**5. Lidar observations**

## **5.1. MSP-Lidar**

288 Atmospheric Aerosols – Regional Characteristics – Chemistry and Physics

polluted conditions [57].

**4.3. Coastal regions** 

Although local sources are by far the most important contributors to particulate air pollution in São Paulo city, back-trajectory analysis has shown that the atmosphere of the city can also be affected by the advection of air masses from distant regions where agricultural biomass burning is practiced, especially northeast Brazil [62]. This could explain the finding that the relative contribution of ammonium sulphate is higher under less

An important consequence of the prevalence of fine mode particles in the atmosphere of the city is that the indoor environment provides little or no protection against exposure to these pollutants, since they easily infiltrate buildings. This was observed [63] using simultaneous indoor and outdoor measurements of a range of ionic species associated with both primary emissions (potassium, magnesium, sodium and calcium) and secondary aerosol formation (chloride, acetate, nitrate, formate, pyruvate, nitrite, sulphate, oxalate and ammonium). The measurements were made in offices, restaurants and a hotel. In the fine mode, only oxalate and ammonium showed significantly lower concentrations indoors. In the coarse mode, lower concentrations were normally found indoors (with the exception of acetate, chloride

Polycyclic aromatic hydrocarbons are an important component of the urban aerosols. Chrysene, benzo(e)pyrene and benzo(b)fluoranthene were found to be the predominant PAHs in PM10, originating from industry, vehicles and long-range transport [64]. Levels of PM10 similar to those in São Paulo were measured in a city (Araraquara) situated in the rural biomass burning zone, although here PAH concentrations were lower. In both cases, dry deposition appeared to be the main mechanism of removal of PAH-containing aerosols from the atmosphere [65].

Bourotte *et al.* [66] measured the concentrations of 13 PAHs in fine (PM2.5) and coarse (PM2.5-10) aerosols. In both fractions, the predominant compounds were indeno(1,2,3-cd)pyrene, benzo(ghi)perylene and benzo(b)fluoranthene and PAH ratios suggested that automobile exhaust was the main source of the compounds. Factor analysis revealed four source components for the PM2.5 fraction: diesel emissions, stationary combustion, vehicle emissions, and combustion of natural gas and biomass. For the coarse fraction, two components were

Although measurements of atmospheric aerosol are scarce in most of the coastal regions, an exception is the industrialized town of Cubatão, located near sea level at the base of the Serra do Mar scarp, where there is a large industrial complex comprising over 20 heavy industries (petrochemical, chemical, iron and steel, fertilizer, cement, coking and others). The monitoring stations in this area register regular episodes of particulate pollution, with the emissions from the industrial installations being entrained into a sea breeze circulation, when PM10 concentrations can increase by as much as an order of magnitude [67]. Pollutants absorbed into cloud water and precipitation are subsequently deposited to the vegetation of

identified, corresponding to vehicles and a mixture of gas, oil, and waste combustion.

the Serra do Mar Atlantic rainforest, causing extensive ecological damage [68].

and potassium), reflecting the less efficient infiltration of larger aerosols.

In 2001, an elastic backscattering lidar system (MSP-Lidar) was installed in a suburban area of São Paulo city, on the Campus of the University of São Paulo (23°33' S, 46°44' W; Figure 6) and is being operated by the *Centro de Lasers e Aplicações* (CLA) of the *Instituto de Pesquisas Energéticas e Nucleares* (IPEN). The lidar is collocated with an AERONET sunphotometer, which provides the vertical profile of the aerosol backscatter coefficient at 532 nm up to an altitude of 4–6 km above sea level [72]. The MSP-Lidar comprises a Nd:YAG laser with a wavelength of 532 nm, and is operated with a repetition rate (PRF) of 20 Hz and an energy pulse of up to 120 mJ. The backscattering signal is captured by a Newtonian telescope with 1,5 m focal length. Attached to the telescope is a photomultiplier optimized for the visible spectrum with a 1 nm FWHM interference filter. Observations are being made whenever atmospheric conditions (absence of low or middle clouds; no rain) permit the operation of the lidar, resulting in a vast amount of data having been accumulated, which have so far been exploited in 5 MSc and 4 PhD theses, the most relevant being [73-76].

In January 2004, the IPEN MSP-Lidar system was installed for 6 weeks at IPMet in Bauru (Figure 6), located in the central part of São Paulo State, to provide the first measurements of

aerosol layers in the interior of the State [77]. At the beginning of the campaign, the lidar was operated in its original configuration and the data were digitized using a digital oscilloscope with 1 GHz bandwidth and 11-bit resolution; at the end of January 2004, this device was replaced by a transient recorder, capable of simultaneous analog and photon counting measurements at higher resolution (12-bit). The system was operated on 31 different days, during periods of about 4 hours in the morning, 4 hours in the afternoon and 6-8 hours during the night, depending on the occurrence of cloud and/or precipitation. The daytime measurements had a 15-30 m spatial resolution and maximum altitude of 10 km, yielding information on the diurnal variation of the Planetary Boundary Layer (PBL), while the measurements at night had a 30-60 m resolution, reaching up to 30-35 km maximum altitude. The diurnal variation of the PBL during the austral summer could be documented, as well as some background concentrations of aerosols, because very little biomass burning takes place during the rainy period.

