases – Emission, Measurement and Management

Shimada, K., T. Kamoshida, M. Itoh, S. Nishino, E. Carmack, F. McLaughlin, S. Zimmerman,

Simmonds, I. and Keay, K. (2009), Extraordinary September Arctic sea ice reductions and

Speilhagen, R.F., and others (2011), Enhanced modern heat transfer to the Arctic by warm

Stroeve, J.C., M.C. Serreze, J.E. Kay, M.M. Holland, W.N. Meier and A.P. Barrett, (2011a).

Stroeve, J.C., J. Maslanik, M.C. Serreze, I. Rigor and W. Meier, (2011b), Sea ice response to an

Stroeve, J.C., M.C. Serreze and D.N. Kindig, (2011c), Attribution of recent changes in

Stroeve, J., M. Serreze, S. Drobot, S. Gearheard, M. Holland, J. Maslanik, W. Meier, T.

Stroeve, J.C., M.C. Serreze, F. Fetterer, T. Arbetter, W. Meier, J. Maslanik and K. Knowles,

Thompson, D.W.J., and J.M. Wallace, (1998), The Arctic Oscillation signature in the

Tietsche, S., D. Notz, J. H. Jungclaus, and J. Marotzke (2011), Recovery mechanisms of Arctic summer sea ice, *Geophys. Res. Lett.*, *38*, L02707, doi:10.1029/2010GL045698. Wang, M., and J. E. Overland (2009), A sea ice free summer Arctic within 30 years?, *Geophys.* 

Wang, J., J. Zhang, E. Watanabe, M. Ikeda, K. Mizobata, J.E. Walsh, X. Bai, and B. Wu, (2009).

Wu, B., J. Wang, and J. E. Walsh (2006), Dipole anomaly in the winter Arctic atmosphere and

Yang, X-Y., J.C. Fyfe, and G.M. Flato (2010), The role of poleward energy transport in Arctic temperature evolution, *Geophys. Res. Lett*., 37, L14803, doi:10.1029/2010GL042487. Yin, J.H. (2005), A consistent poleward shift of the storm tracks in simulations of 21st century

Zhang, T., R.G. Barry, K. Knowles, F. Ling, and R.L. Armstrong, (2003), Distribution of

Permafrost. Phillips, Springman, and Arenson, eds. Swets and Zeitlinger. Zhang, X. D., and J. E. Walsh (2006), Toward a seasonally ice-covered Arctic Ocean: Scenarios from the IPCC AR4 model simulations, *J. Clim*., 19(9), 1730–1747.

seasonally and perennially frozen ground in the Northern Hemisphere. In

climate. *Geophys. Res. Lett*. 32, L18701, doi:10.1029/2005GL023684.

*Res. Lett*., 36, L07502, doi:10.1029/2009GL037820.

*Geophys. Res. Lett*., 36, L05706, doi:10.1029/2008GL036706.

its association with sea ice motion, *J. Climate*, 19, 210–225.

Atlantic water, *Science*, 331 (6016), doi:10.116/science.1197397.

doi:10.1029/2005GL025624.

L19715, doi:10.1029/2009GL039810.

doi: 10.1007/s10584-011-0101-1.

*Res. Lett.*, doi: 2010GL045662.

0870.2011.00515.x.

89(2), 13-14.

2005.

1297-1300.

and A. Proshutinsky, (2006), Pacific Ocean inflow: Influence on catastrophic reduction of sea ice cover in the Arctic Ocean, *Geophys. Res. Lett.,* 33, L08605,

their relationships with storm behavior over 1979-2008*. Geophys. Res. Lett*. 36,

The Arctic's rapidly shrinking sea ice cover: A research synthesis, *Climatic Change*,

extreme negative phase of the Arctic Oscillation during winter 2009/2010, *Geophys.* 

autumn cyclone associated precipitation in the Arctic, *Tellus*, DOI: 10.1111/j.1600-

Scambos, (2008), Arctic sea ice extent plummets in 2007, *EOS Transactions, AGU*,

(2005), Tracking the Arctic's shrinking ice cover: Another extreme September minimum in 2004, *Geophys. Res. Lett.,* 32(4), L04501, doi:10.1029/2004GL021810,

wintertime geopotential height and temperature fields, *Geophs. Res. Lett.*, 25(9),

Is the dipole anomaly a major drier to record lows in Arctic summer sea ice extent?,

Post-combustion capture through chemical absorption is the most technologically-mature CO2 capture method, developed about 70 years ago to remove acid gases from natural gas streams (Herzog, 2001). The prominent disadvantage of the method is the significant amount of thermal energy required for the regeneration of the chemicals used, resulting in a considerable efficiency penalty. In recent years, many studies have reviewed different postcombustion technologies and compared the effectiveness of different types of absorbents for chemical absorption (e.g., Kothandaraman et al., 2009; Rubin & Rao, 2002). Analyses have yet to reveal any significant breakthrough, leaving chemical absorption as one of the most energy intensive methodologies for CO2 capture. Nevertheless, the significant advantage of chemical absorption is that existing plants can be retrofitted with a capture unit without further rearrangements. This straightforward application of the technology makes it interesting from both a practical and an economic point of view.

This study evaluates the performance of a combined-cycle power plant with postcombustion CO2 capture using monoethanolamine (MEA plant). The structure and operating conditions of the plant are based on a reference power plant that does not include CO2 capture. The methods used in the evaluation process are exergy-based. In an exergetic analysis the physical and chemical exergies of process streams are calculated and the performance of the plant components is assessed using exergy destruction and exergetic efficiency. The combination of an exergetic analysis with an economic analysis and a life cycle assessment (LCA) constitutes an exergoeconomic and an exergoenvironmental

<sup>1</sup> The views expressed are purely those of the authors and may not in any circumstances be regarded as stating an official position of the European Commission

Post-Combustion CO2 Capture with Monoethanolamine in

is captured. The solution used consists of 40% MEA (w/w).

**STs**

Fig. 2. Simplified diagram of the MEA plant (grey box highlights differences from the

In the CAU, the CO2-rich gas enters the absorber flowing upwards, counter-current to the lean MEA solution (Figure 3). After the CO2 is absorbed, the clean gas is exhausted to the atmosphere and the CO2–rich solution is heated in a heat exchanger (HX) and sent to a regenerator. In the regenerator, low-pressure steam extracted from the ST of the plant provides the necessary thermal energy ( *Q* ) to regenerate the absorption medium. In this study, all of the components included in the CAU have been simulated as a black box with the embedded Equations (1)-(4) derived from Rubin & Rao (2002). Two input streams (the steam that provides the regeneration heat and the exhaust gas of the power plant) and three output streams (the exiting liquid water, the CO2 stream and the stream containing the

CO2 2 L /G exp 1.4352 0.1239 y 3.4863 0.0174 lean CO

 

0.0027 Tfg ,in (1)

–0.0397 C

**Natural gas**

**De-aerator**

remaining elements of the flue gas) have been considered.

**Combustion chamber**

**C1 GT1**

reference plant).

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 465

of the chemical solvent used, and (3) the addition of steam turbines (STs) to drive the flue gas and compressors (C2 & C3-C6 in Figure A.2). The last two points result in a significant decrease in the power output and, consequently, in the efficiency of the overall system. A simple diagram of the plant with chemical absorption capture is shown in Figure 2. The grey box highlights the additional parts of this plant, when this is compared to the reference plant. The detailed flow diagram of the plant is shown in Figure A.2. The flue gas entering the CAU of the plant consists of 3.9% (v/v) CO2, resulting in 38 kg/s of CO2, 85% of which

**Chemical Absorption Unit** 

**CO2 -depleted gas** 

**flue gas**

**HRSG 3 pressure levels**

**1 reheat stage**

**CO2**

CO2 Compression (C3-C6)

**FG COND CO2 H2O**

**H2O**

analysis, respectively. In an exergoeconomic analysis, costs are assigned to the exergy of the streams and to the exergy destruction, revealing trade-offs between capital investment and cost of exergy destruction within each component of an energy conversion system. Analogously, in an exergoenvironmental analysis, environmental impacts are assigned to the construction of the components and to the exergy destruction and trade-offs between the environmental impact of component construction and exergy destruction are identified. In all cases, the evaluation takes place both at the component and the plant level.

## **2. The power plants**

## **2.1 The reference plant**

The reference plant is a combined-cycle power plant without CO2 capture that is used as the base case for the simulation of the MEA plant. A simplified flow diagram of the reference plant is shown in Figure 1, while its detailed diagram can be found in the Appendix (Figure A.1).

Fig. 1. Simplified diagram of the reference plant.

The plant includes a three-pressure-level heat-recovery steam generator (HRSG) and one reheat stage. A description of the operation and important operating parameters of the plant can be found in Petrakopoulou (2010) and Petrakopoulou et al. (2011a).

#### **2.2 The plant with chemical absorption using monoethanolamine (MEA plant)**

The plant with post-combustion capture bears minimal structural changes when compared to the reference plant. The modifications needed to incorporate CO2 capture here are: (1) the addition of a chemical absorption unit (CAU) at the outlet of the exhaust gases, (2) the extraction of low-pressure steam to produce adequate thermal energy for the regeneration

analysis, respectively. In an exergoeconomic analysis, costs are assigned to the exergy of the streams and to the exergy destruction, revealing trade-offs between capital investment and cost of exergy destruction within each component of an energy conversion system. Analogously, in an exergoenvironmental analysis, environmental impacts are assigned to the construction of the components and to the exergy destruction and trade-offs between the environmental impact of component construction and exergy destruction are identified. In

The reference plant is a combined-cycle power plant without CO2 capture that is used as the base case for the simulation of the MEA plant. A simplified flow diagram of the reference plant is shown in Figure 1, while its detailed diagram can be found in the Appendix (Figure

**STs**

The plant includes a three-pressure-level heat-recovery steam generator (HRSG) and one reheat stage. A description of the operation and important operating parameters of the plant

The plant with post-combustion capture bears minimal structural changes when compared to the reference plant. The modifications needed to incorporate CO2 capture here are: (1) the addition of a chemical absorption unit (CAU) at the outlet of the exhaust gases, (2) the extraction of low-pressure steam to produce adequate thermal energy for the regeneration

**HRSG 3 pressure levels**

**exhaust gas**

**1 reheat stage**

all cases, the evaluation takes place both at the component and the plant level.

**C1 GT1**

Fig. 1. Simplified diagram of the reference plant.

**Combustion chamber**

**Natural gas**

**De-aerator**

can be found in Petrakopoulou (2010) and Petrakopoulou et al. (2011a).

**2.2 The plant with chemical absorption using monoethanolamine (MEA plant)** 

**2. The power plants 2.1 The reference plant** 

**Air**

A.1).

of the chemical solvent used, and (3) the addition of steam turbines (STs) to drive the flue gas and compressors (C2 & C3-C6 in Figure A.2). The last two points result in a significant decrease in the power output and, consequently, in the efficiency of the overall system. A simple diagram of the plant with chemical absorption capture is shown in Figure 2. The grey box highlights the additional parts of this plant, when this is compared to the reference plant. The detailed flow diagram of the plant is shown in Figure A.2. The flue gas entering the CAU of the plant consists of 3.9% (v/v) CO2, resulting in 38 kg/s of CO2, 85% of which is captured. The solution used consists of 40% MEA (w/w).

Fig. 2. Simplified diagram of the MEA plant (grey box highlights differences from the reference plant).

In the CAU, the CO2-rich gas enters the absorber flowing upwards, counter-current to the lean MEA solution (Figure 3). After the CO2 is absorbed, the clean gas is exhausted to the atmosphere and the CO2–rich solution is heated in a heat exchanger (HX) and sent to a regenerator. In the regenerator, low-pressure steam extracted from the ST of the plant provides the necessary thermal energy ( *Q* ) to regenerate the absorption medium. In this study, all of the components included in the CAU have been simulated as a black box with the embedded Equations (1)-(4) derived from Rubin & Rao (2002). Two input streams (the steam that provides the regeneration heat and the exhaust gas of the power plant) and three output streams (the exiting liquid water, the CO2 stream and the stream containing the remaining elements of the flue gas) have been considered.

$$\begin{aligned} \text{(L/G)} &= \exp\left(-1.4352 + 0.1239 \times \text{y}\_{\text{CO}\_2} + 3.4863 \times \text{q}\_{\text{an}} + 0.0174 \times \text{q}\_{\text{CO}\_2} - 0.0397 \times \text{C} \\\\ &+ 0.0027 \times \text{T}\_{\text{\textdegree{q},in}} \end{aligned} \right) \tag{1}$$

Post-Combustion CO2 Capture with Monoethanolamine in

relative to the lean sorbent CO2 loading.

**3. Methodology** 

**3.1 Exergetic analysis** 

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 467

Fig. 4. Exergetic efficiency (%, blue line) and energy requirement (MJ/kg CO2, grey line)

An exergetic analysis reveals the magitudes, locations and causes of inefficiencies and losses in an energy conversion system and provides insights that cannot be obtained from an energetic analysis. For a considered process, an exergetic analysis begins with a system of balance equations, formulated at the component level. The rate of exergy of product of component *k*, , is the exergy of the desired output resulting from the operation of the component, while the rate of exergy of fuel exergy of the same component, *EF k*, , is the expense in exergetic resources for the generation of the desired output. The rate of exergy destruction within component *k*, *ED k*, , is calculated as the difference between its rate of exergy of fuel and product ( *E EE Dk Fk Pk* ,,, ). For the analysis at the component level, streams exiting a component are considered either as part of the product, or they are used in the definition of the component's fuel. Thereafter, exergy

The exergetic efficiency of component *k* and that of the overall system consisting of n-

 

> , 1 , , 1

*E E* 

, , , , 1 *Pk Dk*

(5)

(6)

*Fk Fk E E E E*

*n*

*F tot F tot*

, ,

*E E*

*D k L tot*

loss is only defined for the overall system (tot): *EEEE L tot F tot P tot D tot* ,,,, .

*k*

*P tot <sup>k</sup> tot*

*E*

components are defined by Equations (5) and (6), respectively:

$$\mathbf{p}\left(\mathbf{Q}/\mathbf{L}\right) = \exp\left(-2.4452 - 0.0037 \times \mathbf{y}\_{\odot\_2} - 6.2743 \times \boldsymbol{\uprho}\_{\text{ana}} + 0.0254 \times \mathbf{C}\right) \times 100 \tag{2}$$

$$\left(\mathbf{T}\_{\rm fg,out}\right) = 41.15 + 0.062 \times \mathbf{T}\_{\rm fg,in} + 1.507 \times \mathbf{y}\_{\rm CO\_2} - 18.872 \times \boldsymbol{\uprho}\_{\rm lean} + 0.270 \times \mathbf{C} \tag{3}$$

$$\left(\text{mw}\_{\text{lan}}\right) = 16.907 + 2.333 \times \rho\_{\text{lan}} + 0.204 \times \text{C} \tag{4}$$

Here, *C* is the MEA concentration in the sorbent (w/w, %), *G* is the total inlet flue gas flow rate (kmol/h), *L* is the total sorbent flow rate (kmol/h), mwlean is the average molecular weight of the lean sorbent (kg/kmol), *Q* is the total sorbent regeneration heat requirement (GJ/h), *Tfg,in* is the temperature of the flue gas entering the CO2 absorber (°C), *Tfg,out* is the temperature of the flue gas leaving the CO2 absorber (°C), CO2 y is the CO2 concentration in the inlet flue gas (v/v, %) and lean is the lean sorbent CO2 loading that represents the part of the leftover CO2 within the regenerated solvent (mol CO2/mol MEA).

MEA is not included as a chemical compound in the simulation software (EBSILONProfessional), therefore its thermodynamic properties cannot be calculated. This leaves us with two choices: either no consideration of solvent losses (lean =0) or consideration of losses that cause a minor violation of mass conservation because no MEA input stream is considered. Without lean solvent CO2 loading, the MEA is assumed to be fully regenerated. This results in a relatively large amount of regeneration energy (6 MW/kg of CO2 captured). Therefore, the lean solvent CO2 loading has been varied from 0.0-0.3 mol CO2/mol MEA. The influence of this variation on the exergetic efficiency and on the energy requirement of the plant is shown in Figure 4. As can be seen, with a mean value of the lean solvent CO2 loading (0.2 mol CO2/mol MEA), the energy requirement is reduced from 6 MW/kg of CO2, calculated without losses, to 3.7 MW/kg of CO2. In section 4, the MEA plant is evaluated with both 0.0 and 0.2 mol CO2/mol MEA (MEA-0 and MEA-0.2), to further assess the effect of this parameter.

Fig. 3. Simplified diagram of a CAU.

Fig. 4. Exergetic efficiency (%, blue line) and energy requirement (MJ/kg CO2, grey line) relative to the lean sorbent CO2 loading.

#### **3. Methodology**

466 Greenhouse Gases – Emission, Measurement and Management

lean is the lean sorbent CO2 loading that represents the part

**R E G E N E R A T O R**

**Low-pressure steam** 

(2)

lean (3)

lean =0) or

lean (4)

CO2 Q /L exp 2.4452 0.0037 y lean 6.2743 0.0254 C 100

<sup>2</sup> T 41.15 0.062 T 1.307 y 18.872 0.270 C fg,out fg,in CO

mw 16.907 2.333 0.204 C lean

Here, *C* is the MEA concentration in the sorbent (w/w, %), *G* is the total inlet flue gas flow rate (kmol/h), *L* is the total sorbent flow rate (kmol/h), mwlean is the average molecular weight of the lean sorbent (kg/kmol), *Q* is the total sorbent regeneration heat requirement (GJ/h), *Tfg,in* is the temperature of the flue gas entering the CO2 absorber (°C), *Tfg,out* is the temperature of the flue gas leaving the CO2 absorber (°C), CO2 y is the CO2 concentration in

MEA is not included as a chemical compound in the simulation software (EBSILONProfessional), therefore its thermodynamic properties cannot be calculated. This

consideration of losses that cause a minor violation of mass conservation because no MEA input stream is considered. Without lean solvent CO2 loading, the MEA is assumed to be fully regenerated. This results in a relatively large amount of regeneration energy (6 MW/kg of CO2 captured). Therefore, the lean solvent CO2 loading has been varied from 0.0-0.3 mol CO2/mol MEA. The influence of this variation on the exergetic efficiency and on the energy requirement of the plant is shown in Figure 4. As can be seen, with a mean value of the lean solvent CO2 loading (0.2 mol CO2/mol MEA), the energy requirement is reduced from 6 MW/kg of CO2, calculated without losses, to 3.7 MW/kg of CO2. In section 4, the MEA plant is evaluated with both 0.0 and 0.2 mol CO2/mol MEA (MEA-0 and MEA-0.2), to

**HX**

**CO2 -depleted gas CO2 + H2O**

**Rich CO2 solution** 

the inlet flue gas (v/v, %) and

further assess the effect of this parameter.

**Flue gas cooler**

**Flue gas**

Fig. 3. Simplified diagram of a CAU.

of the leftover CO2 within the regenerated solvent (mol CO2/mol MEA).

**A B S O R B E R**

leaves us with two choices: either no consideration of solvent losses (

#### **3.1 Exergetic analysis**

An exergetic analysis reveals the magitudes, locations and causes of inefficiencies and losses in an energy conversion system and provides insights that cannot be obtained from an energetic analysis. For a considered process, an exergetic analysis begins with a system of balance equations, formulated at the component level. The rate of exergy of product of component *k*, , is the exergy of the desired output resulting from the operation of the component, while the rate of exergy of fuel exergy of the same component, *EF k*, , is the expense in exergetic resources for the generation of the desired output. The rate of exergy destruction within component *k*, *ED k*, , is calculated as the difference between its rate of exergy of fuel and product ( *E EE Dk Fk Pk* ,,, ). For the analysis at the component level, streams exiting a component are considered either as part of the product, or they are used in the definition of the component's fuel. Thereafter, exergy loss is only defined for the overall system (tot): *EEEE L tot F tot P tot D tot* ,,,, .

The exergetic efficiency of component *k* and that of the overall system consisting of ncomponents are defined by Equations (5) and (6), respectively:

$$
\omega\_k = \frac{\dot{E}\_{P,k}}{\dot{E}\_{F,k}} = 1 - \frac{\dot{E}\_{D,k}}{\dot{E}\_{F,k}} \tag{5}
$$

$$\dot{\varepsilon}\_{tot} = \frac{\dot{E}\_{P,tot}}{\dot{E}\_{F,tot}} = 1 - \frac{\sum\_{k=1}^{n} \dot{E}\_{D,k} + \dot{E}\_{L,tot}}{\dot{E}\_{F,tot}} \tag{6}$$

Post-Combustion CO2 Capture with Monoethanolamine in

Here, , 1

> , 1

with costs:

cost, *Zk*

related *Zk*

.

Equation (13).