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 291

**Figure 11.** Nocturnal lidar observation above Bauru on 04 March 2004. Vertical range from 855 m to

In early 2008, the MSP-Lidar system was upgraded to a Raman lidar, and in its present 3-channel configuration it can measure elastic backscatter at 355 nm, together with nitrogen and water vapour Raman backscatters at 387 nm and 408 nm, respectively. Therefore, the PBL data now available include aerosol backscattering and extinction coefficients, as well as the Lidar Ratio (LR) and water vapour mixing ratio. Figures 12 and 13 present typical results of Raman lidar measurements recorded during night-time of 09/10 January 2008, during the austral summer season. This period of the year is characterized by a very well defined boundary layer throughout the day and relatively high humidity. The major part of aerosols and water vapour is contained within the boundary layer, while the scattering above the PBL is mainly due to molecules. Figure 12 shows the aerosol extinction and backscattering coefficient profiles at 355 nm, where one can see a residual aerosol layer between 900 m and 2000 m AGL, indicating a very pronounced presence of aerosols, overlaid by another discrete layer above it between 2500 m and 3500 m AGL. The height profile of the Lidar Ratio is shown in Figure 13a. The Lidar Ratio is about 80 sr and stable throughout the PBL up to about 3000 m AGL. The vertical profile of the lidar-derived water vapour mixing ratio can be seen in Figure 13b. The calibration of the lidar was performed using radiosonde data from the nearby São Paulo Campo de Marte airport. Although the sonde had a relatively low height resolution, integrating the water vapour content with height made such

21,5 km AGL, with a resolution of 30 m (after [78]).

calibration possible.

Figure 10 shows a typical example of the diurnal variation of the height of the PBL on a cloudless day in Bauru. It should be noted that, due to the latitude of -22,3, the lidar cannot be operated during the midday period in summer, but as a result of turbulent mixing, the PBL could easily reach a maximum height of ≥3,5 km above ground level (AGL) during the early afternoon. The top of the PBL starts decreasing well before sunset, until it stabilizes at around 1,5 km AGL during the night. Times are indicated in Local Time (LT = UT-3h).

**Figure 10.** Height of the PBL over Bauru on 01/02 March 2004 for four different periods (10:19-11:08; 18:05-18:55; 19:13-19:54; 00:27-02:02 LT), with a vertical resolution of 30 m (after [78]).

The increased vertical range of the lidar during nocturnal operation permitted the detection of thin clouds and layers of aerosols, as shown in Figure 11. A cloud layer is clearly visible at around 4,5 km, while aerosols were detected at 3 and 5 km AGL, respectively. The top of the PBL is at about 1850 m AGL, with the faint layering being indicated in shades of green and light-blue colours.

takes place during the rainy period.

light-blue colours.

**0**

**10:00**

**10:19**

**10:30**

**10:36**

**10:52**

**11:05**

**500**

**1000**

**1500**

**2000**

**PBL Height AGL (m)**

**2500**

**3000**

**3500**

aerosol layers in the interior of the State [77]. At the beginning of the campaign, the lidar was operated in its original configuration and the data were digitized using a digital oscilloscope with 1 GHz bandwidth and 11-bit resolution; at the end of January 2004, this device was replaced by a transient recorder, capable of simultaneous analog and photon counting measurements at higher resolution (12-bit). The system was operated on 31 different days, during periods of about 4 hours in the morning, 4 hours in the afternoon and 6-8 hours during the night, depending on the occurrence of cloud and/or precipitation. The daytime measurements had a 15-30 m spatial resolution and maximum altitude of 10 km, yielding information on the diurnal variation of the Planetary Boundary Layer (PBL), while the measurements at night had a 30-60 m resolution, reaching up to 30-35 km maximum altitude. The diurnal variation of the PBL during the austral summer could be documented, as well as some background concentrations of aerosols, because very little biomass burning

Figure 10 shows a typical example of the diurnal variation of the height of the PBL on a cloudless day in Bauru. It should be noted that, due to the latitude of -22,3, the lidar cannot be operated during the midday period in summer, but as a result of turbulent mixing, the PBL could easily reach a maximum height of ≥3,5 km above ground level (AGL) during the early afternoon. The top of the PBL starts decreasing well before sunset, until it stabilizes at around 1,5 km AGL during the night. Times are indicated in Local Time (LT = UT-3h).

**Figure 10.** Height of the PBL over Bauru on 01/02 March 2004 for four different periods (10:19-11:08;

The increased vertical range of the lidar during nocturnal operation permitted the detection of thin clouds and layers of aerosols, as shown in Figure 11. A cloud layer is clearly visible at around 4,5 km, while aerosols were detected at 3 and 5 km AGL, respectively. The top of the PBL is at about 1850 m AGL, with the faint layering being indicated in shades of green and

18:05-18:55; 19:13-19:54; 00:27-02:02 LT), with a vertical resolution of 30 m (after [78]).

**18:05**

**18:08**

**18:14**

**18:20**

**18:29**

**18:36**

**18:43**

**Local Time (hh:mm LT)**

**18:49**

**19:18**

**19:29**

**19:35**

**19:46**

**19:53**

**00:33**

**00:43**

**01:03**

**01:59**

**Figure 11.** Nocturnal lidar observation above Bauru on 04 March 2004. Vertical range from 855 m to 21,5 km AGL, with a resolution of 30 m (after [78]).

In early 2008, the MSP-Lidar system was upgraded to a Raman lidar, and in its present 3-channel configuration it can measure elastic backscatter at 355 nm, together with nitrogen and water vapour Raman backscatters at 387 nm and 408 nm, respectively. Therefore, the PBL data now available include aerosol backscattering and extinction coefficients, as well as the Lidar Ratio (LR) and water vapour mixing ratio. Figures 12 and 13 present typical results of Raman lidar measurements recorded during night-time of 09/10 January 2008, during the austral summer season. This period of the year is characterized by a very well defined boundary layer throughout the day and relatively high humidity. The major part of aerosols and water vapour is contained within the boundary layer, while the scattering above the PBL is mainly due to molecules. Figure 12 shows the aerosol extinction and backscattering coefficient profiles at 355 nm, where one can see a residual aerosol layer between 900 m and 2000 m AGL, indicating a very pronounced presence of aerosols, overlaid by another discrete layer above it between 2500 m and 3500 m AGL. The height profile of the Lidar Ratio is shown in Figure 13a. The Lidar Ratio is about 80 sr and stable throughout the PBL up to about 3000 m AGL. The vertical profile of the lidar-derived water vapour mixing ratio can be seen in Figure 13b. The calibration of the lidar was performed using radiosonde data from the nearby São Paulo Campo de Marte airport. Although the sonde had a relatively low height resolution, integrating the water vapour content with height made such calibration possible.

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 293

**Figure 13. (a)** shows the 355 nm Lidar Ratio profile on 10 January 2008 at 00:25 UT (21:25 LT). **(b)** shows the water vapour mixing ratio extracted on the same day from the 408 nm channel (00:25 UT)

(b)

(a)

and from a radiosonde ascent (00:00 UT).

**Figure 12.** Aerosol backscatter and extinction profiles at 355 nm observed on 10 January 2008 at 00:25 UT (21:25 LT). The PBL top height is considered to be at 2000 m.

The MSP-Lidar system has contributed to several studies concerning the properties of aerosols and their influence on the air quality index of the city of São Paulo. Lidar measurements conducted daily provided observations of the PBL variation, which could be compared to corresponding air quality index values from local air quality monitoring and management agencies, as well as identifying potential air dispersion conditions [79]. It has also been deployed to monitor the long-range transport of aerosol plumes from different regions of Brazil to the RMSP and to evaluate the contribution of aerosol pollutants from remote sources. Landulfo and Lopes [80] have analyzed an event during the period 02 - 09 August 2007 when the AOD (Aerosol Optical Depth) and AE (Ångström Exponent) values retrieved from the AERONET sunphotometer indicated that high aerosol loads at five different locations in the Brazilian territory corresponded to biomass-burning particles. This was validated by the mean values of the Total Attenuated Backscatter Coefficient at 532 nm, the mean depolarization ratio and also the Lidar Ratio (about 70 sr) for all sites over-flown by the CALIOP sensor onboard the CALIPSO satellite.

In another case study during the dry winter season of 2008, fire plumes attributed to sugar cane fires were frequently observed by IPMet's radars in the absence of rain echoes and documented in terms of radar reflectivity, time and location [12]. On several occasions, IPEN's Elastic Backscatter Lidar in São Paulo observed layers of aerosols of varying strength and heights above the city. The most significant days were selected for calculating

UT (21:25 LT). The PBL top height is considered to be at 2000 m.

by the CALIOP sensor onboard the CALIPSO satellite.