*j k*

*m*

*j C* 

*i C* 

*l*

*i k*

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 469

*ik jk k*

is the sum of the cost rates associated with the *m* streams leaving the component and

In the system of balance equations, when the number of unknown stream costs is larger than the number of equations, auxiliary statements are required. For each component, streams entering are assumed to be known, while streams leaving the component are unknown. When the number of the outgoing exergy streams of a component is higher than one (m > 1), m-1 auxiliary equations are needed. The P-principle (on the product side) and the F-principle (on the fuel side) are used to determine the auxiliary equations (Lazzaretto & Tsatsaronis, 2006). The P-principle states that the cost per unit of exergy is supplied to all streams that belong to the definition of the product of the component at the same cost. The F-principle states that the cost, associated with the exergy removed from a component, has

An important outcome of the exergoeconomic analysis is the relation of exergy destruction

The calculation of the cost of exergy destruction facilitates the evaluation of plant components and allows comparisons between the cost of exergy destruction and the investment cost for the most important components. The components are first ranked and evaluated based on their total costs *C Z Dk k* , . The higher the sum of these costs is, the more significant the effect of the component on the overall plant. The contribution of the capital

*k*

*k*

*r*

*<sup>Z</sup> <sup>f</sup> Z C* 

Another important variable in the exergoeconomic evaluation is the relative cost difference, *kr* . For a given component *k*, the difference between the specific cost of product, *P k*, *c* , and the specific cost of fuel, *F k*, *<sup>c</sup>* , depends on the cost of exergy destruction, *CD k*, , and the

,, ,

, ,, *Pk Fk Dk k*

*F k Fk Pk cc CZ*

*c cE* 

, to the sum of costs is expressed by the exergoeconomic factor *kf* , defined by

, *k*

*k Dk*

is the sum of the cost rates associated with the *l* steams entering component *k,*

*C CZ*

0

(11)

*C cE Dk Fk Dk* , ,, (12)

(13)

(14)

, , 1 1

*l m*

*i j*

Zk is the rate of investment cost associated with the component.

the same specific cost as the exergy supplied to the upstream components.

where, *F k*, *c* is the average specific cost of fuel of component *k.* 

General guidelines for the definition of exergetic efficiencies have been proposed in (Lazzaretto & Tsatsaronis, 2006). In dissipative components, such as condensers, intercoolers and throttling valves, exergy is destroyed without any useful product in the component itself; thus, no exergetic purpose can be defined (Bejan et al., 1996; Lazzaretto & Tsatsaronis, 2006). The essential role of these components is to serve other plant components, leading to a more efficient or cost effective operation of the overall system.

Variables related to exergy destruction and exergy loss are the exergy destruction ratio (defined both at the component level and the overall system with, Equations (7) and (8) and the exergy loss ratio (defined only for the overall plant with, Equation 9).

$$\mathcal{Y}\_{D,k} = \frac{\dot{E}\_{D,k}}{\dot{E}\_{F,\text{tot}}} \tag{7}$$

$$y\_{D,tot} = \frac{\dot{E}\_{D,tot}}{\dot{E}\_{F,tot}} \tag{8}$$

$$y\_{L,tot} = \frac{\dot{E}\_{L,tot}}{\dot{E}\_{F,tot}} \tag{9}$$

The exergy destruction ratio is a measure of the contribution of the exergy destruction within each component to the reduction of the overall exergetic efficiency. It can be used to compare dissimilar components of the same system, while the ratios of total exergy destruction and exergy loss can be used to compare different thermodynamic systems.

With an exergetic analysis the main sources of thermodynamic irreversibilities within a plant are identified. If necessary, modifications to the plant can then be applied, in order to reduce these inefficiencies. Since the adoption and/or the development of systems are mainly driven by economics, the thermodynamically optimal design can be used as the starting point for cost reduction and eventually for cost minimization. Nowadays, the concept of *cost* could also be substituted with *environmental impact*, since a rapid increase in energy demand is foreseen that will impact the environment significantly.

#### **3.2 Exergoeconomic analysis**

An exergoeconomic analysis is an appropriate combination of an exergetic analysis with economic principles. This is achieved through exergy costing, by which a specific cost *c* is assigned to each exergy stream of the plant. The specific cost of stream *i*, *<sup>i</sup> c* , multiplied by the exergy rate of the same stream, *Ei* , provides the cost rate *Ci* , associated with stream *i*:

$$
\dot{\mathbf{C}}\_i = \mathbf{c}\_i \dot{\mathbf{E}}\_i \tag{10}
$$

To perform an exergoeconomic analysis on a plant, cost balances are formulated at the component level resulting in a system of balance equations. For example, the cost balance of component *k* is stated as follows:

$$\sum\_{i=1}^{l} \dot{\mathbf{C}}\_{i,k} - \sum\_{j=1}^{m} \dot{\mathbf{C}}\_{j,k} + \dot{Z}\_k = \mathbf{0} \tag{11}$$

Here, , 1 *l i k i C* is the sum of the cost rates associated with the *l* steams entering component *k,*

, 1 *m j k j C* is the sum of the cost rates associated with the *m* streams leaving the component and

Zk is the rate of investment cost associated with the component.

468 Greenhouse Gases – Emission, Measurement and Management

General guidelines for the definition of exergetic efficiencies have been proposed in (Lazzaretto & Tsatsaronis, 2006). In dissipative components, such as condensers, intercoolers and throttling valves, exergy is destroyed without any useful product in the component itself; thus, no exergetic purpose can be defined (Bejan et al., 1996; Lazzaretto & Tsatsaronis, 2006). The essential role of these components is to serve other plant components, leading to

Variables related to exergy destruction and exergy loss are the exergy destruction ratio (defined both at the component level and the overall system with, Equations (7) and (8) and

,

*D k*

,

,

*L tot*

*D tot*

,

(7)

(8)

(9)

*E <sup>y</sup> <sup>E</sup>*

, *D k*

*F tot E <sup>y</sup> <sup>E</sup>*

,

, *D tot*

,

*E <sup>y</sup> <sup>E</sup>*

The exergy destruction ratio is a measure of the contribution of the exergy destruction within each component to the reduction of the overall exergetic efficiency. It can be used to compare dissimilar components of the same system, while the ratios of total exergy destruction and exergy loss can be used to compare different thermodynamic systems.

With an exergetic analysis the main sources of thermodynamic irreversibilities within a plant are identified. If necessary, modifications to the plant can then be applied, in order to reduce these inefficiencies. Since the adoption and/or the development of systems are mainly driven by economics, the thermodynamically optimal design can be used as the starting point for cost reduction and eventually for cost minimization. Nowadays, the concept of *cost* could also be substituted with *environmental impact*, since a rapid increase in

An exergoeconomic analysis is an appropriate combination of an exergetic analysis with economic principles. This is achieved through exergy costing, by which a specific cost *c* is assigned to each exergy stream of the plant. The specific cost of stream *i*, *<sup>i</sup> c* , multiplied

To perform an exergoeconomic analysis on a plant, cost balances are formulated at the component level resulting in a system of balance equations. For example, the cost balance of

, provides the cost rate *Ci*

*C cE i ii* (10)

, associated with

energy demand is foreseen that will impact the environment significantly.

**3.2 Exergoeconomic analysis** 

component *k* is stated as follows:

stream *i*:

by the exergy rate of the same stream, *Ei*

, *L tot*

*F tot*

*F tot*

a more efficient or cost effective operation of the overall system.

the exergy loss ratio (defined only for the overall plant with, Equation 9).

In the system of balance equations, when the number of unknown stream costs is larger than the number of equations, auxiliary statements are required. For each component, streams entering are assumed to be known, while streams leaving the component are unknown. When the number of the outgoing exergy streams of a component is higher than one (m > 1), m-1 auxiliary equations are needed. The P-principle (on the product side) and the F-principle (on the fuel side) are used to determine the auxiliary equations (Lazzaretto & Tsatsaronis, 2006). The P-principle states that the cost per unit of exergy is supplied to all streams that belong to the definition of the product of the component at the same cost. The F-principle states that the cost, associated with the exergy removed from a component, has the same specific cost as the exergy supplied to the upstream components.

An important outcome of the exergoeconomic analysis is the relation of exergy destruction with costs:

$$
\dot{\mathcal{C}}\_{D,k} = \sigma\_{F,k} \dot{E}\_{D,k} \tag{12}
$$

where, *F k*, *c* is the average specific cost of fuel of component *k.* 

The calculation of the cost of exergy destruction facilitates the evaluation of plant components and allows comparisons between the cost of exergy destruction and the investment cost for the most important components. The components are first ranked and evaluated based on their total costs *C Z Dk k* , . The higher the sum of these costs is, the more significant the effect of the component on the overall plant. The contribution of the capital cost, *Zk* , to the sum of costs is expressed by the exergoeconomic factor *kf* , defined by Equation (13).

$$f\_k = \frac{\dot{Z}\_k}{\dot{Z}\_k + \dot{\mathcal{C}}\_{D,k}} \tag{13}$$

Another important variable in the exergoeconomic evaluation is the relative cost difference, *kr* . For a given component *k*, the difference between the specific cost of product, *P k*, *c* , and the specific cost of fuel, *F k*, *<sup>c</sup>* , depends on the cost of exergy destruction, *CD k*, , and the related *Zk* .

$$r\_k = \left(\frac{c\_{P,k} - c\_{F,k}}{c\_{F,k}}\right) = \frac{\dot{\mathcal{C}}\_{D,k} + \dot{Z}\_k}{c\_{F,k}\dot{E}\_{P,k}}\tag{14}$$

Post-Combustion CO2 Capture with Monoethanolamine in

relative environmental impact difference, *b k*, *r* .

when the value of *b k*, *f* is relatively high, *Yk*

operation of the plant (exergy destruction).

**4. Results** 

**4.1 Exergetic analysis** 

components in the Appendix.

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 471

Here, *bF,k* is the specific environmental impact of the fuel provided to component *k. BD k*, can

The exergoenvironmental analysis not only identifies the components with the highest environmental impact, but also reveals the possibilities for improvement, in order to decrease the environmental impact of the overall plant. These improvement possibilities can be identified through the sum of the component-related environmental impact and the impact of exergy destruction, *Y B k Dk* , , the exergoenvironmental factor, *b k*, *<sup>f</sup>* , and the

*k*

*k D,k <sup>Y</sup> f =Y B* 

*F,k P,k Dk k*

*b b B Y*

*b b E* 

*F,k F,k P k*

,

,

With the exergoenvironmental factor, the contribution of the component-related impact, *Yk*

to the total environmental impact, *Y B k Dk* , is expressed at the component level. In theory,

low, exergy destruction is dominant. Thus, the higher the exergoenvironmental factor, the higher the influence of the component-related impact on the overall performance of the plant. In practice, when a system works with fossil fuels, the results of the exergoenvironmental factor differ from those of the exergoeconomic factor significantly: the component-related impact is very low when compared to the impact associated with the

The environmental impact difference of component *k*, *b k*, *r* , depends on the impact of its exergy destruction and its component-related impact. Thus, it is an indicator of the environmental impact reduction potential of the component. After the calculation and evaluation of the mentioned variables, design changes are suggested, in order to reduce the

In the analysis it was assumed that all power plants are provided with the same amount of fuel. Thus, the derived rate of exergy of product ( *EP tot* , ) depends on the operating characteristics of the plant and the requirements of the CO2 capture technology. Selected results of the analysis for the overall plants are presented in Table 1 and for the individual

As previously discussed, the MEA plant is considered with both zero and 0.2 lean sorbent CO2 loading (MEA-0 and MEA-0.2). The results of MEA-0.2 agree better with published

*b,k*

then be compared to the component-related impact of component *k*, *Yk*

,

environmental impact associated with the product of the overall process.

*b k*

*r*

*B bE Dk Fk Dk* , ,, (18)

.

(19)

is dominant, whereas when the value of *b k*, *<sup>f</sup>* is

(20)

,

Information about compromises between the cost of exergy destruction and the investment cost of components, resulting from the exergoeconomic evaluation, can be used in an iterative design improvement of the plant. The objective is to reduce the cost associated with the product of the overall plant.

#### **3.3 Exergoenvironmental analysis**

In an exergoenvironmental analysis, the concepts of exergy and environmental impact are combined (Meyer et al., 2009). The component-related environmental impact of component *k*, *Yk* , is obtained in an LCA considering the entire life cycle of each plant component. It is the sum of the environmental impact of: (a) construction, *CO Yk* , (including manufacturing, transport and installation), (b) operation and maintenance, *OM Yk* and (c) the disposal, *DI Yk* , of component *k*:

$$
\dot{Y}\_k = \dot{Y}\_k^{CO} + \dot{Y}\_k^{OM} + \dot{Y}\_k^{DI} \tag{15}
$$

Similar to the exergoeconomic analysis, the exergoenvironmental analysis is performed with a system of equations written at the component level. The environmental impact balance for component *k* states that the sum of the environmental impacts associated with all input streams of the component equals the sum of the environmental impacts associated with all output streams of the same component:

$$\sum\_{i=1}^{l} \dot{B}\_{i,k} - \sum\_{j=1}^{m} \dot{B}\_{j,k} + \dot{Y}\_k + \dot{B}\_k^{PF} = 0 \tag{16}$$

Here, *B bE <sup>i</sup>*/ // *<sup>j</sup> <sup>i</sup> <sup>j</sup> <sup>i</sup> <sup>j</sup>* (*b*: specific environmental impact of stream *i/j*), , 1 *l i k i B* is the sum of the

environmental impacts associated with the *l* steams entering component *k*, , 1 *m j k j B* is the

sum of the environmental impacts associated with the *m* streams leaving component *k* and *PF Bk* is the impact of pollutant formation. The latter is related to the production of pollutants within a component and is charged to the specific component, representing the potential impact that could be caused if the generated pollutants were exhausted. Pollutant formation is defined only when a chemical reaction takes place; in any other case, it is zero. It is calculated as:

$$
\dot{B}\_k^{PF} = \sum\_i b\_i^{PF} \left( \dot{m}\_{i,out} - \dot{m}\_{i,in} \right) \tag{17}
$$

where, *min* and *mout* are the mass flow rates of pollutants entering and exiting component *k*, respectively. The pollutant streams that are taken into account here include CO2 and NOX.

When auxiliary equations need to be formulated, to make the number of the unknowns equal to the number of equations, the same principles are valid as for the exergoeconomic analysis. The environmental impact of the exergy destruction is calculated as:

$$
\dot{B}\_{\rm D,k} = b\_{\rm F,k} \dot{E}\_{\rm D,k} \tag{18}
$$

Here, *bF,k* is the specific environmental impact of the fuel provided to component *k. BD k*, can then be compared to the component-related impact of component *k*, *Yk* .

The exergoenvironmental analysis not only identifies the components with the highest environmental impact, but also reveals the possibilities for improvement, in order to decrease the environmental impact of the overall plant. These improvement possibilities can be identified through the sum of the component-related environmental impact and the impact of exergy destruction, *Y B k Dk* , , the exergoenvironmental factor, *b k*, *<sup>f</sup>* , and the relative environmental impact difference, *b k*, *r* .

$$f\_{h,k} = \frac{\dot{Y}\_k}{\dot{Y}\_k + \dot{B}\_{D,k}} \tag{19}$$

$$r\_{b,k} = \frac{\left(b\_{F,k} - b\_{P,k}\right)}{b\_{F,k}} = \frac{\mathring{B}\_{D,k} + \mathring{Y}\_k}{b\_{F,k}\mathring{E}\_{P,k}}\tag{20}$$

With the exergoenvironmental factor, the contribution of the component-related impact, *Yk* , to the total environmental impact, *Y B k Dk* , is expressed at the component level. In theory, when the value of *b k*, *f* is relatively high, *Yk* is dominant, whereas when the value of *b k*, *<sup>f</sup>* is low, exergy destruction is dominant. Thus, the higher the exergoenvironmental factor, the higher the influence of the component-related impact on the overall performance of the plant. In practice, when a system works with fossil fuels, the results of the exergoenvironmental factor differ from those of the exergoeconomic factor significantly: the component-related impact is very low when compared to the impact associated with the operation of the plant (exergy destruction).

The environmental impact difference of component *k*, *b k*, *r* , depends on the impact of its exergy destruction and its component-related impact. Thus, it is an indicator of the environmental impact reduction potential of the component. After the calculation and evaluation of the mentioned variables, design changes are suggested, in order to reduce the environmental impact associated with the product of the overall process.

#### **4. Results**

470 Greenhouse Gases – Emission, Measurement and Management

Information about compromises between the cost of exergy destruction and the investment cost of components, resulting from the exergoeconomic evaluation, can be used in an iterative design improvement of the plant. The objective is to reduce the cost associated with

In an exergoenvironmental analysis, the concepts of exergy and environmental impact are combined (Meyer et al., 2009). The component-related environmental impact of component

Similar to the exergoeconomic analysis, the exergoenvironmental analysis is performed with a system of equations written at the component level. The environmental impact balance for component *k* states that the sum of the environmental impacts associated with all input streams of the component equals the sum of the environmental impacts associated with all

> , , 1 1

environmental impacts associated with the *l* steams entering component *k*, ,

*ik jk k k*

sum of the environmental impacts associated with the *m* streams leaving component *k* and

 is the impact of pollutant formation. The latter is related to the production of pollutants within a component and is charged to the specific component, representing the potential impact that could be caused if the generated pollutants were exhausted. Pollutant formation is defined only when a chemical reaction takes place; in any other case, it is zero. It is

> , , *PF PF k i i out i in*

where, *min* and *mout* are the mass flow rates of pollutants entering and exiting component *k*, respectively. The pollutant streams that are taken into account here include CO2 and NOX. When auxiliary equations need to be formulated, to make the number of the unknowns equal to the number of equations, the same principles are valid as for the exergoeconomic

*i*

analysis. The environmental impact of the exergy destruction is calculated as:

*B B YB*

*l m*

*i j*

Here, *B bE <sup>i</sup>*/ // *<sup>j</sup> <sup>i</sup> <sup>j</sup> <sup>i</sup> <sup>j</sup>* (*b*: specific environmental impact of stream *i/j*), ,

, is obtained in an LCA considering the entire life cycle of each plant component. It is

, (including manufacturing,

,

and (c) the disposal, *DI Yk*

*CO OM DI YY Y Y kk k k* (15)

0

(16)

*B bm m* (17)

1

is the sum of the

*m*

1

*j B* 

*j k*

is the

*i B* 

*l i k*

*PF*

the product of the overall plant.

*k*, *Yk*

*PF Bk*

calculated as:

of component *k*:

**3.3 Exergoenvironmental analysis** 

output streams of the same component:

the sum of the environmental impact of: (a) construction, *CO Yk*

transport and installation), (b) operation and maintenance, *OM Yk*

#### **4.1 Exergetic analysis**

In the analysis it was assumed that all power plants are provided with the same amount of fuel. Thus, the derived rate of exergy of product ( *EP tot* , ) depends on the operating characteristics of the plant and the requirements of the CO2 capture technology. Selected results of the analysis for the overall plants are presented in Table 1 and for the individual components in the Appendix.

As previously discussed, the MEA plant is considered with both zero and 0.2 lean sorbent CO2 loading (MEA-0 and MEA-0.2). The results of MEA-0.2 agree better with published

Post-Combustion CO2 Capture with Monoethanolamine in

be found in Petrakopoulou (2010).

materials is considered.

2010).

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 473

plant. Additionally, the cost of the HRSG of both MEA plants is similar to that of the reference plant. Detailed results of the economic analysis can be found in (Petrakopoulou,

An important outcome of the exergoeconomic analysis is the correlation of exergy destruction with costs. The cost rate of exergy destruction is calculated at the component level and is compared to the respective investment cost rates. In the reference plant, the three components with the highest cost rates are those constituting the GT system: CC, GT1 and C1. The components that follow the GT system in order of importance are the LPST and the HPHRSG. In the MEA plants, the CAU presents the second highest cost of exergy destruction and total cost, right after the CC of the plant. GT1, C1 and the group of the CO2 compressors follow the CAU. In the MEA plants, the HPHRSG exceeds the STs in cost. Results for selected components of the exergoeconomic analysis are shown in Tables A.1 - A.3 of the Appendix, while the complete results at the component and the stream level can

The relative cost difference, *kr* , is found to be high for compressors and pumps, where electric power is used as fuel. This variable shows the theoretical improvement potential of the components. Nevetheless, the exergoeconomic factor, *kf* , is the main tool for evaluating the cost effectiveness of a considered component. High values of the exergoeconomic factor for components with high total cost suggest that a reduction of the investment cost should be considered. On the other hand, low values of the factor suggest that a reduction in the exergy destruction should be considered, even if this would increase the investment cost of the component. The low exergoeconomic factors of the CCs of the plants show that most of the components' total cost is related to exergy destruction. This, however, is common for chemical reactors, due to the high level of irreversibilities present there. The exergoeconomic factor of reactors increases when a design with different, more expensive and/or rare

In general, the values of the exergoeconomic factor are within the expected value ranges for the most influential components (Bejan et al., 1996). Exceptions could be the high exergoeconomic factors calculated for the CO2 compressors, suggesting that a decrease in the investment cost of these components (if less expensive components could be employed) should be considered in an attempt to improve the cost effectiveness of the overall plant. A low exergoeconomic factor is calculated for ST4 of the MEA plants. This is due to the exergy destruction within this component, which is found to be high both on its own and when it is compared to the other STs of the plants. Thus, to improve the overall operation of the plants, the efficiency of this ST should be increased. Additionally, for all plants, low factors are calculated in coolers and condensers, where relatively high exergy destruction is found.