**Figure 12.** Aerosol backscatter and extinction profiles at 355 nm observed on 10 January 2008 at 00:25

The MSP-Lidar system has contributed to several studies concerning the properties of aerosols and their influence on the air quality index of the city of São Paulo. Lidar measurements conducted daily provided observations of the PBL variation, which could be compared to corresponding air quality index values from local air quality monitoring and management agencies, as well as identifying potential air dispersion conditions [79]. It has also been deployed to monitor the long-range transport of aerosol plumes from different regions of Brazil to the RMSP and to evaluate the contribution of aerosol pollutants from remote sources. Landulfo and Lopes [80] have analyzed an event during the period 02 - 09 August 2007 when the AOD (Aerosol Optical Depth) and AE (Ångström Exponent) values retrieved from the AERONET sunphotometer indicated that high aerosol loads at five different locations in the Brazilian territory corresponded to biomass-burning particles. This was validated by the mean values of the Total Attenuated Backscatter Coefficient at 532 nm, the mean depolarization ratio and also the Lidar Ratio (about 70 sr) for all sites over-flown

In another case study during the dry winter season of 2008, fire plumes attributed to sugar cane fires were frequently observed by IPMet's radars in the absence of rain echoes and documented in terms of radar reflectivity, time and location [12]. On several occasions, IPEN's Elastic Backscatter Lidar in São Paulo observed layers of aerosols of varying strength and heights above the city. The most significant days were selected for calculating

**Figure 13. (a)** shows the 355 nm Lidar Ratio profile on 10 January 2008 at 00:25 UT (21:25 LT). **(b)** shows the water vapour mixing ratio extracted on the same day from the 408 nm channel (00:25 UT) and from a radiosonde ascent (00:00 UT).

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 295

backward, as well as forward trajectories, deploying the Flextra 3.3 Trajectory Model [81], which was initiated with ECMWF historical data with a 0,25 x 0,25 grid spacing [12]. The results showed an excellent match between the radar-detected sources of the plumes and lidar observations in São Paulo. Figure 14 presents a typical case study, when emissions from biomass fires were identified by the radars on 11 July 2008 in the central parts of the State, and were subsequently monitored by IPEN's lidar over Metropolitan São Paulo on 14 July 2008, deploying forward and backward trajectories. The forward trajectories, initiated at different heights ranging from 930 hPa (close to ground level) up to 450 hPa (ca 6,7 km amsl) at 30 hPa intervals (only the most significant 10 heights are shown in Figure 14), indicated a transport duration of approximately 70 hours under the prevailing meteorological conditions (Figure 14b). The arrival of the plume over the RMSP on 14 July 2008, as observed by the lidar at IPEN,

**Figure 15.** Lidar observations at IPEN in São Paulo, showing the range-corrected signal in arbitrary units, on 14 July 2008 between 11:00 and 18:13 LT. The plume identified in Figure 14 can be seen

The MSP-Lidar system in São Paulo has also been contributing to CALIPSO satellite validation procedures [75, 82]. During 2007, correlative measurements were carried out with special attention to the dry season (May-October), when most of the days have poor dispersion conditions and long distance transport is more frequent. From a total of 28 days of measurements, on only 10 days were no clouds present below 4 km. Figure 16a presents a typical example, showing the range-corrected signal retrieved by the lidar system at São Paulo on 10 October 2007 between 03:34 and 05:35 UT, which contains the CALIPSO overpass window, beginning at 04:30 UT (Figure 16b). On this day, the closest distance of the satellite ground-track from the lidar site was about 48 km. The presence of aerosol layers above the PBL at 4-5 km, 6 km and 9 km is noticeable. The same features are also observed in the CALIOP 532 nm Total Attenuated Backscattering plot, as shown in Figure 16b. Both systems detected a cirrus structure between 12 and 13 km AGL, but the strong cirrus cloud signal observed in the CALIOP "plot-curtain" is much weaker in the lidar image. The red box in Figure 16b represents the CALIPSO ground-track region over Metropolitan São Paulo

with coordinates of -22,5625 latitude and -46,0247 longitude at about 04:35 UT.

is shown in Figure 15.

between 4-5 km AGL.

**Figure 14.** Forward trajectories initiated at different heights where a large fire was observed by IPMet's radars on 11 July 2008, 22:00 UT (19:00 LT). **(a)** The + indicates the position of the PPR and BRU radars; o indicates the position of the lidar in São Paulo (IPEN). **(b)** Forward trajectories plotted against height and time. The + indicates the position of IPEN, marking height and time of arrival matching exactly with the lidar observation (Figure 15).

backward, as well as forward trajectories, deploying the Flextra 3.3 Trajectory Model [81], which was initiated with ECMWF historical data with a 0,25 x 0,25 grid spacing [12]. The results showed an excellent match between the radar-detected sources of the plumes and lidar observations in São Paulo. Figure 14 presents a typical case study, when emissions from biomass fires were identified by the radars on 11 July 2008 in the central parts of the State, and were subsequently monitored by IPEN's lidar over Metropolitan São Paulo on 14 July 2008, deploying forward and backward trajectories. The forward trajectories, initiated at different heights ranging from 930 hPa (close to ground level) up to 450 hPa (ca 6,7 km amsl) at 30 hPa intervals (only the most significant 10 heights are shown in Figure 14), indicated a transport duration of approximately 70 hours under the prevailing meteorological conditions (Figure 14b). The arrival of the plume over the RMSP on 14 July 2008, as observed by the lidar at IPEN, is shown in Figure 15.

294 Atmospheric Aerosols – Regional Characteristics – Chemistry and Physics

**Figure 14.** Forward trajectories initiated at different heights where a large fire was observed by IPMet's radars on 11 July 2008, 22:00 UT (19:00 LT). **(a)** The + indicates the position of the PPR and BRU radars; o indicates the position of the lidar in São Paulo (IPEN). **(b)** Forward trajectories plotted against height and time. The + indicates the position of IPEN, marking height and time of arrival matching exactly

(b)

(a)

with the lidar observation (Figure 15).

**Figure 15.** Lidar observations at IPEN in São Paulo, showing the range-corrected signal in arbitrary units, on 14 July 2008 between 11:00 and 18:13 LT. The plume identified in Figure 14 can be seen between 4-5 km AGL.

The MSP-Lidar system in São Paulo has also been contributing to CALIPSO satellite validation procedures [75, 82]. During 2007, correlative measurements were carried out with special attention to the dry season (May-October), when most of the days have poor dispersion conditions and long distance transport is more frequent. From a total of 28 days of measurements, on only 10 days were no clouds present below 4 km. Figure 16a presents a typical example, showing the range-corrected signal retrieved by the lidar system at São Paulo on 10 October 2007 between 03:34 and 05:35 UT, which contains the CALIPSO overpass window, beginning at 04:30 UT (Figure 16b). On this day, the closest distance of the satellite ground-track from the lidar site was about 48 km. The presence of aerosol layers above the PBL at 4-5 km, 6 km and 9 km is noticeable. The same features are also observed in the CALIOP 532 nm Total Attenuated Backscattering plot, as shown in Figure 16b. Both systems detected a cirrus structure between 12 and 13 km AGL, but the strong cirrus cloud signal observed in the CALIOP "plot-curtain" is much weaker in the lidar image. The red box in Figure 16b represents the CALIPSO ground-track region over Metropolitan São Paulo with coordinates of -22,5625 latitude and -46,0247 longitude at about 04:35 UT.

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 297

**Figure 17.** Total Attenuated Backscatter Coefficient profiles at 532 nm for the horizontal coverage of CALIPSO level 1 data compared to the Attenuated Backscatter Coefficient retrieved by the ground-

The mobile bi-axial Raman lidar system uses a commercial pulsed Nd:YAG laser, operating at a wavelength of 532 nm in the elastic channel and 607 nm in the Nitrogen Raman channel, with a pulse energy of 130 mJ at 20 Hz PRF. The pulse width is 25 ns, yielding a spatial resolution of 7,5 m. A detailed description of the system is found in [83]. The system allows the determination of the optical properties of the atmosphere, including aerosol backscatter and extinction coefficients, as well as an indication of the type of aerosol present, based on the Lidar Ratio. This lidar has so far been deployed during specific campaigns at three different sites within the central region of São Paulo State, *viz*., Rio Claro [84], Bauru and Ourinhos [85-87], as well as in Cubatão, an industrial hub at the coast, near Santos [88], as

A one-month pilot study was undertaken during August 2010 in Ourinhos (Figure 6), which is situated in one of the State's major sugar cane producing regions, where biomass burning is a regular occurrence. The objective was to characterize the effects of these emissions on the atmosphere, considering the local circulation and the consequences for the region [85]. In the absence of rain, the plumes were tracked by IPMet´s two S-band Doppler radars within their quantitative ranges of 240 km (BRU = Bauru, PPR = Presidente Prudente; Figure 6), using the TITAN (*Thunderstorm Identification, Tracking, Analysis, and Nowcasting*) Radar Software [89]. A large range of meteorological, physical and chemical instrumentation, including the mobile Raman lidar, was used to observe elevated layers and the type of aerosols. A medium-sized

based MSP-Lidar system in Metropolitan São Paulo on 10 October 2007.

**5.2. Mobile Raman lidar** 

shown in Figure 6.

**Figure 16. (a)** Range-corrected lidar signal (plot-curtain) measured by the MSP-Lidar on 10 October 2007, 03:34 - 05:35 UT. **(b)** Total Attenuated Backscattering signal measured by the CALIOP at 532 nm during the period 04:30 - 04:41 UT on the same day, when it was closest to the MSP-Lidar site (red box).

Figure 17 compares the attenuated backscatter coefficient profile retrieved by CALIOP on board the CALIPSO satellite and the corrected one obtained from the ground-based MSP-Lidar system in São Paulo. The satellite profile has a 5 km horizontal resolution. The attenuated backscatter profile from the MSP-Lidar site was derived under cloud-free conditions from the range-corrected and background noise-subtracted lidar return signal. Both profiles are in good agreement, presenting similar layer patterns in the profiles observed at 5-6 km and about 7 km AGL. Since it can be assumed with reasonable confidence that, at higher altitudes, the horizontal atmospheric structure is more homogeneous, the good agreement between the two systems demonstrates the possibility that they were probing the same air masses for this specific measurement. At lower altitudes, observation of some differences between the two profiles is more likely due to local effects. In this case, the localized effects are more pronounced, and the fact that the systems are not covering the exact same region becomes evident.