Since the plants have the same *Fc* , the total cost rate of exergy destruction, *CD tot* , , depends on the exergy destruction ( *ED tot* , ) within the plants (i.e., *C cE D tot F D tot* , , ). The cost of exergy destruction of the MEA plants is larger when compared to that of the reference plant. Specifically, the *CD tot* , of MEA-0 is higher by 23% and that of MEA-0.2 by 16% (see Appendix). As expected, the cost differences are representative of the differences between

the *ED tot* , of the MEA plants and that of the reference plant.



Table 1. Selected results of the exergetic analysis.

The MEA plant results in an efficiency of eight percentage points lower than that of the reference plant. As expected, the main exergy destruction in the plant occurs within its combustion chamber (CC). When the reactants are preheated, the exergy destruction within the reactors decreases (Petrakopoulou, 2010). The exergy destruction within the CC of the MEA plant is the same as that of the reference plant, because the gas turbine (GT) systems of the two plants are identical. The CC is followed by the expander and the compressor of the GT system (GT1 and C1) in descending order of exergy destruction. Apart from the GT system that has a dominant influence due to its high values of *ED k*, , other components appear to be equally important. The CAU has the second highest value of exergy destruction among the plant components with 8% of the plant's *EF tot* , being destroyed there. The highpressure level of the HRSG (HPHRSG) is the most important part of the HRSG in the plant, followed by its respective low-pressure level part (LPHRSG). The low-pressure steam turbine (LPST) also presents relatively significant values of exergy destruction. Lastly, the exergy destruction within the CO2 compression unit is approximately 2% of the *EF tot* , of the plant.

#### **4.2 Exergoeconomic analysis**

The investment cost of the reference plant is calculated to be 213 € million (Petrakopoulou et al., 2011c). For the MEA plants the investment cost is increased by 50%, which is related to the cost of the CAU. Specifically, the investment cost of MEA-0.2 has been found to be 326 € million and that of MEA-0 319 € million. Comparing the two MEA plants, the regeneration requirement in MEA-0.2 is reduced and the CAU is smaller and, therefore, cheaper. However, since a smaller steam mass flow is needed for the CAU, more steam will flow through the condenser of the plant (COND, Figure A.2). Thus, the cooling water requirement of the plant increases, resulting in a larger condenser and a larger cooling tower (CT). Moreover, the first CO2 compressor (C3) is larger in MEA-0.2 than in MEA-0, because the outlet temperature of the CAU is calculated to be higher. The CO2 compression unit, i.e., the CO2 compressors and coolers, is accountable for 13% of the investment cost in the MEA

<sup>2 0</sup> and 0.2 stand for the assumed lean sorbent CO2 loading (0 or 0.2 mol CO2/mol MEA).

2010).

materials is considered.

472 Greenhouse Gases – Emission, Measurement and Management

work (Rubin & Rao, 2002). Therefore, although MEA-0 will sometimes be used for comparison purposes, MEA-0.2 is considered as the main representative plant for chemical

**Ref. Plant MEA-0.2**<sup>2</sup> **MEA-01** 

 *(%)* 56.5 48.4 45.8 *EP tot* ,  *(MW)* 412.5 353.8 334.6 *ED tot* ,  *(MW)* 300.4 349.1 368.3 *EL tot* ,  *(MW)* 17.6 27.6 27.7 *D tot* , *y (%)* 41.1 47.8 50.4

The MEA plant results in an efficiency of eight percentage points lower than that of the reference plant. As expected, the main exergy destruction in the plant occurs within its combustion chamber (CC). When the reactants are preheated, the exergy destruction within the reactors decreases (Petrakopoulou, 2010). The exergy destruction within the CC of the MEA plant is the same as that of the reference plant, because the gas turbine (GT) systems of the two plants are identical. The CC is followed by the expander and the compressor of the GT system (GT1 and C1) in descending order of exergy destruction. Apart from the GT system that has a dominant influence due to its high values of *ED k*, , other components appear to be equally important. The CAU has the second highest value of exergy destruction among the plant components with 8% of the plant's *EF tot* , being destroyed there. The highpressure level of the HRSG (HPHRSG) is the most important part of the HRSG in the plant, followed by its respective low-pressure level part (LPHRSG). The low-pressure steam turbine (LPST) also presents relatively significant values of exergy destruction. Lastly, the exergy destruction within the CO2 compression unit is approximately 2% of the *EF tot* , of the

The investment cost of the reference plant is calculated to be 213 € million (Petrakopoulou et al., 2011c). For the MEA plants the investment cost is increased by 50%, which is related to the cost of the CAU. Specifically, the investment cost of MEA-0.2 has been found to be 326 € million and that of MEA-0 319 € million. Comparing the two MEA plants, the regeneration requirement in MEA-0.2 is reduced and the CAU is smaller and, therefore, cheaper. However, since a smaller steam mass flow is needed for the CAU, more steam will flow through the condenser of the plant (COND, Figure A.2). Thus, the cooling water requirement of the plant increases, resulting in a larger condenser and a larger cooling tower (CT). Moreover, the first CO2 compressor (C3) is larger in MEA-0.2 than in MEA-0, because the outlet temperature of the CAU is calculated to be higher. The CO2 compression unit, i.e., the CO2 compressors and coolers, is accountable for 13% of the investment cost in the MEA

2 0 and 0.2 stand for the assumed lean sorbent CO2 loading (0 or 0.2 mol CO2/mol MEA).

absorption. If not otherwise stated, *MEA plant* refers to MEA-0.2.

*tot* 

Table 1. Selected results of the exergetic analysis.

plant.

**4.2 Exergoeconomic analysis** 

An important outcome of the exergoeconomic analysis is the correlation of exergy destruction with costs. The cost rate of exergy destruction is calculated at the component level and is compared to the respective investment cost rates. In the reference plant, the three components with the highest cost rates are those constituting the GT system: CC, GT1 and C1. The components that follow the GT system in order of importance are the LPST and the HPHRSG. In the MEA plants, the CAU presents the second highest cost of exergy destruction and total cost, right after the CC of the plant. GT1, C1 and the group of the CO2 compressors follow the CAU. In the MEA plants, the HPHRSG exceeds the STs in cost. Results for selected components of the exergoeconomic analysis are shown in Tables A.1 - A.3 of the Appendix, while the complete results at the component and the stream level can be found in Petrakopoulou (2010).

The relative cost difference, *kr* , is found to be high for compressors and pumps, where electric power is used as fuel. This variable shows the theoretical improvement potential of the components. Nevetheless, the exergoeconomic factor, *kf* , is the main tool for evaluating the cost effectiveness of a considered component. High values of the exergoeconomic factor for components with high total cost suggest that a reduction of the investment cost should be considered. On the other hand, low values of the factor suggest that a reduction in the exergy destruction should be considered, even if this would increase the investment cost of the component. The low exergoeconomic factors of the CCs of the plants show that most of the components' total cost is related to exergy destruction. This, however, is common for chemical reactors, due to the high level of irreversibilities present there. The exergoeconomic factor of reactors increases when a design with different, more expensive and/or rare

In general, the values of the exergoeconomic factor are within the expected value ranges for the most influential components (Bejan et al., 1996). Exceptions could be the high exergoeconomic factors calculated for the CO2 compressors, suggesting that a decrease in the investment cost of these components (if less expensive components could be employed) should be considered in an attempt to improve the cost effectiveness of the overall plant. A low exergoeconomic factor is calculated for ST4 of the MEA plants. This is due to the exergy destruction within this component, which is found to be high both on its own and when it is compared to the other STs of the plants. Thus, to improve the overall operation of the plants, the efficiency of this ST should be increased. Additionally, for all plants, low factors are calculated in coolers and condensers, where relatively high exergy destruction is found.

Since the plants have the same *Fc* , the total cost rate of exergy destruction, *CD tot* , , depends on the exergy destruction ( *ED tot* , ) within the plants (i.e., *C cE D tot F D tot* , , ). The cost of exergy destruction of the MEA plants is larger when compared to that of the reference plant. Specifically, the *CD tot* , of MEA-0 is higher by 23% and that of MEA-0.2 by 16% (see Appendix). As expected, the cost differences are representative of the differences between the *ED tot* , of the MEA plants and that of the reference plant.

Post-Combustion CO2 Capture with Monoethanolamine in

Pts/t.

(EIE).

capture.

Ref. Plant 38.41

The component-related impact (*Ytot*

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 475

For the LCA, the environmental impact of pollutant formation *PF B* of the reactors of each plant has been calculated separately. The specific environmental impact associated with each pollutant and the results of the calculations, including the impact that is avoided due to CO2 capture, are shown in Table 4. As can be seen, 60% of pollutant formation in the reference plant is related to the CO2 emissions of the reference plant, while the remaining 40% is related to its NOX emissions. The same NOX emissions are considered for the MEA plants. The environmental impact of pollutants, such as CO2, can affect the result of the overall analysis. A sensitivity analysis of the impact of CO2 emissions for the reference and MEA-0.2 plants (with and without consideration of the environmental impact of CO2 sequestration) is presented in Petrakopoulou (2010). It was found that CO2 capture becomes meaningful when the environmental impact of CO2 is higher than 20 Pts/t (when storage is also accounted for). This is a specific environmental impact of CO2 approximately four times higher than that provided by Goedkoop & Spriensma (2000). Additionally, when CO2 transport and sequestration are not accounted for, the limit for a positive environmental impact of CO2 capture decreases the required specific environmental impact of CO2 to 14

 **Ref. Plant MEA 0.2 MEA 0.0**  *Total environmental impact (103 Pts)* 2,592 3,223 2,871 *Total environmental impact (Pts/kW)* 6.3 9.1 8.6 *EIE (mPts/kWh)* 25.1 27.4 29.0

Table 3. Component-related environmental impact and environmental impact of electricity

0.05

almost negligible and differences among the total impact ( *B Y D tot tot* , ) of the plants are determined by the impact of exergy destruction (see Tables A.1 - A.3). This indicates that the construction is not the key area for reducing the environmental impact of the plants.

In the reference plant, the highest environmental impact ( *B Y Dk k* , ) corresponds to the CC, GT1, the LPST and C1. In the MEA plant, the CC is followed by the CAU, which presents a high environmental impact of exergy destruction. In the exergoenvironmental analysis, dissipative components become more important than in the exergoeconomic analysis: a high

MEA 0.2 38.42 0.05 1270 -646 MEA 0.0 38.42 0.05 1268 -646 Table 4. Environmental impact of overall and avoided pollutant formation due to CO2

**(kg/s) (mPts/kg) (kg/s) (Pts/t) (Pts/h) (Pts/h)** 

**2749.4** 

*PF <sup>B</sup>* <sup>2</sup> \_

1259 0

) differs among the plants. However, this difference is

*PF BCO capt* 

**CO2 NOX**

**5.4** 

Values for the overall plants are shown under *Total* in Tables A.1 - A.3. The overall exergoeconomic factor of the reference plant is calculated to be 40%, while the MEA plants result in overall exergoeconomic factors of 43 and 45%. Additionally, the overall relative cost difference is higher for the plants with CO2 capture than for the reference plant. This is justified with the additional charges of the supplementary equipment used.

To further compare the costs of the plants, the COE and the cost of avoided CO2 (COA-CO2) are considered. The latter shows the added cost of electricity per ton of CO2 avoided based on net plant capacity (Rubin & Rao, 2002):

 COA-CO2= 2 2 . . € / € / / / *capture ref plant emitted emitted CO CO ref plant plant capture kWh kWh t kWh t kWh* (21)

The COA-CO2 relates only to the capture of the CO2 and it does not include transportation or storage costs.

The resulting levelized COE and the COA-CO2 for the plants are shown in Table 2. Between the two MEA plants, a lower COE is achieved by MEA-0.2. The COE of this plant is 28% higher than that of the reference plant. When compared to other plants with CO2 capture (Petrakopoulou et al., 2010b; Petrakopoulou et al., 2011a, 2011b, 2011c), the MEA plants present relatively high costs. These costs are mainly associated with the high energy demand of the solvent regeneration in the CAU and the relatively low percentage of CO2 capture (85%).


COA-CO2: Cost of avoided CO2, COE: Cost of electricity

Table 2. Overall results of the exergoeconomic analysis for the plants.

#### **4.3 Results of the exergoenvironmental analysis**

The component-related environmental impacts determined in the LCAs of the plants differ in relative magnitude from costs obtained in the economic analysis. While in the economic analysis, the cost rates (calculated in €/h), are relatively substantial, in the LCA, the component-related environmental impact rates, (*Yk* , in Pts/h) are much lower in scale. Relatively high values are calculated for components constructed with materials of higher environmental impact and for the CTs of the plants, due to their large size. As can be seen in Table 3, the MEA plants have a relatively low increase in relative total environmental impact (Pts/kW), when compared to the reference plant, because of the similar equipment used in both plants. Comparing MEA-0 with MEA-0.2, the differences are also small. While the absorber of MEA-0.2 is smaller and results in a lower impact, its COND and CT are larger. This happens because of the larger mass of steam flowing through the COND, which is a direct result of the lower mass of steam extracted and used in the CAU of this plant (Petrakopoulou, 2010).

Values for the overall plants are shown under *Total* in Tables A.1 - A.3. The overall exergoeconomic factor of the reference plant is calculated to be 40%, while the MEA plants result in overall exergoeconomic factors of 43 and 45%. Additionally, the overall relative cost difference is higher for the plants with CO2 capture than for the reference plant. This is

To further compare the costs of the plants, the COE and the cost of avoided CO2 (COA-CO2) are considered. The latter shows the added cost of electricity per ton of CO2 avoided based

2 2

/ /

The COA-CO2 relates only to the capture of the CO2 and it does not include transportation

The resulting levelized COE and the COA-CO2 for the plants are shown in Table 2. Between the two MEA plants, a lower COE is achieved by MEA-0.2. The COE of this plant is 28% higher than that of the reference plant. When compared to other plants with CO2 capture (Petrakopoulou et al., 2010b; Petrakopoulou et al., 2011a, 2011b, 2011c), the MEA plants present relatively high costs. These costs are mainly associated with the high energy demand of the solvent regeneration in the CAU and the relatively low percentage of CO2

> *COE (€/MWh)* 74.1 94.6 99.5 *COA-CO2 (€/t)* N/A 73.5 92.2

The component-related environmental impacts determined in the LCAs of the plants differ in relative magnitude from costs obtained in the economic analysis. While in the economic analysis, the cost rates (calculated in €/h), are relatively substantial, in the LCA, the

Relatively high values are calculated for components constructed with materials of higher environmental impact and for the CTs of the plants, due to their large size. As can be seen in Table 3, the MEA plants have a relatively low increase in relative total environmental impact (Pts/kW), when compared to the reference plant, because of the similar equipment used in both plants. Comparing MEA-0 with MEA-0.2, the differences are also small. While the absorber of MEA-0.2 is smaller and results in a lower impact, its COND and CT are larger. This happens because of the larger mass of steam flowing through the COND, which is a direct result of the lower mass of steam extracted and used in the CAU of this plant

Table 2. Overall results of the exergoeconomic analysis for the plants.

€ / € /

*kWh kWh*

*t kWh t kWh*

.

.

**Ref. Plant MEA 0.2 MEA 0.0** 

, in Pts/h) are much lower in scale.

(21)

*capture ref plant emitted emitted CO CO ref plant plant capture*

justified with the additional charges of the supplementary equipment used.

COA-CO2=

on net plant capacity (Rubin & Rao, 2002):

COA-CO2: Cost of avoided CO2, COE: Cost of electricity

**4.3 Results of the exergoenvironmental analysis** 

component-related environmental impact rates, (*Yk*

or storage costs.

capture (85%).

(Petrakopoulou, 2010).

For the LCA, the environmental impact of pollutant formation *PF B* of the reactors of each plant has been calculated separately. The specific environmental impact associated with each pollutant and the results of the calculations, including the impact that is avoided due to CO2 capture, are shown in Table 4. As can be seen, 60% of pollutant formation in the reference plant is related to the CO2 emissions of the reference plant, while the remaining 40% is related to its NOX emissions. The same NOX emissions are considered for the MEA plants. The environmental impact of pollutants, such as CO2, can affect the result of the overall analysis. A sensitivity analysis of the impact of CO2 emissions for the reference and MEA-0.2 plants (with and without consideration of the environmental impact of CO2 sequestration) is presented in Petrakopoulou (2010). It was found that CO2 capture becomes meaningful when the environmental impact of CO2 is higher than 20 Pts/t (when storage is also accounted for). This is a specific environmental impact of CO2 approximately four times higher than that provided by Goedkoop & Spriensma (2000). Additionally, when CO2 transport and sequestration are not accounted for, the limit for a positive environmental impact of CO2 capture decreases the required specific environmental impact of CO2 to 14 Pts/t.


Table 3. Component-related environmental impact and environmental impact of electricity (EIE).


Table 4. Environmental impact of overall and avoided pollutant formation due to CO2 capture.

The component-related impact (*Ytot* ) differs among the plants. However, this difference is almost negligible and differences among the total impact ( *B Y D tot tot* , ) of the plants are determined by the impact of exergy destruction (see Tables A.1 - A.3). This indicates that the construction is not the key area for reducing the environmental impact of the plants.

In the reference plant, the highest environmental impact ( *B Y Dk k* , ) corresponds to the CC, GT1, the LPST and C1. In the MEA plant, the CC is followed by the CAU, which presents a high environmental impact of exergy destruction. In the exergoenvironmental analysis, dissipative components become more important than in the exergoeconomic analysis: a high

Post-Combustion CO2 Capture with Monoethanolamine in

percentage of CO2 capture (85%).

emissions.

considered.

**6. Nomenclature** 

*<sup>E</sup>* Exergy rate (MW)

*b* Environmental impact per unit of exergy (Pts/GJ)

*C* Cost rate associated with an exergy stream, (€/h)

*B* Rate of environmental impact (Pts/h)

*c* Cost per unit of exergy (€/GJ)

*f* Exergoeconomic factor (%)

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 477

(921 €/kW). Additionally, it was found that the cost of exergy destruction of the MEA plant is larger than that of the reference plant. It should be noted that this difference is mainly representative of the differences between the rates of exergy destruction of the plants. Larger differences between the overall costs of the reference and MEA plants are caused by the high investment cost of components used in the CO2 separation and compression units. The cost of electricity of the MEA plant is found to be 28% higher that of the reference plant. When compared to other capture technologies, and on the basis of the cost of avoided CO2, large cost differences are observed (Petrakopoulou, 2010). These differences are mainly associated with the high energy demand of the solvent regeneration and the relatively low

In the exergoenvironmental analysis, it was found that the MEA plant presents a low unit increase of the component-related environmental impact (Pts/kW), when compared to the reference plant. However, the component-related environmental impact of the plants is negligible when compared to the impact associated with the exergy destruction that takes place during the operation phase of the plants. The calculation of the overall environmental impact is mainly influenced by the impacts of fuel processing (methane) and the impact of pollutant emission. With data provided by Goepkoop & Spiensma (2000), the environmental impact of the electricity generated in the MEA plant is found to be significantly higher than that of the reference plant (2.3 mPts/kWh higher), due its high efficiency penalty. This raises questions concerning the real environmental and cost viability of chemical absorption with MEA for CO2 capture in power plants. A sensitivity analysis concerning the variation of the environmental impact of CO2 emissions showed that post-combustion technology will not decrease the environmental impact of power production, unless a specific environmental impact approximately four times higher than the present estimate is assigned to the CO2

In general, CO2 capture is a costly process, since it involves either expensive equipment that increases the overall investment cost of the facility or energy-demanding processes that decrease the efficiency, in turn increasing the fuel consumption (i.e., the fuel costs) of a plant. Moreover, the environmental analysis shows that high efficiency reductions result in significant environmental penalties. Thus, with present data, the environmental viability of post-combustion CO2 capture with chemical absorption using monoethanolamine is questionable, especially when the associated cost expenditure of the technology is also

impact is calculated for the condensers (COND and FG COND) of the plants. As already mentioned, in the exergoeconomic and exergoenvironmental analyses, the influence of the non-exergy related costs/impacts (investment cost rate and rate of the component-related impact) is different. Because in the exergoenvironmental analysis the component-related environmental impact is almost negligible, the exergy destruction and the specific environmental impact of fuel are the main deciding factors of the significance of a component. Differences between the results of the exergoenvironmental analysis and that of the exergetic analysis can be noted only for components with high environmental impacts (Petrakopoulou et al., 2010a; Petrakopoulou et al., 2011b).