**Figure 17.** Total Attenuated Backscatter Coefficient profiles at 532 nm for the horizontal coverage of CALIPSO level 1 data compared to the Attenuated Backscatter Coefficient retrieved by the groundbased MSP-Lidar system in Metropolitan São Paulo on 10 October 2007.

## **5.2. Mobile Raman lidar**

296 Atmospheric Aerosols – Regional Characteristics – Chemistry and Physics

**Figure 16. (a)** Range-corrected lidar signal (plot-curtain) measured by the MSP-Lidar on 10 October 2007, 03:34 - 05:35 UT. **(b)** Total Attenuated Backscattering signal measured by the CALIOP at 532 nm during the period 04:30 - 04:41 UT on the same day, when it was closest to the MSP-Lidar site (red box).

(b)

(a)

Figure 17 compares the attenuated backscatter coefficient profile retrieved by CALIOP on board the CALIPSO satellite and the corrected one obtained from the ground-based MSP-Lidar system in São Paulo. The satellite profile has a 5 km horizontal resolution. The attenuated backscatter profile from the MSP-Lidar site was derived under cloud-free conditions from the range-corrected and background noise-subtracted lidar return signal. Both profiles are in good agreement, presenting similar layer patterns in the profiles observed at 5-6 km and about 7 km AGL. Since it can be assumed with reasonable confidence that, at higher altitudes, the horizontal atmospheric structure is more homogeneous, the good agreement between the two systems demonstrates the possibility that they were probing the same air masses for this specific measurement. At lower altitudes, observation of some differences between the two profiles is more likely due to local effects. In this case, the localized effects are more pronounced, and the

fact that the systems are not covering the exact same region becomes evident.

The mobile bi-axial Raman lidar system uses a commercial pulsed Nd:YAG laser, operating at a wavelength of 532 nm in the elastic channel and 607 nm in the Nitrogen Raman channel, with a pulse energy of 130 mJ at 20 Hz PRF. The pulse width is 25 ns, yielding a spatial resolution of 7,5 m. A detailed description of the system is found in [83]. The system allows the determination of the optical properties of the atmosphere, including aerosol backscatter and extinction coefficients, as well as an indication of the type of aerosol present, based on the Lidar Ratio. This lidar has so far been deployed during specific campaigns at three different sites within the central region of São Paulo State, *viz*., Rio Claro [84], Bauru and Ourinhos [85-87], as well as in Cubatão, an industrial hub at the coast, near Santos [88], as shown in Figure 6.

A one-month pilot study was undertaken during August 2010 in Ourinhos (Figure 6), which is situated in one of the State's major sugar cane producing regions, where biomass burning is a regular occurrence. The objective was to characterize the effects of these emissions on the atmosphere, considering the local circulation and the consequences for the region [85]. In the absence of rain, the plumes were tracked by IPMet´s two S-band Doppler radars within their quantitative ranges of 240 km (BRU = Bauru, PPR = Presidente Prudente; Figure 6), using the TITAN (*Thunderstorm Identification, Tracking, Analysis, and Nowcasting*) Radar Software [89]. A large range of meteorological, physical and chemical instrumentation, including the mobile Raman lidar, was used to observe elevated layers and the type of aerosols. A medium-sized sodar, as well as 6 automatic weather stations, were also deployed in the region. Various gases and aerosol size fractions were sampled, providing an atmospheric chemistry database and thus documenting the impact of the harvesting practice on the region. The aerosol load of the atmosphere was quantified by hourly mean AOD values and hourly mean backscatter profiles. Several case studies have already been analyzed, but the one of 25-26 August 2010 will be shown in this Chapter to illustrate how the various remote sensing instruments are being deployed to generate a complete picture of events.

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 299

the pollutants (aerosols), trapped above the inversion, only commenced at around 09:00- 09:30 LT, since from 00:00-09:00 LT the atmosphere was extremely stable below 300 m AGL

**(a)** First TITAN centroid of the *queimada* (actual fire, blue) at 03:22 UT (00:22 LT; annotation:

**(b)** The *queimada* reached the Ourinhos region at 05:45 UT (02:45 LT, blue; annotation: maximum

**(c)** Vertical cross-sections at 03:45 UT (00:45 LT), showing the horizontal and vertical extent along the

The lidar observed the arrival of the plume at 02:40 LT between 350 and 600 m AGL (Figure 19a). The top of the PBL extended to about 2,6 km AGL, above which a very dry and relatively warm and clean air mass was advected from the west, creating an elevated inversion which blocked further upward mixing. The lowest layer ≤250 m AGL appeared clean, being trapped within the surface inversion, inhibiting downward mixing, also confirmed by the sodar measurements, indicating a very stable layer. Lidar data from the Raman Channel (non-elastic signal at 607 nm) were integrated into hourly means until 09:00 LT to obtain the AOD. The results confirmed a high aerosol load of the atmosphere, with hourly mean values of AOD varying between 0,265 and 0,288 until 07:00 LT, after which they increased to 0,433 by 09:00 LT. Hourly means of the Lidar Ratio confirmed the arrival of the plume between 02:00 and 03:00 LT (example shown in Figure 19b), while an almost 20% increase of LR to 72 sr after 07:00 LT was probably due to downward mixing of the aerosols accumulated above the inversion, also confirmed by an increase of AOD values from the Raman signal [85]. LR values of around 70 sr suggest aerosols originating from

**Figure 18.** Examples of the tracks generated by TITAN on 26 August 2010. The envelopes (10 dBZ reflectivity) show the position of the *queimada* (smoke plume) in intervals of 7,5 min (blue = actual time;

(w = 0 m.s-1).

propagation velocity in km.h-1);

reflectivity in dBZ).

base lines A-B and C-D.

green = future; yellow = past).

biomass burning [90, 91].

During the second half of August 2010, the weather was dominated by a high pressure system, resulting in a rise in temperatures, with low humidity favoring the accumulation of pollutants in the atmosphere of the region [25]. IPMet's radars have a 2° beam width and a quantitative range of 240 km, generating a volume-scan every 7,5 minutes, with a resolution of 250 m radially and 1° in azimuth. Reflectivities and radial velocities are recorded at 16 elevations. However, in order to detect and track the biomass burning plumes, a special scanning cycle was configured to provide a better vertical resolution up to the anticipated detectable top of the plumes: 10,0, 8,0, 6,5, 5,0, 4,0, 3,2, 2,4, 1,6, 0,8 and 0,3, with each "sweep" (Plan Position Indicator - PPI) having 360 rays with 957 range bins each. Two different software systems were deployed, *viz.,* IRIS (*Interactive Radar Information System*) Analysis was used first to generate CAPPIs (Constant Altitude PPIs) at 1,5 and 2,0 km amsl, in order to identify all smoke plumes within the 240 km range of the radars. Once a plume was identified as likely to pass over the monitoring site, it was tracked using TITAN Software to determine its intensity (based on radar reflectivity in dBZ), horizontal and vertical dimensions, and the velocity of approach. The thresholds used for tracking were 10 dBZ with a minimum volume of 2 km3. It should be noted that TITAN uses Universal Time (Local Time LT = UT-3h).

A typical case study of a sugar cane fire in the Ourinhos region is now presented, demonstrating the integration of all types of data into one coherent event. The first echo of a smoke plume was detected by the Bauru radar on 26 August 2010 at 00:08 LT, about 35 km north-northeast of Ourinhos and about 85 km southwest of the radar (Figure 18), rapidly gaining in area and intensity (≤40 dBZ near its origin). By 00:22 LT, the TITAN Software could already identify its centroid of 10 dBZ reflectivity and tracked it until 02:45 LT, when the plume had already spread over Ourinhos, where the Raman lidar and sodar were located. As the plume moved southwards with the northerly winds, the aerosols spread out (dispersed) and the reflectivity dropped gradually, but it could still be detected by the radar until 03:46 LT, >20 km south of Ourinhos, using a reflectivity threshold of –6 dBZ [85].

Furthermore, it can be deduced from Figure 18a that while the plume was at a low height during the initial phase of transport, it moved very slowly (3-4 km.h-1), since the wind speed in the first few hundred meters was very low (5 m.s-1), as observed by the sodar. There was also a shift of the wind direction from easterly to northerly winds above 300 m AGL. These northerly winds were above the nighttime surface inversion, confirmed by the "Skew T x Log P" profiles of the Meso-Eta model in the 900-800 hPa layer (650–1650 m AGL) as shown in [85]. The vertical velocity (w), measured by the sodar, indicated that downward mixing of the pollutants (aerosols), trapped above the inversion, only commenced at around 09:00- 09:30 LT, since from 00:00-09:00 LT the atmosphere was extremely stable below 300 m AGL (w = 0 m.s-1).

298 Atmospheric Aerosols – Regional Characteristics – Chemistry and Physics

deployed to generate a complete picture of events.