The total exergoenvironmental factor is similar for the reference and MEA plants because of their similar component-related environmental impact. A reduction in the overall environmental impact could be achieved by increasing the exergetic efficiency of the GT system and of the reactors. In general, a significant decrease in the irreversibilities present in reactors is difficult to be achieved, because these irreversibilities are mostly unavoidable (Petrakopoulou et al., 2011d). However, the preheating of the reactants, as well as the use of different GT systems (e.g., steam-cooled expanders) would lead to better efficiencies, thus decreasing the incurred exergy destruction. In general, in order to reduce the overall impact of the plants, more attention should be given to the effectiveness of the component operation, thus to the exergetic efficiencies of the components.

To compare the overall environmental performance of the plants, the environmental impact of the electricity (EIE) has been calculated (Table 3). The EIE produced by the reference plant is found to be 25.1 Pts/MWh. This is comparable to the European average impact of low voltage electricity: 26 Pts/MWh (Goedkoop & Spriensma, 2000). When compared to the reference plant, the EIE of MEA-0.2 is higher by 2.3 Pts/MWh. Considering that no impact has been considered for pollutants generated by the processing of the solvent used in the plant, the case presented is considered the *best case scenario* of this plant.

## **5. Conclusion**

In this study, a post-combustion CO2 capture technology (chemical absorption with monoethanolamine) has been evaluated with two different possible energy requirements. A plant incorporating this technology (MEA plant) has been compared with a reference power plant of similar configuration that does not include CO2 capture. The plants have been analyzed using conventional exergetic, exergoeconomic and exergoenvironmental analyses.

The plant with CO2 capture presented an (exergetic) efficiency penalty of approximately eight percentage points (48.4%3), when compared to the reference plant (56.5%). The relatively high efficiency penalty of the MEA plant is caused by the high energy requirements of chemical absorption that decrease its net power output.

When comparing the cost of the plant with other CO2 capture technologies (Petrakopoulou, 2010), the MEA plant presents the most economical relative costs based on its power output

 3 Here, only MEA-0.2 is considered.

(921 €/kW). Additionally, it was found that the cost of exergy destruction of the MEA plant is larger than that of the reference plant. It should be noted that this difference is mainly representative of the differences between the rates of exergy destruction of the plants. Larger differences between the overall costs of the reference and MEA plants are caused by the high investment cost of components used in the CO2 separation and compression units. The cost of electricity of the MEA plant is found to be 28% higher that of the reference plant. When compared to other capture technologies, and on the basis of the cost of avoided CO2, large cost differences are observed (Petrakopoulou, 2010). These differences are mainly associated with the high energy demand of the solvent regeneration and the relatively low percentage of CO2 capture (85%).

In the exergoenvironmental analysis, it was found that the MEA plant presents a low unit increase of the component-related environmental impact (Pts/kW), when compared to the reference plant. However, the component-related environmental impact of the plants is negligible when compared to the impact associated with the exergy destruction that takes place during the operation phase of the plants. The calculation of the overall environmental impact is mainly influenced by the impacts of fuel processing (methane) and the impact of pollutant emission. With data provided by Goepkoop & Spiensma (2000), the environmental impact of the electricity generated in the MEA plant is found to be significantly higher than that of the reference plant (2.3 mPts/kWh higher), due its high efficiency penalty. This raises questions concerning the real environmental and cost viability of chemical absorption with MEA for CO2 capture in power plants. A sensitivity analysis concerning the variation of the environmental impact of CO2 emissions showed that post-combustion technology will not decrease the environmental impact of power production, unless a specific environmental impact approximately four times higher than the present estimate is assigned to the CO2 emissions.

In general, CO2 capture is a costly process, since it involves either expensive equipment that increases the overall investment cost of the facility or energy-demanding processes that decrease the efficiency, in turn increasing the fuel consumption (i.e., the fuel costs) of a plant. Moreover, the environmental analysis shows that high efficiency reductions result in significant environmental penalties. Thus, with present data, the environmental viability of post-combustion CO2 capture with chemical absorption using monoethanolamine is questionable, especially when the associated cost expenditure of the technology is also considered.

## **6. Nomenclature**

476 Greenhouse Gases – Emission, Measurement and Management

impact is calculated for the condensers (COND and FG COND) of the plants. As already mentioned, in the exergoeconomic and exergoenvironmental analyses, the influence of the non-exergy related costs/impacts (investment cost rate and rate of the component-related impact) is different. Because in the exergoenvironmental analysis the component-related environmental impact is almost negligible, the exergy destruction and the specific environmental impact of fuel are the main deciding factors of the significance of a component. Differences between the results of the exergoenvironmental analysis and that of the exergetic analysis can be noted only for components with high environmental impacts

The total exergoenvironmental factor is similar for the reference and MEA plants because of their similar component-related environmental impact. A reduction in the overall environmental impact could be achieved by increasing the exergetic efficiency of the GT system and of the reactors. In general, a significant decrease in the irreversibilities present in reactors is difficult to be achieved, because these irreversibilities are mostly unavoidable (Petrakopoulou et al., 2011d). However, the preheating of the reactants, as well as the use of different GT systems (e.g., steam-cooled expanders) would lead to better efficiencies, thus decreasing the incurred exergy destruction. In general, in order to reduce the overall impact of the plants, more attention should be given to the effectiveness of the component

To compare the overall environmental performance of the plants, the environmental impact of the electricity (EIE) has been calculated (Table 3). The EIE produced by the reference plant is found to be 25.1 Pts/MWh. This is comparable to the European average impact of low voltage electricity: 26 Pts/MWh (Goedkoop & Spriensma, 2000). When compared to the reference plant, the EIE of MEA-0.2 is higher by 2.3 Pts/MWh. Considering that no impact has been considered for pollutants generated by the processing of the solvent used in the

In this study, a post-combustion CO2 capture technology (chemical absorption with monoethanolamine) has been evaluated with two different possible energy requirements. A plant incorporating this technology (MEA plant) has been compared with a reference power plant of similar configuration that does not include CO2 capture. The plants have been analyzed using conventional exergetic, exergoeconomic and exergoenvironmental

The plant with CO2 capture presented an (exergetic) efficiency penalty of approximately eight percentage points (48.4%3), when compared to the reference plant (56.5%). The relatively high efficiency penalty of the MEA plant is caused by the high energy

When comparing the cost of the plant with other CO2 capture technologies (Petrakopoulou, 2010), the MEA plant presents the most economical relative costs based on its power output

(Petrakopoulou et al., 2010a; Petrakopoulou et al., 2011b).

operation, thus to the exergetic efficiencies of the components.

**5. Conclusion** 

analyses.

3 Here, only MEA-0.2 is considered.

plant, the case presented is considered the *best case scenario* of this plant.

requirements of chemical absorption that decrease its net power output.


Post-Combustion CO2 Capture with Monoethanolamine in

G

49

29

42

**GEN2**

41

M

**CT**

50

44

**HPST**

**IPST**

**LPST**

52

**COND**

**CT P**

30

51

48

46 47

M

**COND P**

43

45

G

**GT1**

**CC**

5

53

1

2

**C1**

Fig. A.1. The reference plant.

54 55

56 57

4

**GEN1**

6

**RH**

8

3

7

9

40

**HPSH**

10

**M3**

28

**M2**

40

**7. Appendix** 

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 479

M

21

19

20

11

39

**HPEC**

27

**De-aerator**

**HPEV**

M

**IPP**

24

**HPP**

Chimney

M

23 37

31

**M1**

22

35

34

**LPP**

12

**η: 58.9%, ε: 56.5%**

**Wnet: 412.5 MW**

38

13

26

25

33

32

36

14

**IPSH**

**IPEV** 15 **IPEC**

16

**LPEV**

**LPSH**

17

18

**LPEC**


#### **Subscripts**


#### **Greek symbols**

ε Exergetic efficiency (%)

#### **Abbreviations**


## **7. Appendix**

478 Greenhouse Gases – Emission, Measurement and Management

*m* Mass flow (kg/s)

D Exergy destruction F Fuel (exergy) P Product (exergy)

ε Exergetic efficiency (%)

CAU Chemical absorption unit CC Combustion chamber COA-CO2 Cost of avoided CO2 COE Cost of electricity COND Condenser CT Cooling tower EC Economizer

EIE Environmental impact of electricity

HRSG Heat recovery steam generator

HP, IP, LP High pressure, intermediate pressure, low pressure

**Subscripts** 

i,j Stream k Component

L Loss

**Greek symbols** 

**Abbreviations** 

C (1-6) Compressor

EV Evaporator FG Flue gas GT Gas turbine

HX Heat exchanger LCA Life cycle assessment MEA Monoethanolamine

PF Pollutant formation

NG Natural gas

PH Preheater RH Reheater SH Superheater ST Steam turbine

*r* Relative cost difference (%) *y* Exergy destruction ratio (%)

*Y* Component-related environmental impact (Pts/h) *Z* Cost rate associated with capital investment (€/h)

Fig. A.1. The reference plant.

Fig. A.2. The MEA plant.

1

70

Post-Combustion CO2 Capture with Monoethanolamine in

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 481

Table A.1. Selected results at the component level for the reference plant.


Table A.1. Selected results at the component level for the reference plant.

64

81

**CT**

65

82

62 63

80

79

89

90

91 99

92

93

M

88

78

87

98

96

100

86

85

**CT P**

59

58

55

52

94

83

57

**C6**

**COOL3**

56

54

**C5**

**COOL2**

**C4**

97

95 51

**COOL1**

**C3** 50 49

53

60

**COOL4**

61

G

> **LPST**

> > **IPST HPST**

73

**ST4**

**ST5**

**GEN2**

68

101

30

44

**M4**

G

**CC**

5

102

1

2

103

**C1**

Fig. A.2. The MEA plant.

4

106 105

**GT1**

104

**GEN1**

29

76

71

70

72

 **MEA-0.2**

**η: 50.5%, ε: 48.4%**

**Wnet: 353.8 MW**

 **MEA-0**

**η: 47.8%, ε: 45.8%**

**Wnet: 334.6 MW**

75 74

42

77

41

6

**RH**

8

3

7

9

40

**HPSH**

10

28

**M3**

**M2**

40

M

**HPEC**

39

21

11

27

**HPEV**

**De-aerator**

M

**COND P**

43

45

**COND**

19

69

20

67

Chimney

48

M

24

**HPP**

M

23 37

**IPP**

31

22

35

34

**M1**

**LPP**

12

38

13

26

25

33

32

36

14

**IPEV**

**IPSH**

15 **IPEC**

16

17

18

**C2**

**CAU**

> **LPEC**

> > **LPEV**

**LPSH**

47

46

66

**COOL** 

107


Table A.2. Selected results at the component level for MEA-0.2.

Post-Combustion CO2 Capture with Monoethanolamine in

Table A.3. Selected results at the component level for MEA-0.

a Combined-Cycle Power Plant: Exergetic, Economic and Environmental Assessment 483


Table A.3. Selected results at the component level for MEA-0.

482 Greenhouse Gases – Emission, Measurement and Management

Table A.2. Selected results at the component level for MEA-0.2.

**22** 

*1France 2USA* 

**The Greenhouse Stakes of Globalization** 

As 2012 approaches, twenty-five years have passed since the Brundtland report defining sustainable development as a progress that meets the needs of the present without compromising the ability of future generations to meet their own needs (United Nations). Currently, the fight against Climate change constitutes one of the main features of worldwide policies towards sustainable development. Furthermore, the global climate change issue benefits from an international discussion area through the United Nations framework convention on climate change (UNFCCC). This convention was created during the first Earth summit, which was held in Rio in 1992 and benefits from the scientific support of the Intergovernmental Panel on Climate Change (IPCC) created in November 1988. The ultimate objective of this convention is to achieve stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system (United Nations, 1992). Thus far, the 1997 Kyoto protocol constitutes the main achievement of this convention of which we propose to briefly

During the same period, the world is known to have experienced an unprecedented wave of globalization. Such globalization is characterized by the increasing number of countries taking part in international trade and the growing number of traded items both in variety, quantity, and value. **Figure 1** shows the world production and the associated globalization expressed in terms of world GDP fraction associated with exports have been respectively multiplied by 45 and 2.5 over the last fifty years. Nevertheless, over the 1960 – 2008 periods, it should be noted the same three regions1 cover over half of the world GDP. It thus seems the globalization process has so far mainly consisted in a redistribution of GDP strength

Another key point in the globalization process is the increasing sophistication of modern supply chains. Indeed, the application of modern management principles throughout the supply chain such as Just-In-Time, lean production and Efficient Consumer Response have led to more responsive, more flexible supply chains that enable firms to compete in a global

<sup>1</sup> **Western Europe**: Austria, Belgium, France, Germany, Luxembourg, Netherlands, Switzerland.

**East Asia**: China, Hong Kong, Macao, North Korea, Japan, Mongolia, South Korea. **Northern America**: Bermuda, Canada, Greenland, Saint Pierre and Miquelon, USA.

remind the objectives and analyze the policy patterns in **2.1**.

**1. Introduction** 

between these regions.

Sébastien Dente1 and Troy Hawkins2

*1Université de Technologie de Troyes 2Environmental Protection Agency* 

#### **8. References**


## **The Greenhouse Stakes of Globalization**

Sébastien Dente1 and Troy Hawkins2

*1Université de Technologie de Troyes 2Environmental Protection Agency 1France 2USA* 

## **1. Introduction**

484 Greenhouse Gases – Emission, Measurement and Management

Bejan, A., Tsatsaronis, G., & Moran, M. (1996). *Thermal Design and Optimization* (p. 560). New

http://www.steag-systemtechnologies.com/ebsilon\_professional.html Goedkoop, M., & Spriensma, R. (2000). *Eco-indicator 99 impact assessment method for Life Cycle* 

Herzog, H. J. (2001). What future for carbon capture and sequestration? *Environmental science* 

Kothandaraman, A., Nord, L., Bolland, O., Herzog, H. J., & McRae, G. J. (2009). Comparison

Lazzaretto, A., & Tsatsaronis, G. (2006). SPECO: A systematic and general methodology for calculating efficiencies and costs in thermal systems. *Energy*, *31*, 1257-1289. Meyer, L., Tsatsaronis, G., Buchgeister, J., & Schebek, L. (2009). Exergoenvironmental

Petrakopoulou, F. (2010). *Comparative Evaluation of Power Plants with CO2 Capture:* 

Petrakopoulou, F., Tsatsaronis, G., Boyano, A., & Morosuk, T. (2010b). Exergoeconomic and

Petrakopoulou, F., Tsatsaronis, G., & Morosuk, T. (2011b). Exergoeconomic Analysis of an

Petrakopoulou, F., Tsatsaronis, G., & Morosuk, T. (2011c). Conventional Exergetic and

Petrakopoulou, F., Tsatsaronis, G., Morosuk, T., & Carassai, A. (2011d). Conventional and

Rubin, E. S., & Rao, A. B. (2002). *A Technical, Economic and Environmental Assessment of* 

Retrieved from http://opus.kobv.de/tuberlin/volltexte/2011/2946/ Petrakopoulou, F., Boyano, A., Cabrera, M., & Tsatsaronis, G. (2010a). Exergy-based

*Procedia*, *1*(1), 1373-1380. Elsevier. doi:10.1016/j.egypro.2009.01.180

Design-Optimization-Adrian-Bejan/dp/0471584673

EBSILONProfessional. (n.d.) Retrieved August 8, 2011, from

*& technology*, *35*(7), 148A-153A. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11348092

*Energy*, *34*(1), 75-89. doi:10.1016/j.energy.2008.07.018

*Technologies*, *5*(4), 231-238. doi:10.1093/ijlct/ctq028

doi:10.1016/j.ijggc.2010.06.008

doi:10.1016/j.energy.2011.05.028

doi:10.2172/804932

*133*(11), 113001-12. ASME. Retrieved from http://link.aip.org/link/?GTP/133/113001/1

York: Wiley-Interscience. Retrieved from http://www.amazon.com/Thermal-

*Impact Assessment*. Amersfoort. Retrieved from http://www.pre.nl/content/eco-

of solvents for post-combustion capture of CO2 by chemical absorption. *Energy* 

analysis for evaluation of the environmental impact of energy conversion systems.

*Thermodynamic, Economic and Environmental Performance*. Institut für Energietechnik.

analyses of an advanced zero emission plant. *International Journal of Low-Carbon* 

exergoenvironmental evaluation of power plants including CO2 capture. *Chemical Engineering Research and Design*, *89*(9), 1469-1461. doi:10.1016/j.cherd.2010.08.001 Petrakopoulou, F., Boyano, A., Cabrera, M., & Tsatsaronis, G. (2011a). Exergoeconomic and

exergoenvironmental analyses of a combined cycle power plant with chemical looping technology. *International Journal of Greenhouse Gas Control*, *5*(3), 475-482.

Advanced Zero Emission Plant. *Journal of Engineering for Gas Turbines and Power*,

Exergoeconomic Analyses of a Power Plant with Chemical Looping Combustion for CO2 Capture. *International Journal of Thermodynamics*, *13*(3), 77-86. doi:10.5541/ijot.204

advanced exergetic analyses applied to a combined cycle power plant. *Energy*.

*Amine-based CO2 Capture Technology for Power Plant Greenhouse Gas Control*.

**8. References** 

indicator-99

As 2012 approaches, twenty-five years have passed since the Brundtland report defining sustainable development as a progress that meets the needs of the present without compromising the ability of future generations to meet their own needs (United Nations). Currently, the fight against Climate change constitutes one of the main features of worldwide policies towards sustainable development. Furthermore, the global climate change issue benefits from an international discussion area through the United Nations framework convention on climate change (UNFCCC). This convention was created during the first Earth summit, which was held in Rio in 1992 and benefits from the scientific support of the Intergovernmental Panel on Climate Change (IPCC) created in November 1988. The ultimate objective of this convention is to achieve stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system (United Nations, 1992). Thus far, the 1997 Kyoto protocol constitutes the main achievement of this convention of which we propose to briefly remind the objectives and analyze the policy patterns in **2.1**.

During the same period, the world is known to have experienced an unprecedented wave of globalization. Such globalization is characterized by the increasing number of countries taking part in international trade and the growing number of traded items both in variety, quantity, and value. **Figure 1** shows the world production and the associated globalization expressed in terms of world GDP fraction associated with exports have been respectively multiplied by 45 and 2.5 over the last fifty years. Nevertheless, over the 1960 – 2008 periods, it should be noted the same three regions1 cover over half of the world GDP. It thus seems the globalization process has so far mainly consisted in a redistribution of GDP strength between these regions.

Another key point in the globalization process is the increasing sophistication of modern supply chains. Indeed, the application of modern management principles throughout the supply chain such as Just-In-Time, lean production and Efficient Consumer Response have led to more responsive, more flexible supply chains that enable firms to compete in a global

<sup>1</sup> **Western Europe**: Austria, Belgium, France, Germany, Luxembourg, Netherlands, Switzerland.

**East Asia**: China, Hong Kong, Macao, North Korea, Japan, Mongolia, South Korea.

**Northern America**: Bermuda, Canada, Greenland, Saint Pierre and Miquelon, USA.

The Greenhouse Stakes of Globalization 487

countries around global climate change issue constitutes without doubt an international policy success but only 41 countries2 grouped under the "Annex I" label are mandated with GHG emission reduction targets. Following the principles of historical responsibility and the right to development of developing and emerging countries, the UNFCCC determined on December 11th 1997 in Kyoto that reduction targets should first applied to developed countries. Therefore, Annex I countries are mainly developed countries and the Article 3 of the UNFCCC Kyoto protocol sets their GHG emissions must be at least 5 % below 1990 levels for the commitment period 2008 to 2012(United Nations). On the other hand, non-Annex I countries are only encouraged to elaborate and publish methodologies and datasets of the GHG emissions associated with their activities (United Nations). Therefore, build a comprehensive worldwide dataset of GHG emissions may be complicated due to the uncertainties associated with non-Annex I GHG emissions. Nevertheless, the 1990 reference emissions associated with Annex I countries represent as much as 63.7% of 1990 worldwide GHG emissions (United Nations). Furthermore, the Kyoto Protocol entered only into force on February 16th 2005 after more than 55% of 1990 GHG emissions of Annex I countries was concerned by the Protocol. These last figures means the lowest worldwide GHG emissions reduction target set by the Kyoto Protocol corresponds to a 1.8% reduction of the 1990 world GHG emissions3. Consequently, three main questions need to be addressed in regards to the

Kyoto policy success:

emissions


countries with targets, four main trends clearly appear:




The Annex I countries GHG emissions dropped from 18.73 GtCO2 in 1990 to 17.76 GtCO2 in 2008 thus achieving a 5.2% reduction in emissions. When comparing the average GHG annual emission level of the 1990-2008 periods with the 1990 emission level, the reduction achieved is 5.3%. Therefore, from the viewpoint of the reduction targets set for Annex I countries, the Kyoto protocol has reached its objectives. However, this success covers different stories as shown in **Figure 2**. Actually, Annex I countries without emission reduction targets has known a 10% increase of their average GHG annual emission while those with emissions reduction targets followed an opposite trend with a 13% decrease of average annual emission level. Consequently, Annex I countries without GHG emission targets of which the United States constitutes the main contributor (95%) represent 42% of total Annex I GHG emissions in 2008 against 34% in 1990. Furthermore, within the Annex I

<sup>2</sup> *Non-EU Europe*: Belarus\*, Croatia (-5%), Iceland (+10%), Liechtenstein (-8%), Monaco (-8%), Norway (+1%), Switzerland (-8%), Russian Federation (+0%), Ukraine (0%) *EU27*: Bulgaria (-8%), Romania (-8%) *EU25*: Czech Republic (-8%), Estonia (-8%), Hungary (-6%), Latvia (-8%), Lithuania (-8%), Malta\*, Poland (-6%), Slovakia (-8%), Slovenia (-8%) *EU15*: Austria (-8%), Finland (-8%), Sweden (-8%) *EU12* : Denmark (-8%), Greece (-8%), Ireland (-8%), Portugal (-8%), Spain (-8%), United Kingdom (-8%) *EU6*: Belgium (-8%), France (-8%), Germany (-8%), Italy (-8%), Luxembourg (-8%), Netherlands (-8%) *Oceania:* Australia (+8%), New Zealand (+0%) *Asia*: Japan (-6%), Turkey\* *America*: Canada (-6%), United States\* (-7%)*\* Without target Annex I countries* (United Nations) 3 5% reduction\*55 % of Annex I GHG emissions\*63.7% of world GHG emission=1.8% of world GHG

market. Furthermore, the potential extension of trade partners encourages companies to experience global sourcing strategies and wider distribution of finished products (Cetinkaya, et al., 2011). All these new aspects of trade tend to increase the need for freight transportation and particularly for international freight transportation whose associated emissions were not addressed by the Kyoto Protocol. Indeed, the inclusion of these emissions constitutes the main debate of Post-Kyoto mitigation policies.

The objective of this chapter is thus to link globalization to global climate change via:


Fig. 1. The evolution of world, Northern America, East Asia and Western Europe GDP together with the corresponding fraction of world GDP in the 1960-2008 period

## **2. How to frame reduction of GHG emissions from freight transportation in a global economy?**

#### **2.1 Kyoto Protocol and GHG mitigation policy backgrounds**

In 2011, the 193 member states of the United Nations have signed the Kyoto Protocol except Afghanistan, Andorra, and South Sudan (United Nations). Nevertheless, there are 193 signatories of the Kyoto Protocol as Cook Islands and Niue are considered as additional states and European Union as an added economic region. Among these 193 signatories, only the United States has not ratified the Protocol (United Nations). Being able to unify all these

market. Furthermore, the potential extension of trade partners encourages companies to experience global sourcing strategies and wider distribution of finished products (Cetinkaya, et al., 2011). All these new aspects of trade tend to increase the need for freight transportation and particularly for international freight transportation whose associated emissions were not addressed by the Kyoto Protocol. Indeed, the inclusion of these



Fig. 1. The evolution of world, Northern America, East Asia and Western Europe GDP together with the corresponding fraction of world GDP in the 1960-2008 period

**2.1 Kyoto Protocol and GHG mitigation policy backgrounds** 

**2. How to frame reduction of GHG emissions from freight transportation in a** 

**0% 3% 6% 9% 12% 15% 18% 21% 24% 27% 30% 33% 36% 39% 42% 45% 48% 51% 54% 57% 60% 63% 66% 69%**

**Share of World GDP**

In 2011, the 193 member states of the United Nations have signed the Kyoto Protocol except Afghanistan, Andorra, and South Sudan (United Nations). Nevertheless, there are 193 signatories of the Kyoto Protocol as Cook Islands and Niue are considered as additional states and European Union as an added economic region. Among these 193 signatories, only the United States has not ratified the Protocol (United Nations). Being able to unify all these

emissions constitutes the main debate of Post-Kyoto mitigation policies.

definition of Post-Kyoto GHG reduction strategies.

transportation.

**Billions US\$**

**global economy?** 

**1960**

**1962**

**1964**

**1966**

**1968**

**1970**

**1972**

**1974**

**1976**

**1978**

**Northern America GDP**

**East Asia GDP Western Europe GDP**

**ROW GDP world Export NA, EA and WE** 

**1980**

**1982**

**1984**

**1986**

**1988**

**1990**

**1992**

**1994**

**1996**

**1998**

**2000**

**2002**

**2004**

**2006**

**2008**

The objective of this chapter is thus to link globalization to global climate change via:

countries around global climate change issue constitutes without doubt an international policy success but only 41 countries2 grouped under the "Annex I" label are mandated with GHG emission reduction targets. Following the principles of historical responsibility and the right to development of developing and emerging countries, the UNFCCC determined on December 11th 1997 in Kyoto that reduction targets should first applied to developed countries. Therefore, Annex I countries are mainly developed countries and the Article 3 of the UNFCCC Kyoto protocol sets their GHG emissions must be at least 5 % below 1990 levels for the commitment period 2008 to 2012(United Nations). On the other hand, non-Annex I countries are only encouraged to elaborate and publish methodologies and datasets of the GHG emissions associated with their activities (United Nations). Therefore, build a comprehensive worldwide dataset of GHG emissions may be complicated due to the uncertainties associated with non-Annex I GHG emissions. Nevertheless, the 1990 reference emissions associated with Annex I countries represent as much as 63.7% of 1990 worldwide GHG emissions (United Nations). Furthermore, the Kyoto Protocol entered only into force on February 16th 2005 after more than 55% of 1990 GHG emissions of Annex I countries was concerned by the Protocol. These last figures means the lowest worldwide GHG emissions reduction target set by the Kyoto Protocol corresponds to a 1.8% reduction of the 1990 world GHG emissions3. Consequently, three main questions need to be addressed in regards to the Kyoto policy success:


The Annex I countries GHG emissions dropped from 18.73 GtCO2 in 1990 to 17.76 GtCO2 in 2008 thus achieving a 5.2% reduction in emissions. When comparing the average GHG annual emission level of the 1990-2008 periods with the 1990 emission level, the reduction achieved is 5.3%. Therefore, from the viewpoint of the reduction targets set for Annex I countries, the Kyoto protocol has reached its objectives. However, this success covers different stories as shown in **Figure 2**. Actually, Annex I countries without emission reduction targets has known a 10% increase of their average GHG annual emission while those with emissions reduction targets followed an opposite trend with a 13% decrease of average annual emission level. Consequently, Annex I countries without GHG emission targets of which the United States constitutes the main contributor (95%) represent 42% of total Annex I GHG emissions in 2008 against 34% in 1990. Furthermore, within the Annex I countries with targets, four main trends clearly appear:


<sup>2</sup> *Non-EU Europe*: Belarus\*, Croatia (-5%), Iceland (+10%), Liechtenstein (-8%), Monaco (-8%), Norway (+1%), Switzerland (-8%), Russian Federation (+0%), Ukraine (0%) *EU27*: Bulgaria (-8%), Romania (-8%) *EU25*: Czech Republic (-8%), Estonia (-8%), Hungary (-6%), Latvia (-8%), Lithuania (-8%), Malta\*, Poland (-6%), Slovakia (-8%), Slovenia (-8%) *EU15*: Austria (-8%), Finland (-8%), Sweden (-8%) *EU12* : Denmark (-8%), Greece (-8%), Ireland (-8%), Portugal (-8%), Spain (-8%), United Kingdom (-8%) *EU6*: Belgium (-8%), France (-8%), Germany (-8%), Italy (-8%), Luxembourg (-8%), Netherlands (-8%) *Oceania:* Australia (+8%), New Zealand (+0%) *Asia*: Japan (-6%), Turkey\* *America*: Canada (-6%), United States\* (-7%)*\* Without target Annex I countries* (United Nations) 3 5% reduction\*55 % of Annex I GHG emissions\*63.7% of world GHG emission=1.8% of world GHG

emissions

The Greenhouse Stakes of Globalization 489

German decreasing emission trend has been going on after 1998 though at a slower pace. **Figure 3a)** allows a clearer distinction between Japan and Oceania historical evolution of emissions. While Oceania emissions are growing at a high rate especially in the 1995-2000 periods, Japan experienced their main rise in emission in the 1990-1994 periods before remaining around the same level until a huge drop in 2008 which may be due to financial crisis. Therefore, contrary to Oceania, Japan, though not fulfilling in 2008 its Kyoto target, seems to have at least stabilized its emissions. Finally, France, after some chaotic variation corresponding to the 1993-1994 recession period, experienced a slow decrease in its emissions between 1998 and 2005 followed by a higher decrease during the 2005-2008 periods. Interestingly, the beginning of this higher decrease period coincides with the introduction into the French Vth republic constitution of an environment charter (Digithèque MJP) and the entry into force of the Kyoto Protocol. Since then, environment has taken a stronger place in French policy debate but it should be noted that in spite of these late

Fig. 3. a) A historical perspective of the evolution of GHG emissions compared to 2008

**2 007**

**2 008**

**42% 27%**

**1990 2008**

**24% 20% 2.6% 3.0% 2.8% 3.3% 5.1% 4.1% 1.7% 7.2% 5.0% 12% 11%**

**20 965 MtCO2**

**1.7%**

**1.2%**

**22%**

**2.0%**

**1.5%**

**29 381 MtCO2**

**AnnexI with targets Annex I without targets Africa Middle East Non-OECD Europe/Other FSU Latin America Asia (excl.China)**

**China MARINE AVIATION**

b) Evolution of the structure worldwide energy-related emissions between 1990 and 2008

According to IEA report (IEA, 2010), the world carbon dioxide emissions associated with fuel combustion has increased by 40 percent during the period 1990 – 2008, from 21.0 GtCO2 in 1990 to 29.4 GtCO2 in 2008. During this period, the Annex I CO2 emissions associated with fuel combustion has remained the same, thus showing the difficulties to implement CO2 reduction when it comes to fuel combustion processes. Indeed, if the 2008 GHG emissions of Annex I countries are 5.2% less than those of 1990, their 2008 CO2 emissions are only 2.2% less than those of 1990. This demonstrates that most of the GHG emission reduction achieved concerned GHGs other than CO2 such as methane and nitrous oxide and that among CO2 emission reduction, those concerning fuel combustion has particularly difficult to achieve. Furthermore, while Annex I countries represented 66% of world CO2 emissions associated with fuel combustion in 1990, they represent only 47% of these in 2008.

efforts, France has not in 2008 met its Kyoto targets.

**Japan Oceania**

**France Germany**

**Russian Federation**

emission level.

**1 990**

**1 991**

**1 992**

**1 993**

**1 994**

**1 995**

**1 996**

**1 997**

**1 998**

**1 999**

**2 000**

**2 001**

**2 002**

**2 003**

**2 004**

**2 005**

**2 006**

**Index 100=Average 1990 -2008 GHG emissions**


Interestingly, while a 6.1% reduction of French GHG emissions is achieved in 2008 compared to 1990 level, this reduction drops to 0.9% when considering the average annual French GHG emissions of the 1990-2008 periods. Respectively, Japanese emissions increase from 1% to 4.7%. Therefore, though most statistics regarding climate change compares directly the 2008 emission level to those of 1990, we argue that considering the level of emissions averaged over the 1990-2008 periods better captures the historical evolution of GHG emissions. Indeed, the 2008 GHG emissions may correspond to a crisis year with low level of activities, thus setting a low GHG emission reference. Furthermore, in the perspective of an achievement of GHG emission reduction over the 2008-2012 periods and beyond, one needs to ensure the achieved GHG emission reduction are not cyclical and will not rise up as soon as the crisis period has ended.

Fig. 2. Has the reduction targets been reached for Annex I countries? A direct ratio and cumulated ratio perspective

In this framework, as shown in **Figure 3a)**, the historical perspective of Russian federation emissions is straightforward: a huge decrease due to the former Soviet Union reconstruction until 1998 and a slow increase since then. The importance of the former Soviet Union reconstruction in the rapid decrease of European emissions is further confirmed by the profile of the evolution of German emissions4. Nevertheless, contrary to Russian federation,

<sup>4</sup> Indeed, the former DRG experienced a huge decrease of emission in the beginning of the nineties, 50% decrease in terms of CO2 emission over the 1987 – 1993 period while the 1993 CO2 emissions level of the former FRG is 2% more than 1987 level (United Nations, UNFCCC, 1994)


Interestingly, while a 6.1% reduction of French GHG emissions is achieved in 2008 compared to 1990 level, this reduction drops to 0.9% when considering the average annual French GHG emissions of the 1990-2008 periods. Respectively, Japanese emissions increase from 1% to 4.7%. Therefore, though most statistics regarding climate change compares directly the 2008 emission level to those of 1990, we argue that considering the level of emissions averaged over the 1990-2008 periods better captures the historical evolution of GHG emissions. Indeed, the 2008 GHG emissions may correspond to a crisis year with low level of activities, thus setting a low GHG emission reference. Furthermore, in the perspective of an achievement of GHG emission reduction over the 2008-2012 periods and beyond, one needs to ensure the achieved GHG emission reduction are not cyclical and will

Fig. 2. Has the reduction targets been reached for Annex I countries? A direct ratio and

**-46**

**Other Non-EU**

**-39**

**-6**

**-0.9**

**France**

**With targets**

**2008/1990 GHG emission(%) Average 1990-2008/1990 GHG emission (%)**

**-22**

**-14**

**Germany**

**1 2.1**

**Other EU6**

**3.1 3 3.1**

**3.6 3.2 6.0 5.1 7.8 1.3 4.4 1.4**

**-8 -8 -8 -8 -8 -8 -8 -8**

**EU12**

**-0.1**

**EU15**

**-23**

**-18**

**EU25**

**-39**

**-32**

**EU27**

In this framework, as shown in **Figure 3a)**, the historical perspective of Russian federation emissions is straightforward: a huge decrease due to the former Soviet Union reconstruction until 1998 and a slow increase since then. The importance of the former Soviet Union reconstruction in the rapid decrease of European emissions is further confirmed by the profile of the evolution of German emissions4. Nevertheless, contrary to Russian federation,

4 Indeed, the former DRG experienced a huge decrease of emission in the beginning of the nineties, 50% decrease in terms of CO2 emission over the 1987 – 1993 period while the 1993 CO2 emissions level of the

former FRG is 2% more than 1987 level (United Nations, UNFCCC, 1994)



not rise up as soon as the crisis period has ended.

**30**

**15**

**8**

**Oceania**

**-33**

**-32**

**Russian Federation**

**Objectives (%) 1990-2008 share (%)**

**0**

**13**

**1**

**4.7**

**Japan**

**2.1 3.8 7.5 3.1**

**Germany, EU25, EU27)**

cumulated ratio perspective

**Without targets**

**Turkey and Belarus**

**13**

**United States**

**40**

**38**

**9.9 <sup>13</sup> <sup>15</sup>**

**24**

**-7 -6 -6**

**Canada**

German decreasing emission trend has been going on after 1998 though at a slower pace. **Figure 3a)** allows a clearer distinction between Japan and Oceania historical evolution of emissions. While Oceania emissions are growing at a high rate especially in the 1995-2000 periods, Japan experienced their main rise in emission in the 1990-1994 periods before remaining around the same level until a huge drop in 2008 which may be due to financial crisis. Therefore, contrary to Oceania, Japan, though not fulfilling in 2008 its Kyoto target, seems to have at least stabilized its emissions. Finally, France, after some chaotic variation corresponding to the 1993-1994 recession period, experienced a slow decrease in its emissions between 1998 and 2005 followed by a higher decrease during the 2005-2008 periods. Interestingly, the beginning of this higher decrease period coincides with the introduction into the French Vth republic constitution of an environment charter (Digithèque MJP) and the entry into force of the Kyoto Protocol. Since then, environment has taken a stronger place in French policy debate but it should be noted that in spite of these late efforts, France has not in 2008 met its Kyoto targets.

Fig. 3. a) A historical perspective of the evolution of GHG emissions compared to 2008 emission level.

b) Evolution of the structure worldwide energy-related emissions between 1990 and 2008

According to IEA report (IEA, 2010), the world carbon dioxide emissions associated with fuel combustion has increased by 40 percent during the period 1990 – 2008, from 21.0 GtCO2 in 1990 to 29.4 GtCO2 in 2008. During this period, the Annex I CO2 emissions associated with fuel combustion has remained the same, thus showing the difficulties to implement CO2 reduction when it comes to fuel combustion processes. Indeed, if the 2008 GHG emissions of Annex I countries are 5.2% less than those of 1990, their 2008 CO2 emissions are only 2.2% less than those of 1990. This demonstrates that most of the GHG emission reduction achieved concerned GHGs other than CO2 such as methane and nitrous oxide and that among CO2 emission reduction, those concerning fuel combustion has particularly difficult to achieve. Furthermore, while Annex I countries represented 66% of world CO2 emissions associated with fuel combustion in 1990, they represent only 47% of these in 2008.

The Greenhouse Stakes of Globalization 491

The fact that GNI is generally different from GDP is a sign of the globalization process. However, even if imports and exports have roughly doubled between 2000 and 2009, the main contributing countries to globalization covering more than 50% of worldwide monetary exports and imports remain those of Western Europe, East Asia, and Northern America1 thus creating three globalization driven attractive economic areas (dat). Nevertheless, the globalization process is still expanding and new globalization driven attractive economic areas will appear in the future. Notably, one of the consequences of this on-going process will result in an even more increasing inter-connection of countries GHG emissions than the one

presented in section **2.1**. Consequently, two perspectives are offered to policy-makers:

consumption (**GHGC**) and those for exports (**GHGE**).

(Lenzen, et al., 2007).

the developed Annex I countries.

 6 consumption (**GHGC**) and those associated with imports (**GHGI**).



The difference between CP and PP results from the difference of GHG emissions embodied in exports and those embodied in imports. Therefore, while PP policy options only aims at improving national production GHG efficiency, CP policy options are dependent on the trade partners' production GHG efficiency. Consequently, the CP policy options would have to be more collaborative raising the issue of emission allocations between actors

For 2004, Davis and Caldeira determined that 23% of CO2 worldwide emissions were traded internationally primarily as exports from china and other emerging markets to consumers in developed countries (Davis, et al., 2010). This result was further confirmed by Peters and Hertwich who determined that almost one-quarter of carbon dioxide emissions released to the atmosphere is emitted in the production of internationally traded goods and services (Peters, et al., 2008) thus calling for the introduction of trade within post-Kyoto policy debate (Peters, et al., 2008). Following this trade perspective at the Chinese national level, Ackerman determined that if Chinese exports and Chinese imports were produced with the U.S. carbon intensity, the CO2 trade surplus would be close to zero as CO2 emissions associated with Chinese exports would decreased by around 75%(Ackerman, 2009). Other studies regarding carbon embodied in trade were realized in developed countries6. All of these studies show as in **Figure 4** that CP leads to higher GHG emissions than PP for most of

On a policy level, this means the risk of carbon leakage is high as the developed countries could relocate GHG-intensive activities to a cheaper labor but less GHG efficient market and

Austria (Kratena, et al., 2010), U.S (Weber, et al., 2007), France(Lenglart, et al., 2010)

Consequently, post-Kyoto policies must necessarily consider that more parties should get emission reduction targets and not only Annex I countries. As shown in **Figure 3b)**, Asian countries and particularly China should be among the first to get these emission reduction targets as their associated fuel combustion CO2 emissions raised up from 19% of 1990 world emission to 34% of 2008 world emission. Indeed, when looking at the top20 CO2 emitters in 2007, which alone cover 81% of the world CO2 emissions, a worldwide pattern of CO2 emissions clearly appear, Asia (35%), North America (23%), Europe (15%), Middle and Near Eastern countries (3.1%), Africa (1.5%), Oceania (1.3%) and South America (1.3%)**5**. Furthermore, among the top20 CO2 emitters in 2007, Annex I countries represent 55% of the GHG emissions and only 30% if the United States who did not ratify the Kyoto protocol are taken out of Annex I countries emissions. Consequently, efficient post-Kyoto policies will have to deal with the increasing distribution of GHG emissions worldwide.

The globalization of trade and associated GHG emissions are logically connected to an increase of GHG emissions associated with international transportation. Actually, the 2008 CO2 emission associated with international marine bunkers fuel combustion are 63% higher than those of 1990 while those of international aviation bunkers have increased by 76% over the same period. Moreover, transport represents 22% of world CO2 emissions in 2008 thus constituting an important potential for mitigation policy. Nevertheless, global demand for transport appears unlikely to decrease in the foreseeable future as the *WEO 2009* projects that transport will grow by 45% by 2030(IEA, 2010). Noting that international bunkers emissions are not currently part of Kyoto protocol framework, the observed globalization of GHG emissions calls for the development of methodologies able to track down national and international bunkers emissions throughout the patterns of trade so as to ensure the best post-Kyoto GHG emission reduction international policy. However, current methodologies regarding GHG emission accounting framework mostly focus on national emission patterns and miss the trade perspective. Consequently, the next section presents the evolution of perspective required by the introduction of trade within the GHG emission accounting framework. Then section **2.3** focuses on the transportation aspects needed to frame a comprehensive macro-economic GHG emission accounting framework.

#### **2.2 Production or consumption perspectives?**

The GHG emission reduction targets of Annex I countries is based on a national GHG accounting framework which is thus production driven since all emissions resulting from national residents activities are assigned to the associated country. Particularly, this means that American companies, which are resident in France will contribute to French GHG emissions and not American ones. In the same perspective, French tourists visiting the United States contribute to American GHG emissions. This framework is thus in line with the Gross Domestic Product (GDP) framework which aggregates in monetary terms the results of residents' production but not with the Gross National Income (GNI) which would associate American companies in France with the United States.

<sup>5</sup> **Asia**: China (22%), India (5.5%), Japan\* (4.3%), Republic of Korea (1.7%), Indonesia (1.4%) ; **Europe**: Russian Federation\* (5.2%), Germany\* (2.7%), United Kingdom\* (1.8%), Italy\* (1.6%), France\* (1.3%), Spain\* (1.2%), Ukraine\* (1.1%) ; **North America**: United States\* (20%), Canada\* (1.9%), Mexico (1.6%) ; **Middle and Near Eastern countries**: Islamic republic of Iran (1.7%), Saudi Arabia (1.4%) ; **Africa**: South Africa (1.5%) ; **Oceania**: Australia\* (1.3%) ; **South America**: Brazil (1.3%) *\*Annex I countries*

Consequently, post-Kyoto policies must necessarily consider that more parties should get emission reduction targets and not only Annex I countries. As shown in **Figure 3b)**, Asian countries and particularly China should be among the first to get these emission reduction targets as their associated fuel combustion CO2 emissions raised up from 19% of 1990 world emission to 34% of 2008 world emission. Indeed, when looking at the top20 CO2 emitters in 2007, which alone cover 81% of the world CO2 emissions, a worldwide pattern of CO2 emissions clearly appear, Asia (35%), North America (23%), Europe (15%), Middle and Near Eastern countries (3.1%), Africa (1.5%), Oceania (1.3%) and South America (1.3%)**5**. Furthermore, among the top20 CO2 emitters in 2007, Annex I countries represent 55% of the GHG emissions and only 30% if the United States who did not ratify the Kyoto protocol are taken out of Annex I countries emissions. Consequently, efficient post-Kyoto policies will

The globalization of trade and associated GHG emissions are logically connected to an increase of GHG emissions associated with international transportation. Actually, the 2008 CO2 emission associated with international marine bunkers fuel combustion are 63% higher than those of 1990 while those of international aviation bunkers have increased by 76% over the same period. Moreover, transport represents 22% of world CO2 emissions in 2008 thus constituting an important potential for mitigation policy. Nevertheless, global demand for transport appears unlikely to decrease in the foreseeable future as the *WEO 2009* projects that transport will grow by 45% by 2030(IEA, 2010). Noting that international bunkers emissions are not currently part of Kyoto protocol framework, the observed globalization of GHG emissions calls for the development of methodologies able to track down national and international bunkers emissions throughout the patterns of trade so as to ensure the best post-Kyoto GHG emission reduction international policy. However, current methodologies regarding GHG emission accounting framework mostly focus on national emission patterns and miss the trade perspective. Consequently, the next section presents the evolution of perspective required by the introduction of trade within the GHG emission accounting framework. Then section **2.3** focuses on the transportation aspects needed to frame a

The GHG emission reduction targets of Annex I countries is based on a national GHG accounting framework which is thus production driven since all emissions resulting from national residents activities are assigned to the associated country. Particularly, this means that American companies, which are resident in France will contribute to French GHG emissions and not American ones. In the same perspective, French tourists visiting the United States contribute to American GHG emissions. This framework is thus in line with the Gross Domestic Product (GDP) framework which aggregates in monetary terms the results of residents' production but not with the Gross National Income (GNI) which would

<sup>5</sup> **Asia**: China (22%), India (5.5%), Japan\* (4.3%), Republic of Korea (1.7%), Indonesia (1.4%) ; **Europe**: Russian Federation\* (5.2%), Germany\* (2.7%), United Kingdom\* (1.8%), Italy\* (1.6%), France\* (1.3%), Spain\* (1.2%), Ukraine\* (1.1%) ; **North America**: United States\* (20%), Canada\* (1.9%), Mexico (1.6%) ; **Middle and Near Eastern countries**: Islamic republic of Iran (1.7%), Saudi Arabia (1.4%) ; **Africa**: South

Africa (1.5%) ; **Oceania**: Australia\* (1.3%) ; **South America**: Brazil (1.3%) *\*Annex I countries*

have to deal with the increasing distribution of GHG emissions worldwide.

comprehensive macro-economic GHG emission accounting framework.

associate American companies in France with the United States.

**2.2 Production or consumption perspectives?** 

The fact that GNI is generally different from GDP is a sign of the globalization process. However, even if imports and exports have roughly doubled between 2000 and 2009, the main contributing countries to globalization covering more than 50% of worldwide monetary exports and imports remain those of Western Europe, East Asia, and Northern America1 thus creating three globalization driven attractive economic areas (dat). Nevertheless, the globalization process is still expanding and new globalization driven attractive economic areas will appear in the future. Notably, one of the consequences of this on-going process will result in an even more increasing inter-connection of countries GHG emissions than the one presented in section **2.1**. Consequently, two perspectives are offered to policy-makers:


The difference between CP and PP results from the difference of GHG emissions embodied in exports and those embodied in imports. Therefore, while PP policy options only aims at improving national production GHG efficiency, CP policy options are dependent on the trade partners' production GHG efficiency. Consequently, the CP policy options would have to be more collaborative raising the issue of emission allocations between actors (Lenzen, et al., 2007).

For 2004, Davis and Caldeira determined that 23% of CO2 worldwide emissions were traded internationally primarily as exports from china and other emerging markets to consumers in developed countries (Davis, et al., 2010). This result was further confirmed by Peters and Hertwich who determined that almost one-quarter of carbon dioxide emissions released to the atmosphere is emitted in the production of internationally traded goods and services (Peters, et al., 2008) thus calling for the introduction of trade within post-Kyoto policy debate (Peters, et al., 2008). Following this trade perspective at the Chinese national level, Ackerman determined that if Chinese exports and Chinese imports were produced with the U.S. carbon intensity, the CO2 trade surplus would be close to zero as CO2 emissions associated with Chinese exports would decreased by around 75%(Ackerman, 2009). Other studies regarding carbon embodied in trade were realized in developed countries6. All of these studies show as in **Figure 4** that CP leads to higher GHG emissions than PP for most of the developed Annex I countries.

On a policy level, this means the risk of carbon leakage is high as the developed countries could relocate GHG-intensive activities to a cheaper labor but less GHG efficient market and

<sup>6</sup> Austria (Kratena, et al., 2010), U.S (Weber, et al., 2007), France(Lenglart, et al., 2010)

The Greenhouse Stakes of Globalization 493



In PP, national transportation analysis is straightforward as it can be associated with a particular country. However, it should be noted the transport sector is often not well represented within the national accounting framework leading to the making of IO tables. Moreover, international transport appears difficult to analyze, as they constitute by their international nature a complex allocation problem. To adopt CP changes the way to deal with the transportation allocation problem. Indeed, in this perspective, transportation is seen as part of a product story of life cycle. Consequently, the associated emissions go to the country that consumes the transported goods. However, may it be CP or PP, there is a need for the development of methodologies that better integrate freight transportation. Section 2.3 provides an overview of the last researches done in this direction. Sections 3.1 and 3.2 respectively present the carbon labeling policy framework within which transportation

**2.3 Towards a better integration of GHG emissions from freight transportation** 

application of physical based sounded ratio such as gCO2/Tkm ratios.

The MRIO framework explained in **2.2** though enabling the tracking of emissions embodied in trade with technology differentiations between nations misses to specifically address freight transportation. However, transportation key role in global trade and associated GHG emissions have been increasing ever since 1990. Therefore, new models have been developed to specifically address the issues associated with freight transportation such as food-miles and the connection between product, location of production, choice of transport mode and GHG emissions. These models are based on the IO framework but further differentiate transportation by applying a vector characterizing the Ton-kilometers (Tkm) driven for each sector before calculating the GHG emission associated through the

For example, Webers and Matthews applied a single region IO framework to the U.S. together with an extended transport model to determine the impacts of transportation within the supply chain of food products (Weber, et al., 2008). They found the food-miles problem, namely the trade-off between buying locally or globally, was less important in terms of GHG emissions than the shifting diet problem. Therefore, they confirm the predominance of production climate impact over transportation climate impact for food products. Notably, their study pointed out red meat products top scoring on climate impact is only third when it comes to transportation GHG emissions. Indeed, transportation emissions associated with red mead was found within the same range of fruits, vegetable,

content is cleaner than the German one.

induce production within this country.

takes its full meaning and the related allocation issues.

data intensive.

and cereals products.

carbon efficiency of economies may vary a lot between developed and developing countries but between developed countries. For example, the French electricity carbon

Fig. 4. Carbon trade balance for main GHG emitters' countries ( Sustainable Consumption institute of the University of Manchester)

then import high GHG embodied goods, which are currently not accounted in the Kyoto framework. Therefore, it would be more politically sounded to evaluate CP emissions as it could help convince new emerging countries to join Annex I and get emission reduction targets. Furthermore, a comprehensive framework of the trade patterns is in line with the micro-economic perspective of life cycle assessment (LCA). Indeed, CP corresponds to the product perspective of carbon labeling policy, further explained in section **3.1**., and thus offers possibilities of bi or multi-lateral agreement between countries on the way to achieve reduction of GHG emissions for particular products or sectors.

Nevertheless, from a methodology perspective, CP is more data intensive than PP. Indeed, language, policy issues, non-recorded data, or data non-recorded at the same detailed level are examples of the barriers encountered to comprehensively associate data from trading countries. Furthermore, contrary to PP which only accounts in an aggregate way the national emissions, CP supposes to be able to track down imports, exports and production for national consumption separately thus raising the issue of data availability. Finally, though many environmental assessment methodologies are available such as material flow analysis (MFA), life cycle assessment (LCA) and input-output analysis (IOA) (Hertwich, 2005)(Wiedmann, et al., 2006), they do not cover the same perspectives. LCA and MFA are bottom-up approaches, consequently more suitable to micro-economic or company level, whereas IOA is a top-down approach more suitable to a macro-perspective. Particularly, multi-regional Input-Output analysis (MRIO) lies at the core of CP (Hertwich, et al., 2004), implemented in three ways depending on data availability:


**Carbon trade balance**

Fig. 4. Carbon trade balance for main GHG emitters' countries ( Sustainable Consumption

**6% 8%**

**15%**

**20% 20%**

**23% 24% 26%**

then import high GHG embodied goods, which are currently not accounted in the Kyoto framework. Therefore, it would be more politically sounded to evaluate CP emissions as it could help convince new emerging countries to join Annex I and get emission reduction targets. Furthermore, a comprehensive framework of the trade patterns is in line with the micro-economic perspective of life cycle assessment (LCA). Indeed, CP corresponds to the product perspective of carbon labeling policy, further explained in section **3.1**., and thus offers possibilities of bi or multi-lateral agreement between countries on the way to achieve

Nevertheless, from a methodology perspective, CP is more data intensive than PP. Indeed, language, policy issues, non-recorded data, or data non-recorded at the same detailed level are examples of the barriers encountered to comprehensively associate data from trading countries. Furthermore, contrary to PP which only accounts in an aggregate way the national emissions, CP supposes to be able to track down imports, exports and production for national consumption separately thus raising the issue of data availability. Finally, though many environmental assessment methodologies are available such as material flow analysis (MFA), life cycle assessment (LCA) and input-output analysis (IOA) (Hertwich, 2005)(Wiedmann, et al., 2006), they do not cover the same perspectives. LCA and MFA are bottom-up approaches, consequently more suitable to micro-economic or company level, whereas IOA is a top-down approach more suitable to a macro-perspective. Particularly, multi-regional Input-Output analysis (MRIO) lies at the core of CP (Hertwich, et al., 2004),


institute of the University of Manchester)

**-12%**

**-18%**

**-21%**

reduction of GHG emissions for particular products or sectors.

**-7% -5%**

implemented in three ways depending on data availability:

carbon efficiency of economies may vary a lot between developed and developing countries but between developed countries. For example, the French electricity carbon content is cleaner than the German one.


In PP, national transportation analysis is straightforward as it can be associated with a particular country. However, it should be noted the transport sector is often not well represented within the national accounting framework leading to the making of IO tables. Moreover, international transport appears difficult to analyze, as they constitute by their international nature a complex allocation problem. To adopt CP changes the way to deal with the transportation allocation problem. Indeed, in this perspective, transportation is seen as part of a product story of life cycle. Consequently, the associated emissions go to the country that consumes the transported goods. However, may it be CP or PP, there is a need for the development of methodologies that better integrate freight transportation. Section 2.3 provides an overview of the last researches done in this direction. Sections 3.1 and 3.2 respectively present the carbon labeling policy framework within which transportation takes its full meaning and the related allocation issues.

#### **2.3 Towards a better integration of GHG emissions from freight transportation**

The MRIO framework explained in **2.2** though enabling the tracking of emissions embodied in trade with technology differentiations between nations misses to specifically address freight transportation. However, transportation key role in global trade and associated GHG emissions have been increasing ever since 1990. Therefore, new models have been developed to specifically address the issues associated with freight transportation such as food-miles and the connection between product, location of production, choice of transport mode and GHG emissions. These models are based on the IO framework but further differentiate transportation by applying a vector characterizing the Ton-kilometers (Tkm) driven for each sector before calculating the GHG emission associated through the application of physical based sounded ratio such as gCO2/Tkm ratios.

For example, Webers and Matthews applied a single region IO framework to the U.S. together with an extended transport model to determine the impacts of transportation within the supply chain of food products (Weber, et al., 2008). They found the food-miles problem, namely the trade-off between buying locally or globally, was less important in terms of GHG emissions than the shifting diet problem. Therefore, they confirm the predominance of production climate impact over transportation climate impact for food products. Notably, their study pointed out red meat products top scoring on climate impact is only third when it comes to transportation GHG emissions. Indeed, transportation emissions associated with red mead was found within the same range of fruits, vegetable, and cereals products.

The Greenhouse Stakes of Globalization 495

Fig. 5. Production and transportation emissions associated with French household

**3. How to frame GHG reduction and ensure a greening of supply chains** 

At the most basic level carbon labeling involves associating an amount of CO2 equivalent emissions with a product. Though seemingly simple in principle, carbon labeling schemes raise challenges upon application. These challenges make it difficult to standardize a method for quantifying CO2 equivalents for carbon labeling and open the possibility of

Carbon labeling is closely tied to the complex relationship between companies, consumers, and policy makers. In section **2.2** we discussed two options for framing carbon labels, the producer perspective (PP) and the consumer perspective (CP). PP implies setting guidelines that move companies towards improving carbon efficiencies. These could take the form of thresholds, limitations, bans, externality-related taxation schemes, subsidies for cleaner production processes, and research efforts focused on improving or replacing dirty processes. CP shifts the focus to that of a consumer who is concerned with the overall impact of each product rather than the specific impacts of nodes in the supply network. In contrast to PP where policy actions focus on enhancing the carbon efficiency of individual actors or sectors based on their relevance to total territorial emissions, CP policy is focused on the community of actors involved in an individual product life cycle or basket of goods. From the perspective of a particular company, there are two strategic advantages to viewing their own environmental impacts from a consumer perspective and having policy-makers do so as well.

**Quantifiable evidence of environmental performance is valued by customers:** 

Consumers choose products based on the information they receive. All else being equal, a consumer should choose the least costly option. However, the consumer's perception of

**3.1 The scientific, business and policy challenges of carbon labeling** 

consumption (Hawkins and Dente, 2010)

litigation and controversy upon implementation.

In their study on greenhouse gas emissions driven by the transportation of goods associated with French household consumption, Hawkins and Dente constructed a without reimport/re-export two regional IO model extended with a transportation model accounting for 116 sectors, 175 product types, 5 transport modes, 22 French regions and 42 foreign countries (Hawkins, et al., 2010). The imports coming from the 42 foreign countries were assumed to be produced using German technology, as IO tables were not available at a sufficient level of details for all the 42 countries. It was found that French household consumption was responsible for the emissions of 627 MtCO2e in 2004, 54% more than the 2005 French territorial CO2 emissions and 14% more than the total CO2 emissions driven by total 2005 French final demand as determined by the French National Institute of Statistics and Economic Studies (INSEE) through a six regional classical IO model including Germany, Italy, Belgium, U.K. and Spain in addition to France. Nevertheless, the INSEE study provided low values for agriculture since only dealing with CO2 emissions and not GHG emissions contrary to the Hawkins and Dente study which showed that when dealing with GHG emissions, the agriculture sector accounted for as much as 23% of production and transport related emissions (cf. **Figure 5**). This difference in scope explains most of the result differences observed between the two studies, the rest of the difference being explained through a number of factors: better representation of national and international transportation with the specific transportation model for the Hawkins and Dente study, better representation of the technology differentiation for the INSEE studies as more regional IO tables are taken into account. Furthermore, it is interesting to notice the Hawkins and Dente study estimated to 14MtCO2e the transportation occurring abroad for the purpose of French household consumption.

The above examples shows the use of transport specific models allows for a better tracking of flows within the supply chain of global economies and better represents the patterns of trade and international transportation, the fundamental features of globalization. Moreover, coupled with comprehensive MRIO models, these models allow for differentiation between indirect and direct emissions, transport modes emissions, sector and product differentiations, linkages between the final demand and the CO2 global economic chain. **Figure 5** provides a good description of the achievement available through the application of such models, which provide important analyses and indications for national and supranational policies such as national emission reduction planning and the Kyoto Protocol.

However, the application of transportation model supposes not only to have data to lead a comprehensive MRIO framework but data from transport satellite accounts for all countries. Therefore, the main problem is data availability and initiatives are made notably at the European level through the EXIOPOL project to offer harmonized detailed IO tables for all the EU-27 members (htt4). In this perspective, it should not be forgotten that non-Annex I Kyoto parties were encouraged to establish statistical competencies for the evaluation of their GHG emissions (United Nations). Moreover, the gCO2/tkm ratio taken within transportation models can vary greatly depending on the goods transported and the upstream logistic choices, which determined load factor and empty trips ratio. Thus, take an average cross-technological gCO2/tkm ratio can lead to errors in the determination of emissions and data from meso/micro levels are needed to address the problem of the greening of global supply chains. These questions are the purpose of section 3.

In their study on greenhouse gas emissions driven by the transportation of goods associated with French household consumption, Hawkins and Dente constructed a without reimport/re-export two regional IO model extended with a transportation model accounting for 116 sectors, 175 product types, 5 transport modes, 22 French regions and 42 foreign countries (Hawkins, et al., 2010). The imports coming from the 42 foreign countries were assumed to be produced using German technology, as IO tables were not available at a sufficient level of details for all the 42 countries. It was found that French household consumption was responsible for the emissions of 627 MtCO2e in 2004, 54% more than the 2005 French territorial CO2 emissions and 14% more than the total CO2 emissions driven by total 2005 French final demand as determined by the French National Institute of Statistics and Economic Studies (INSEE) through a six regional classical IO model including Germany, Italy, Belgium, U.K. and Spain in addition to France. Nevertheless, the INSEE study provided low values for agriculture since only dealing with CO2 emissions and not GHG emissions contrary to the Hawkins and Dente study which showed that when dealing with GHG emissions, the agriculture sector accounted for as much as 23% of production and transport related emissions (cf. **Figure 5**). This difference in scope explains most of the result differences observed between the two studies, the rest of the difference being explained through a number of factors: better representation of national and international transportation with the specific transportation model for the Hawkins and Dente study, better representation of the technology differentiation for the INSEE studies as more regional IO tables are taken into account. Furthermore, it is interesting to notice the Hawkins and Dente study estimated to 14MtCO2e the transportation occurring abroad for

The above examples shows the use of transport specific models allows for a better tracking of flows within the supply chain of global economies and better represents the patterns of trade and international transportation, the fundamental features of globalization. Moreover, coupled with comprehensive MRIO models, these models allow for differentiation between indirect and direct emissions, transport modes emissions, sector and product differentiations, linkages between the final demand and the CO2 global economic chain. **Figure 5** provides a good description of the achievement available through the application of such models, which provide important analyses and indications for national and supranational policies such as national emission reduction planning and the Kyoto Protocol.

However, the application of transportation model supposes not only to have data to lead a comprehensive MRIO framework but data from transport satellite accounts for all countries. Therefore, the main problem is data availability and initiatives are made notably at the European level through the EXIOPOL project to offer harmonized detailed IO tables for all the EU-27 members (htt4). In this perspective, it should not be forgotten that non-Annex I Kyoto parties were encouraged to establish statistical competencies for the evaluation of their GHG emissions (United Nations). Moreover, the gCO2/tkm ratio taken within transportation models can vary greatly depending on the goods transported and the upstream logistic choices, which determined load factor and empty trips ratio. Thus, take an average cross-technological gCO2/tkm ratio can lead to errors in the determination of emissions and data from meso/micro levels are needed to address the problem of the

greening of global supply chains. These questions are the purpose of section 3.

the purpose of French household consumption.

Fig. 5. Production and transportation emissions associated with French household consumption (Hawkins and Dente, 2010)

## **3. How to frame GHG reduction and ensure a greening of supply chains**

## **3.1 The scientific, business and policy challenges of carbon labeling**

At the most basic level carbon labeling involves associating an amount of CO2 equivalent emissions with a product. Though seemingly simple in principle, carbon labeling schemes raise challenges upon application. These challenges make it difficult to standardize a method for quantifying CO2 equivalents for carbon labeling and open the possibility of litigation and controversy upon implementation.

Carbon labeling is closely tied to the complex relationship between companies, consumers, and policy makers. In section **2.2** we discussed two options for framing carbon labels, the producer perspective (PP) and the consumer perspective (CP). PP implies setting guidelines that move companies towards improving carbon efficiencies. These could take the form of thresholds, limitations, bans, externality-related taxation schemes, subsidies for cleaner production processes, and research efforts focused on improving or replacing dirty processes. CP shifts the focus to that of a consumer who is concerned with the overall impact of each product rather than the specific impacts of nodes in the supply network. In contrast to PP where policy actions focus on enhancing the carbon efficiency of individual actors or sectors based on their relevance to total territorial emissions, CP policy is focused on the community of actors involved in an individual product life cycle or basket of goods. From the perspective of a particular company, there are two strategic advantages to viewing their own environmental impacts from a consumer perspective and having policy-makers do so as well.

#### **Quantifiable evidence of environmental performance is valued by customers:**

Consumers choose products based on the information they receive. All else being equal, a consumer should choose the least costly option. However, the consumer's perception of

The Greenhouse Stakes of Globalization 497

(Reg11). More recently in France, the Operational Committee on Consumption, Ecological Price and Competitive Advantage of the "Grenelle de l'environnement" announced that environmental information will be required for each product beginning 1st January 2011 (Ministère de l'Ecologie, du Développement durable, des transports et du logement). Though this goal is still in the process of being realized a proposition within the consumption code approved by the French Senate (Article 85) requires carbon cost labeling (htt1). This will be

To answer the raising environmental awareness of their consumers and the compulsory labeling policies driven by policy makers, businesses nevertheless took actions towards the reporting of their emissions, as the increasing number of carbon accounting tools tends to prove. An international benchmarking of carbon accounting tools realized by the French Environment and Energy Management Agency (ADEME) shows that this new market of environmental accounting tools concerns mainly Annex I countries of the Kyoto Protocol7 and the United Nations Environmental Program (UNEP)(ADEME, 2008). On the 62 accounting tools reviewed, 10 are ONG's, 39 related to private businesses activities and 13 are governments'. Consequently, there is seemingly some serious environmental commitment from business companies. Nevertheless, while governments' and ONG's solutions are oriented towards emission reduction, a large majority of private accounting tools corresponds to the needs of companies to compensate emissions thus not enabling exhaustive emission accounting. Furthermore, most of the solutions are site and territory oriented since only ten of the accounting tools reviewed includes a product orientation. This demonstrates that today's accounting tools are in line with the policy framework developed by nations to achieve their Kyoto targets but are not sufficient to embrace the policy objectives of carbon labeling neither those of sustainable consumption. They lack interconnection with the different aspects of their activities to comprehensively accumulate emissions from each node of the supply chain and

However, ideally, the link between PP carbon accounting and CP carbon labeling is straightforward. Actually, let us assume that each company implied in the making and delivery of a product to the final retailer8 is able to provide for each of their products the associated amount of CO2 equivalent emitted. If we consider that N companies are implied in the making and delivery of product i and M as the total number of different products produced by these N companies, our previous assumption allows us to build a M times N dimension table (CO2). Coefficient CO���� represents the amount of CO2 equivalent associated with product i emitted by company j. In this mathematical framework, the PP carbon accounting framework appears as a sum across the rows �CO���� while the CP carbon-

CO���� ⋯ CO���� ⋯ CO����

⋱ ⋮ ⋱ ⋯ CO���� ⋯ ⋱ ⋮ ⋱

CO���� ⋯ CO���� ⋯ CO���� �

8 The final retailer constitutes the final place of purchase but not of consumption. This distinction is

⋮ CO���� ⋮

�� �

(1)

labeling framework corresponds to a sum across the columns �CO����.

⋮ CO���� ⋮

�

7 France, Germany, U.K., Australia, U.S.A, Canada, Netherlands, Japan

�� �

��� =

important for future researches on the subject.

provided on an experimental basis beginning in July 2011 (Afnor-Normalisation).

allocate them to the final consumer products.

characteristics such as quality, durability, aesthetics, others' perceptions, and a number of other factors combine with price in the final decision of which items to purchase. In the absence of reliable environmental information, as is the case today, it might appear that consumers do not factor the environmental impacts of production processes into purchasing decisions (Llerena, et al., 2007). However, as rigorous environmental indicators find their way into the marketplace, the willingness to pay for improvements in the life cycle environmental quality of products will increase as a result of consumers' desire to convey a positive image to others through their choices and perhaps through an internal desire to view themselves as altruistic. When confronted by otherwise equivalent products resulting from different production processes, a well-informed consumer would not be as indifferent as might be predicted by a narrow interpretation of traditional economic theory. Within a society whose environmental conscience is growing, as is the case today, companies that provide information about environmental performance, demonstrate transparency, and continue to improve win market share over time to those that do not.

#### **The enhanced efficiency of supply chains:**

The usual framework to analyze competitiveness is to compare the cost structure and competitive advantage of actors producing similar products with the goal of becoming more cost-efficient while maintaining customer satisfaction. In the current globalized world, this framework still holds true but has evolved from the company level to the supply chain level. Each actor of the supply chain contributes to the comparative advantage of the entire network leading to product sales. In such a complex system often, local optimization does not correspond with the optimization of the global system and can even result in worsening the previous situation. First generation biofuels and the replacement of lead solder exemplify how a narrow approach to the optimization can result in global worsening. While biofuels were seen as attractive replacements for fossil fuels because carbon uptake by plant biomass would off-set the releases upon combustion, further analysis suggests the combined impacts of inputs to agricultural production, soil carbon loss, and other factors may well drive an overall higher carbon footprint. In the case of solder, lead was replaced by costlier substitutes such as silver, with greater extraction- and processing-related impacts, and potentially the same or higher toxicity considering exposure routes along their supply chains. Although these examples are controversial and dependent on which factors are considered in the optimization, we present them here for illustrative purposes.

Although the scope of decision-making for a company lies within its own operations, considering each company relies on the supply network within which it exists, the consumer perspective better addresses overall competitive advantage. Furthermore, CP does not exclude improvements in the carbon efficiency of individual actors, but through the broader perspective, it deals with offers the possibility to transform the investments made in these improvements into market strengths.

In spite of the expected benefits from the implementation of CP for companies, most of the efforts so far have been policy driven. For example, the EU sustainable consumption and production, which includes eco-labeling, aims at reducing the negative impact of consumption on the environment, health, climate, and natural resources. On 24th November 2009, the European parliament resolved to widen the EU eco-label scope in terms of impacts and product categories to avoid the proliferation of environmental labeling schemes and to encourage improvement in all sectors for which environmental impact affects consumer choice

characteristics such as quality, durability, aesthetics, others' perceptions, and a number of other factors combine with price in the final decision of which items to purchase. In the absence of reliable environmental information, as is the case today, it might appear that consumers do not factor the environmental impacts of production processes into purchasing decisions (Llerena, et al., 2007). However, as rigorous environmental indicators find their way into the marketplace, the willingness to pay for improvements in the life cycle environmental quality of products will increase as a result of consumers' desire to convey a positive image to others through their choices and perhaps through an internal desire to view themselves as altruistic. When confronted by otherwise equivalent products resulting from different production processes, a well-informed consumer would not be as indifferent as might be predicted by a narrow interpretation of traditional economic theory. Within a society whose environmental conscience is growing, as is the case today, companies that provide information about environmental performance, demonstrate transparency, and

The usual framework to analyze competitiveness is to compare the cost structure and competitive advantage of actors producing similar products with the goal of becoming more cost-efficient while maintaining customer satisfaction. In the current globalized world, this framework still holds true but has evolved from the company level to the supply chain level. Each actor of the supply chain contributes to the comparative advantage of the entire network leading to product sales. In such a complex system often, local optimization does not correspond with the optimization of the global system and can even result in worsening the previous situation. First generation biofuels and the replacement of lead solder exemplify how a narrow approach to the optimization can result in global worsening. While biofuels were seen as attractive replacements for fossil fuels because carbon uptake by plant biomass would off-set the releases upon combustion, further analysis suggests the combined impacts of inputs to agricultural production, soil carbon loss, and other factors may well drive an overall higher carbon footprint. In the case of solder, lead was replaced by costlier substitutes such as silver, with greater extraction- and processing-related impacts, and potentially the same or higher toxicity considering exposure routes along their supply chains. Although these examples are controversial and dependent on which factors are

continue to improve win market share over time to those that do not.

considered in the optimization, we present them here for illustrative purposes.

Although the scope of decision-making for a company lies within its own operations, considering each company relies on the supply network within which it exists, the consumer perspective better addresses overall competitive advantage. Furthermore, CP does not exclude improvements in the carbon efficiency of individual actors, but through the broader perspective, it deals with offers the possibility to transform the investments made in these

In spite of the expected benefits from the implementation of CP for companies, most of the efforts so far have been policy driven. For example, the EU sustainable consumption and production, which includes eco-labeling, aims at reducing the negative impact of consumption on the environment, health, climate, and natural resources. On 24th November 2009, the European parliament resolved to widen the EU eco-label scope in terms of impacts and product categories to avoid the proliferation of environmental labeling schemes and to encourage improvement in all sectors for which environmental impact affects consumer choice

**The enhanced efficiency of supply chains:** 

improvements into market strengths.

(Reg11). More recently in France, the Operational Committee on Consumption, Ecological Price and Competitive Advantage of the "Grenelle de l'environnement" announced that environmental information will be required for each product beginning 1st January 2011 (Ministère de l'Ecologie, du Développement durable, des transports et du logement). Though this goal is still in the process of being realized a proposition within the consumption code approved by the French Senate (Article 85) requires carbon cost labeling (htt1). This will be provided on an experimental basis beginning in July 2011 (Afnor-Normalisation).

To answer the raising environmental awareness of their consumers and the compulsory labeling policies driven by policy makers, businesses nevertheless took actions towards the reporting of their emissions, as the increasing number of carbon accounting tools tends to prove. An international benchmarking of carbon accounting tools realized by the French Environment and Energy Management Agency (ADEME) shows that this new market of environmental accounting tools concerns mainly Annex I countries of the Kyoto Protocol7 and the United Nations Environmental Program (UNEP)(ADEME, 2008). On the 62 accounting tools reviewed, 10 are ONG's, 39 related to private businesses activities and 13 are governments'. Consequently, there is seemingly some serious environmental commitment from business companies. Nevertheless, while governments' and ONG's solutions are oriented towards emission reduction, a large majority of private accounting tools corresponds to the needs of companies to compensate emissions thus not enabling exhaustive emission accounting. Furthermore, most of the solutions are site and territory oriented since only ten of the accounting tools reviewed includes a product orientation. This demonstrates that today's accounting tools are in line with the policy framework developed by nations to achieve their Kyoto targets but are not sufficient to embrace the policy objectives of carbon labeling neither those of sustainable consumption. They lack interconnection with the different aspects of their activities to comprehensively accumulate emissions from each node of the supply chain and allocate them to the final consumer products.

However, ideally, the link between PP carbon accounting and CP carbon labeling is straightforward. Actually, let us assume that each company implied in the making and delivery of a product to the final retailer8 is able to provide for each of their products the associated amount of CO2 equivalent emitted. If we consider that N companies are implied in the making and delivery of product i and M as the total number of different products produced by these N companies, our previous assumption allows us to build a M times N dimension table (CO2). Coefficient CO���� represents the amount of CO2 equivalent associated with product i emitted by company j. In this mathematical framework, the PP carbon accounting framework appears as a sum across the rows �CO���� while the CP carbonlabeling framework corresponds to a sum across the columns �CO����.

$$\text{CO}\_2 = \begin{pmatrix} \text{CO}\_{\text{2}\_{1,1}} & \cdots & \text{CO}\_{\text{2}\_{1,\text{J}}} & \cdots & \text{CO}\_{\text{2}\_{1,\text{N}}}\\ \vdots & \ddots & \vdots & \ddots & \vdots\\ \text{CO}\_{\text{2}\_{\text{i}\_{\text{i}1}}} & \cdots & \text{CO}\_{\text{2}\_{\text{i}\text{J}}} & \cdots & \text{CO}\_{\text{2}\_{\text{i}\text{N}}}\\ \vdots & \ddots & \vdots & \ddots & \vdots\\ \text{CO}\_{\text{2}\_{\text{M}\_{\text{i}1}}} & \cdots & \text{CO}\_{\text{2}\_{\text{M}\_{\text{J}}}} & \cdots & \text{CO}\_{\text{2}\_{\text{M}\_{\text{N}}}} \end{pmatrix}\_{\text{(1)}} \tag{1}$$

<sup>7</sup> France, Germany, U.K., Australia, U.S.A, Canada, Netherlands, Japan

<sup>8</sup> The final retailer constitutes the final place of purchase but not of consumption. This distinction is important for future researches on the subject.

The Greenhouse Stakes of Globalization 499

temperature pressure and air/gasoline ratio and consequently would need to be associated with real-time measurements. Therefore, an uncertainty is always attached to the use of a physical model so that two companies using the same physical model and reporting two different GHG emission factor within the uncertainty range should be considered as having the same emission factor. Furthermore, direct measurement of GHG emissions should be preferred in the case of highly variable physical phenomena for which physical model are

The management complexity consists in the organization of industry activities, the daily management of tasks and business travels together with the choices of processes. These decisions influence a lot the GHG efficiency of a company. Therefore, the allocation issue consists here in associating emissions with decision-making processes and is dependent on the time-framework chosen for the analysis. Electricity gives a good illustration of this point.

factor determined for the power plant k based on primary energy source j, the carbon

Getting the level of carbon emission at the yearly level would imply to sum the previous framework over time. In this ideal case, we thus obtained a comprehensive framework of emission per energy source with a timeline result. However, in present days, these kind of accurate data are not easily available. The electricity productions for each power plant are often given at the yearly level. Furthermore, there is often only a simple distinction of emission factors based on energy sources. The carbon content of 1 kWh delivered by the

� �������� � ������� ���� ∑ ����

���������������� ������� � ∑ ��������� � �������

CO2���������� = <sup>∑</sup> �����

���� <sup>=</sup> <sup>∑</sup> �����

The difference between (4) and (3) exemplifies how important the time-management framework chosen can influence a GHG analysis. Furthermore, one cannot differentiate electrons on the electric network and it is thus physically not possible to source the origin of electricity used on the network to a specific power plant. Therefore, no one on the network use cleaner energy than the other at a given time. The ways to get cleaner electricity


The example of electricity demonstrates the difficulty of an accurate calculation and allocation of GHG emissions. First in terms of physical complexity, there is a need for standardization of the assumptions used in physical models. The lack of standardization and the difference of scope between studies are indeed the main features of green washing. By green washing, we define any communication using abusively ecological argumentation

�

� ������� ��� (3)

is the per kWh emission

(4)

� �

� �

Indeed, if we consider real-time measurement, assuming that CO2�

CO2�������



content of 1kWh delivered to the network is:

network is thus constant over the year:

delivered at one particular moment are thus:

����� � �

difficult to implement.

$$\text{CO}\_{2,l} = \sum\_{l=1}^{\text{N}} \text{CO}\_{\text{2}\_{\text{l}\downarrow}} \qquad \text{and} \quad \text{CO}\_{2,f} = \sum\_{l=1}^{\text{M}} \text{CO}\_{\text{2}\_{\text{l}\downarrow}} \tag{2}$$

However, the establishment of these coefficients is rarely straightforward. Consequently, one of the major scientific challenges concerning carbon labeling lies in the allocation issues that are associated with the determination of these coefficients. The main features of these allocation issues are presented and analyzed in the next section **3.2**.

#### **3.2 The allocation issues of carbon labeling**

#### **3.2.1 The allocation issue within a company**

The scope of decision-making for a company lies within its own operations such as the transformation of purchased products, the business travels, the storing of raw materials, work-in-progress and finished products, the lightning and daily management of buildings and so on. The first challenge to ensure good management of GHG emissions at the company level is thus to allocate GHG emissions to each operation. This is not easy as different levels of complexity are intertwined within companies:


Regarding physical complexity, a first option would be to provide a continuous measurement of the emissions on every workstation. This would mean that each work station provide measurement of GHG emissions while delivering its work. Such framework is thus heavily based on electronic devices. The fact that many Annex I countries9 have started implementation projects of smart meters as home energy monitors exemplifies how measurement device constitutes the understanding basis of the energy use at consumers' home. Notably, the United Kingdom counts on smart meters to address the greening of everyday energy-consuming behaviors through the completion of the smart meter roll-out in 2019 (Beh11). Nevertheless, smart meters are electronic devices and therefore consume non-renewable resources such as minerals and imply high-energy intensive processes in their making. Consequently, a trade-off between the environmental benefits that could be gained from the use of smart meters with the environmental costs associated with their life cycle must be found. A good way of avoiding the use of measurement systems is to base the determination of GHG emissions on explanatory variables through the use of physical models. One of the main advantages of using physical models is to adopt better manageable variables. For example, knowing that 1 toe corresponds to 1616 kg of coal, a company can at the same time track down their coal purchase in tons and calculate the energy they will use when burning it. Therefore, through calculation, physical models allow to multiply the benefits associated with a particular measurement. Nevertheless, the emissions associated with physical models are often determined in standardized conditions that may reveal quite far from reality. For instance, combustion process is sensitive to the environment

<sup>9</sup> Italy, Japan, Canada, United Kingdom, United States, Australia, New Zealand, Netherlands, Spain, France, Ireland, Italy, Malta, Sweden, Finland, Norway (Sma11)

However, the establishment of these coefficients is rarely straightforward. Consequently, one of the major scientific challenges concerning carbon labeling lies in the allocation issues that are associated with the determination of these coefficients. The main features of these

The scope of decision-making for a company lies within its own operations such as the transformation of purchased products, the business travels, the storing of raw materials, work-in-progress and finished products, the lightning and daily management of buildings and so on. The first challenge to ensure good management of GHG emissions at the company level is thus to allocate GHG emissions to each operation. This is not easy as



Regarding physical complexity, a first option would be to provide a continuous measurement of the emissions on every workstation. This would mean that each work station provide measurement of GHG emissions while delivering its work. Such framework is thus heavily based on electronic devices. The fact that many Annex I countries9 have started implementation projects of smart meters as home energy monitors exemplifies how measurement device constitutes the understanding basis of the energy use at consumers' home. Notably, the United Kingdom counts on smart meters to address the greening of everyday energy-consuming behaviors through the completion of the smart meter roll-out in 2019 (Beh11). Nevertheless, smart meters are electronic devices and therefore consume non-renewable resources such as minerals and imply high-energy intensive processes in their making. Consequently, a trade-off between the environmental benefits that could be gained from the use of smart meters with the environmental costs associated with their life cycle must be found. A good way of avoiding the use of measurement systems is to base the determination of GHG emissions on explanatory variables through the use of physical models. One of the main advantages of using physical models is to adopt better manageable variables. For example, knowing that 1 toe corresponds to 1616 kg of coal, a company can at the same time track down their coal purchase in tons and calculate the energy they will use when burning it. Therefore, through calculation, physical models allow to multiply the benefits associated with a particular measurement. Nevertheless, the emissions associated with physical models are often determined in standardized conditions that may reveal quite far from reality. For instance, combustion process is sensitive to the environment

9 Italy, Japan, Canada, United Kingdom, United States, Australia, New Zealand, Netherlands, Spain,

France, Ireland, Italy, Malta, Sweden, Finland, Norway (Sma11)

��� ��� ����� = ∑ CO����

�

��� (2)

����� = ∑ CO���� �

allocation issues are presented and analyzed in the next section **3.2**.

different levels of complexity are intertwined within companies:

**3.2 The allocation issues of carbon labeling 3.2.1 The allocation issue within a company** 

complexity.

indicators.

temperature pressure and air/gasoline ratio and consequently would need to be associated with real-time measurements. Therefore, an uncertainty is always attached to the use of a physical model so that two companies using the same physical model and reporting two different GHG emission factor within the uncertainty range should be considered as having the same emission factor. Furthermore, direct measurement of GHG emissions should be preferred in the case of highly variable physical phenomena for which physical model are difficult to implement.

The management complexity consists in the organization of industry activities, the daily management of tasks and business travels together with the choices of processes. These decisions influence a lot the GHG efficiency of a company. Therefore, the allocation issue consists here in associating emissions with decision-making processes and is dependent on the time-framework chosen for the analysis. Electricity gives a good illustration of this point. Indeed, if we consider real-time measurement, assuming that CO2� � is the per kWh emission factor determined for the power plant k based on primary energy source j, the carbon content of 1kWh delivered to the network is:

$$\text{COZ}\_{\text{network}}(\mathbf{t}) = \frac{\Sigma\_{\text{Network}}\Big(kWh^{\dagger}\_{\mathbf{k}}(\mathbf{t}) \times \text{COZ}^{\dagger}\_{\mathbf{k}}(\mathbf{t})\Big)}{\Sigma\_{\text{Network}}\, kWh^{\dagger}\_{\mathbf{k}}(\mathbf{t})} \tag{3}$$

Getting the level of carbon emission at the yearly level would imply to sum the previous framework over time. In this ideal case, we thus obtained a comprehensive framework of emission per energy source with a timeline result. However, in present days, these kind of accurate data are not easily available. The electricity productions for each power plant are often given at the yearly level. Furthermore, there is often only a simple distinction of emission factors based on energy sources. The carbon content of 1 kWh delivered by the network is thus constant over the year:

$$\text{CO2}^{\text{year}}\_{\text{network}} = \frac{\Sigma\_{\text{Network}} \Big(k \mathcal{W} h^{l,year}\_k \times \text{CO2}^{l,year}\Big)}{\Sigma\_{\text{Network}} \, k \mathcal{W} h^l\_{k,year}} \tag{4}$$

The difference between (4) and (3) exemplifies how important the time-management framework chosen can influence a GHG analysis. Furthermore, one cannot differentiate electrons on the electric network and it is thus physically not possible to source the origin of electricity used on the network to a specific power plant. Therefore, no one on the network use cleaner energy than the other at a given time. The ways to get cleaner electricity delivered at one particular moment are thus:


The example of electricity demonstrates the difficulty of an accurate calculation and allocation of GHG emissions. First in terms of physical complexity, there is a need for standardization of the assumptions used in physical models. The lack of standardization and the difference of scope between studies are indeed the main features of green washing. By green washing, we define any communication using abusively ecological argumentation

The Greenhouse Stakes of Globalization 501

difficult to achieve as no commercial contract is linked to the moving of empty wagons. Furthermore, the decision regarding the management of empty wagons concerns as well the initial shipper, the owner of the wagon and the carrier. The latter must not only take into account the needs of the initial shipper but those of the other shippers to whom the empty

Such complex patterns would not be a significant problem regarding allocation of GHG emissions if the level of empty trip did not have such an important effect on the GHG emission level. Indeed, according to a Mc Kinnon study (ECTA, Responsible Care and Cefic, 2011), the gCO2/tonne-km emission factor for a 40-44 tonne trucks running 50% of its km empty is between 61% and 87% higher than when there is no vehicle-km run empty. Therefore, in the context of a carbon pricing, such differences will directly result in economic opportunity or losses. The same types of problems are encountered in the international maritime transportation area where the interdependencies of actors are even more complex. Indeed, the diversity of countries involved and notably the possibility to endorse a flag of convenience makes the determination of responsibility utterly complex. The term flag of convenience describes the business practice of registering a merchant ship in a sovereign state different from that of the ship's owners. The main interest for doing so is to reduce operating costs or avoid the regulations, notably taxes and environmental regulations, of the owner's country. Nevertheless, this situation makes it difficult to identify the ship owner and enforce a fair environmental post-Kyoto policy based on the allocation of emissions to

The present chapter has demonstrated the importance of globalization in regards to climate change and the necessity to include the trade perspective to enforce effective post-Kyoto policies. Such macro-policy framework is highly based on a better description of the GHG emissions associated with national and international freight transportation. Furthermore, the chapter revealed the difficulties of allocation associated with current policies notably in multi-actors sectors such as those of transport and logistics. **Figure 6** represents the hierarchy between the different allocation schemes explained in this chapter through an improvement loop. Doing so allows a clearer understanding of the distinction between scientific and policy driven procedures of allocation while maintaining the link between




The non-exhaustive list above offers a glimpse of the path towards more sustainable

them. This framework shows where the opportunity for new researches stand:


grounds for supply chains and society.

environmental and market efficient allocation rules.

wagons will be sent to be loaded.

the ship owner's state.

**4. Conclusion** 

processes.

Service System.

(Afnor Normalisation). The abusive character of green washing can be based on insufficient scientific methods in data collection, modeling, allocation methods, or the use of sounded GHG inventories outside their uncertainty ranges. All these issues can lead to the communication of wrong or not meaningful GHG emission figures. Furthermore, in addition to the previous difficulties in the determination of GHG emissions, the absence of transparency on the method used to calculate GHG emission is often the main obstacle to decide whether or not there is a risk of green washing. Therefore, the reproducibility of the calculation methodology is essential to provide a clear allocation of GHG emissions related to physically complex phenomena. Once the physical characteristics of GHG emissions allocation are solved, it becomes important to understand how to use this information to take decisions that enhance the carbon efficiency of the company and the supply chain within it exists. As said before, such decisions refer to management complexity, which is dependent on the number of actors involved in the decision-making process. Freight transportation is characterized by an important number of actors thus raising interesting allocation issues related to management complexity. Section **3.2.2** presents some of these aspects.

#### **3.2.2 The allocation issues related to transportation**

Transportation is a service allowing the physical connection between two companies through the moving of goods. It is thus characterized by a complex set of actors: 1st party logistic providers (shipper, consignee), 2nd party logistic providers (carriers), 3rd party logistic providers (freight forwarders), and 4th party logistics providers (consulting and IT activities). These actors may have different interests and may not search to optimize the same aspects of the transport chain. A good example is given in the railway sector.


Therefore, optimization trade-offs need to be found between these different perspectives. This would mean that all actors implied in the transportation process are willing to optimize the all process rather than the small part they are responsible for. This holds particularly true for the problem of empty trips, which correspond to the transport of empty wagons. This problem exists for the other transport mode such as road transportation where the term empty truck is used. Who should thus be responsible for the GHG emissions associated with empty trips?

One could take the position of the EcoTransIT world tool intended to be used by shippers that want to estimate the emissions associated with a particular transport activity or a set of different transport options: the shipper or freight owner takes responsibility for the vessel utilization factor that is averaged over the entire journey, from the starting point to the destination as well as the return trip or the entire loop respectively (Eco). This allocation process is coherent with an LCA perspective as it allows for the tracking of emission associated with a particular good. However, accuracy on the routing of empty wagons is

(Afnor Normalisation). The abusive character of green washing can be based on insufficient scientific methods in data collection, modeling, allocation methods, or the use of sounded GHG inventories outside their uncertainty ranges. All these issues can lead to the communication of wrong or not meaningful GHG emission figures. Furthermore, in addition to the previous difficulties in the determination of GHG emissions, the absence of transparency on the method used to calculate GHG emission is often the main obstacle to decide whether or not there is a risk of green washing. Therefore, the reproducibility of the calculation methodology is essential to provide a clear allocation of GHG emissions related to physically complex phenomena. Once the physical characteristics of GHG emissions allocation are solved, it becomes important to understand how to use this information to take decisions that enhance the carbon efficiency of the company and the supply chain within it exists. As said before, such decisions refer to management complexity, which is dependent on the number of actors involved in the decision-making process. Freight transportation is characterized by an important number of actors thus raising interesting allocation issues related to management complexity. Section **3.2.2** presents some of these

Transportation is a service allowing the physical connection between two companies through the moving of goods. It is thus characterized by a complex set of actors: 1st party logistic providers (shipper, consignee), 2nd party logistic providers (carriers), 3rd party logistic providers (freight forwarders), and 4th party logistics providers (consulting and IT activities). These actors may have different interests and may not search to optimize the



Therefore, optimization trade-offs need to be found between these different perspectives. This would mean that all actors implied in the transportation process are willing to optimize the all process rather than the small part they are responsible for. This holds particularly true for the problem of empty trips, which correspond to the transport of empty wagons. This problem exists for the other transport mode such as road transportation where the term empty truck is used. Who should thus be responsible for the GHG emissions associated with

One could take the position of the EcoTransIT world tool intended to be used by shippers that want to estimate the emissions associated with a particular transport activity or a set of different transport options: the shipper or freight owner takes responsibility for the vessel utilization factor that is averaged over the entire journey, from the starting point to the destination as well as the return trip or the entire loop respectively (Eco). This allocation process is coherent with an LCA perspective as it allows for the tracking of emission associated with a particular good. However, accuracy on the routing of empty wagons is

same aspects of the transport chain. A good example is given in the railway sector.


aspects.

reliability, etc.

empty trips?

**3.2.2 The allocation issues related to transportation** 

shippers to minimize costs such as fuel expenses.

difficult to achieve as no commercial contract is linked to the moving of empty wagons. Furthermore, the decision regarding the management of empty wagons concerns as well the initial shipper, the owner of the wagon and the carrier. The latter must not only take into account the needs of the initial shipper but those of the other shippers to whom the empty wagons will be sent to be loaded.

Such complex patterns would not be a significant problem regarding allocation of GHG emissions if the level of empty trip did not have such an important effect on the GHG emission level. Indeed, according to a Mc Kinnon study (ECTA, Responsible Care and Cefic, 2011), the gCO2/tonne-km emission factor for a 40-44 tonne trucks running 50% of its km empty is between 61% and 87% higher than when there is no vehicle-km run empty. Therefore, in the context of a carbon pricing, such differences will directly result in economic opportunity or losses. The same types of problems are encountered in the international maritime transportation area where the interdependencies of actors are even more complex. Indeed, the diversity of countries involved and notably the possibility to endorse a flag of convenience makes the determination of responsibility utterly complex. The term flag of convenience describes the business practice of registering a merchant ship in a sovereign state different from that of the ship's owners. The main interest for doing so is to reduce operating costs or avoid the regulations, notably taxes and environmental regulations, of the owner's country. Nevertheless, this situation makes it difficult to identify the ship owner and enforce a fair environmental post-Kyoto policy based on the allocation of emissions to the ship owner's state.

## **4. Conclusion**

The present chapter has demonstrated the importance of globalization in regards to climate change and the necessity to include the trade perspective to enforce effective post-Kyoto policies. Such macro-policy framework is highly based on a better description of the GHG emissions associated with national and international freight transportation. Furthermore, the chapter revealed the difficulties of allocation associated with current policies notably in multi-actors sectors such as those of transport and logistics. **Figure 6** represents the hierarchy between the different allocation schemes explained in this chapter through an improvement loop. Doing so allows a clearer understanding of the distinction between scientific and policy driven procedures of allocation while maintaining the link between them. This framework shows where the opportunity for new researches stand:


The non-exhaustive list above offers a glimpse of the path towards more sustainable grounds for supply chains and society.

The Greenhouse Stakes of Globalization 503

Hawkins Troy R. et Dente Sébastien M.R. Greenhouse Gas Emissions Driven by the

Hertwich Edgar et Peters Glen Production factors and pollution embodied in trade -

Hertwich Edgar G. Life cycle approaches to Sustainable Consumption: A critical Review [Article] // Environmental Science & Technology. - 2005. - 13 : Vol. 39. IEA CO2 emissions from fuel combustion - Highlights [Rapport]. - [s.l.] : IEA, 2010.

Kratena Kurt et Meyer Ina CO2 Emissions Embodied in Austrian International Trade

Lenglart Fabrice, Lesieur Christophe et Pasquier Jean-Louis Les émissions de CO2 du circuit

Lenzen Manfred [et al.] Shared producer and consumer responsibility - Theory and practice

Llerena Daniel et Michaud Céline Révélation des consentements à payer pour des produits

Daniel et Brissaud Daniel. - [s.l.] : Lavoisier, 2007. - ISBN 978-2-7462-1819-2. Ministère de l'Ecologie, du Développement durable, des transports et du logement Les 34

Noussair C., Robin S. et Ruffieux B. Do consumers really refuse to buy genetically modified

Peters Glen P. et Hertwich Edgar G. Post Kyoto greenhouse gas inventories: production versus consumption [Article] // Climatic Change. - 2008. - 86: 51-66. Peters Glen P. et Hertwich Edgar G. Trading Kyoto [Article] // nature reports climate

Regulation (EC) N°66/2010 of the European Parliamant and of the council of 25th

Sustainable Consumption institute of the University of Manchester Consumer, business and

Ton of oil equivalent [En ligne] // National Institute of Statistic and Economic Studies. - 24

http://www.insee.fr/en/methodes/default.asp?page=definitions/tonne-

United Nations Parties to the Convention and Observer States [En ligne] // UNFCCC. - 19 08 2011. - http://unfccc.int/parties\_and\_observers/parties/items/2352.php.

United Nations [En ligne]. - 1 08 2011. - http://www.un-documents.net/ofc-02.htm#I. United Nations Member States of the United Nations [En ligne] // United Nations. - 19 08

remanufacturés [Section du livre] // Les systèmes de production applications interdisciplinaires et mutations / auteur du livre Boujut jean-François, Llerena

chantiers opérationnels (COMOP) [En ligne] // le Grenelle Environnement. - 05 08

http://www.legrenelle-environnement.fr/IMG/pdf/rapport\_final\_comop\_23.pdf.

November 2009 on the EU Ecolabel [En ligne] // EUR-Lex Access to European Union Law. - 04 08 2011. - http://eur-lex/europa.eu/LexUriServ/LexUriserv.do?

économique en France [Rapport]. - Paris : [s.n.], 2010.

food? [Article] // The Economic Journal. - 2004. - 114.

[Article] // Ecological Economics. - 2007. - 61. - pp. 27-42.

Environ. Sci. Technol.. - 2010.

[Rapport]. - [s.l.] : FIW, 2010.

økologi, 2004.

2011. –

change. - 2008. - Vol. 2.

climate change [Rapport].

equivalent-petrole.htm.

08 2011. –

uri=OJ:L:2010:027:0001:0019:EN:PDF. Smart meter [En ligne] // Wikipédia. - 24 08 2011. –

http://en.wikipedia.org/wiki/Smart\_meter.

2011. - http://www.un.org/en/members/#n.

Transportation of Goods Associated with French Consumption [Article] //

Theoretical Development [Rapport]. - Trondheim : NTNU - Program for industriell

Fig. 6. The hierarchy between the different allocation steps and the on-going improvement process

### **5. References**

[En ligne]. - http://www.feem-project.net/exiopol/.


http://affichage-environnemental.afnor.org/mission.

Behaviour change and energy use [En ligne] // Department of Energy and Climate Change. - 24 08 2011. -

> http://www.decc.gov.uk/en/content/cms/emissions/beh\_change/beh\_change.a spx.


http://www.ecotransit.org/download/ecotransit\_background\_report.pdf.

ECTA, Responsible Care and Cefic Guidelines for measuring and managing CO2 emission from freight transport operaitons [Rapport]. - 2011.

Fig. 6. The hierarchy between the different allocation steps and the on-going improvement

Ackerman Franck Carbon Embedded in China's Trade [Rapport]. - Somerville - US :

ADEME Réalisation d'une veille internationale relative aux outils et méthodes d'évaluation

Afnor Normalisation Guides pratique des allégations environnementales à l'usage des

Afnor-Normalisation Mission [En ligne] // Affichage environnemental des produits de

Cetinkaya Balkan [et al.] Sustainable Supply Chain Management [Livre]. - Heidelberg Dordrecht London New York : Springer, 2011. - ISBN. data trade [En ligne]. Davis Steven J. et Caldeira Ken Consumption-based accounting of CO2- emissions

Digithèque MJP Loi constitutionnelle n°2005-205 du 1er mars 2005 relative à la charte de

EcoTransIT world methodology and data [En ligne] // Ecological Transport Information

http://www.ecotransit.org/download/ecotransit\_background\_report.pdf. ECTA, Responsible Care and Cefic Guidelines for measuring and managing CO2 emission

globale des émissions de gaz à effet de serre et des démarches de réduction

professionnels et des consommateurs [En ligne] // Affichage environnemental des produits de grande consommation - Actualités. - 04 08 2011. - http://affichage-

http://www.decc.gov.uk/en/content/cms/emissions/beh\_change/beh\_change.a

l'environnement [En ligne] // Constitution de la Vème République. - 23 08 2011. -

process

**5. References** 

[En ligne]. - http://www.feem-project.net/exiopol/. [En ligne]. - http://www.senat.fr/rap/l08-552-1117.html.

associées [Rapport]. - 2008.

Change. - 24 08 2011. -

spx.

Stockholm Environment Institute, 2009.

environnemental.afnor.org/actualités.

grande consommation. - 05 08 2011. –

[Article] // PNAS. - 2010. - 12 : Vol. 107.

Tool for Worldwide Transport. –

http://mjp.univ-perp.fr/france/co1958-19.htm.

from freight transport operaitons [Rapport]. - 2011.

http://affichage-environnemental.afnor.org/mission.

Behaviour change and energy use [En ligne] // Department of Energy and Climate


http://www.legrenelle-environnement.fr/IMG/pdf/rapport\_final\_comop\_23.pdf.


http://en.wikipedia.org/wiki/Smart\_meter.


http://www.insee.fr/en/methodes/default.asp?page=definitions/tonneequivalent-petrole.htm.


United Nations Protocole de Kyoto à la convention cadre des nations Unies sur les changements climatiques [En ligne]. - 1 08 2011. –

http://unfccc.int/resource/docs/convkp/kpfrench.pdf.


United Nations Protocole de Kyoto à la convention cadre des nations Unies sur les

United Nations Status of Ratification of the Kyoto Protocol [En ligne] // UNFCCC. - 19 08 2011. - http://unfccc.int/kyoto\_protocol/status\_of\_ratification/items/2613.php. United Nations United Nations Framework Convention on Climate Change [Rapport]. -

United Nations, UNFCCC Résumé analytique de la communication nationale de

Weber C. et Matthews H.S. Embodied Environmental Emissions in U.S. International Trade,

Weber C. L. et Matthews H. S. Food-miles and the relative environmental impact of food choices in the United States [Article] // Environ. Sci. Technol. - 2008. Wiedmann Dr. Thomas [et al.] Sustainable Consumption and Production - Development of an Evidence Base [Rapport]. - York : Stockholm Environment Institute, 2006.

1997 – 2004 [Article] // Environmental Science and Technology. - 2007. - 4875 –

changements climatiques [En ligne]. - 1 08 2011. – http://unfccc.int/resource/docs/convkp/kpfrench.pdf.

1992.

4881 : Vol. 41.

l'Allemagne [Rapport]. - 1994.

## *Edited by Guoxiang Liu*

Understanding greenhouse gas sources, emissions, measurements, and management is essential for capture, utilization, reduction, and storage of greenhouse gas, which plays a crucial role in issues such as global warming and climate change. Taking advantage of the authors' experience in greenhouse gases, this book discusses an overview of recently developed techniques, methods, and strategies: - A comprehensive source investigation of greenhouse gases that are emitted from hydrocarbon reservoirs, vehicle transportation, agricultural landscapes, farms, non-cattle confined buildings, and so on. - Recently developed detection and measurement techniques and methods such as photoacoustic spectroscopy, landfill-based carbon dioxide and methane measurement, and miniaturized mass spectrometer.

Photo by DutchScenery / iStock

Greenhouse Gases - Emission, Measurement and Management

Greenhouse Gases

Emission, Measurement and Management

*Edited by Guoxiang Liu*