Time (Local Time LT = UT-3h).

sodar, as well as 6 automatic weather stations, were also deployed in the region. Various gases and aerosol size fractions were sampled, providing an atmospheric chemistry database and thus documenting the impact of the harvesting practice on the region. The aerosol load of the atmosphere was quantified by hourly mean AOD values and hourly mean backscatter profiles. Several case studies have already been analyzed, but the one of 25-26 August 2010 will be shown in this Chapter to illustrate how the various remote sensing instruments are being

During the second half of August 2010, the weather was dominated by a high pressure system, resulting in a rise in temperatures, with low humidity favoring the accumulation of pollutants in the atmosphere of the region [25]. IPMet's radars have a 2° beam width and a quantitative range of 240 km, generating a volume-scan every 7,5 minutes, with a resolution of 250 m radially and 1° in azimuth. Reflectivities and radial velocities are recorded at 16 elevations. However, in order to detect and track the biomass burning plumes, a special scanning cycle was configured to provide a better vertical resolution up to the anticipated detectable top of the plumes: 10,0, 8,0, 6,5, 5,0, 4,0, 3,2, 2,4, 1,6, 0,8 and 0,3, with each "sweep" (Plan Position Indicator - PPI) having 360 rays with 957 range bins each. Two different software systems were deployed, *viz.,* IRIS (*Interactive Radar Information System*) Analysis was used first to generate CAPPIs (Constant Altitude PPIs) at 1,5 and 2,0 km amsl, in order to identify all smoke plumes within the 240 km range of the radars. Once a plume was identified as likely to pass over the monitoring site, it was tracked using TITAN Software to determine its intensity (based on radar reflectivity in dBZ), horizontal and vertical dimensions, and the velocity of approach. The thresholds used for tracking were 10 dBZ with a minimum volume of 2 km3. It should be noted that TITAN uses Universal

A typical case study of a sugar cane fire in the Ourinhos region is now presented, demonstrating the integration of all types of data into one coherent event. The first echo of a smoke plume was detected by the Bauru radar on 26 August 2010 at 00:08 LT, about 35 km north-northeast of Ourinhos and about 85 km southwest of the radar (Figure 18), rapidly gaining in area and intensity (≤40 dBZ near its origin). By 00:22 LT, the TITAN Software could already identify its centroid of 10 dBZ reflectivity and tracked it until 02:45 LT, when the plume had already spread over Ourinhos, where the Raman lidar and sodar were located. As the plume moved southwards with the northerly winds, the aerosols spread out (dispersed) and the reflectivity dropped gradually, but it could still be detected by the radar until 03:46 LT, >20 km south of Ourinhos, using a reflectivity threshold of –6 dBZ [85].

Furthermore, it can be deduced from Figure 18a that while the plume was at a low height during the initial phase of transport, it moved very slowly (3-4 km.h-1), since the wind speed in the first few hundred meters was very low (5 m.s-1), as observed by the sodar. There was also a shift of the wind direction from easterly to northerly winds above 300 m AGL. These northerly winds were above the nighttime surface inversion, confirmed by the "Skew T x Log P" profiles of the Meso-Eta model in the 900-800 hPa layer (650–1650 m AGL) as shown in [85]. The vertical velocity (w), measured by the sodar, indicated that downward mixing of

**(a)** First TITAN centroid of the *queimada* (actual fire, blue) at 03:22 UT (00:22 LT; annotation: propagation velocity in km.h-1);

**(b)** The *queimada* reached the Ourinhos region at 05:45 UT (02:45 LT, blue; annotation: maximum reflectivity in dBZ).

**(c)** Vertical cross-sections at 03:45 UT (00:45 LT), showing the horizontal and vertical extent along the base lines A-B and C-D.

**Figure 18.** Examples of the tracks generated by TITAN on 26 August 2010. The envelopes (10 dBZ reflectivity) show the position of the *queimada* (smoke plume) in intervals of 7,5 min (blue = actual time; green = future; yellow = past).

The lidar observed the arrival of the plume at 02:40 LT between 350 and 600 m AGL (Figure 19a). The top of the PBL extended to about 2,6 km AGL, above which a very dry and relatively warm and clean air mass was advected from the west, creating an elevated inversion which blocked further upward mixing. The lowest layer ≤250 m AGL appeared clean, being trapped within the surface inversion, inhibiting downward mixing, also confirmed by the sodar measurements, indicating a very stable layer. Lidar data from the Raman Channel (non-elastic signal at 607 nm) were integrated into hourly means until 09:00 LT to obtain the AOD. The results confirmed a high aerosol load of the atmosphere, with hourly mean values of AOD varying between 0,265 and 0,288 until 07:00 LT, after which they increased to 0,433 by 09:00 LT. Hourly means of the Lidar Ratio confirmed the arrival of the plume between 02:00 and 03:00 LT (example shown in Figure 19b), while an almost 20% increase of LR to 72 sr after 07:00 LT was probably due to downward mixing of the aerosols accumulated above the inversion, also confirmed by an increase of AOD values from the Raman signal [85]. LR values of around 70 sr suggest aerosols originating from biomass burning [90, 91].

Visual images from overpasses of the MODIS-AQUA satellite on 25 and 26 August 2010 (at 17:35 and 16:40 UT, respectively; 14:35 and 13:40 LT) showed intense smoke plumes to the west and south of the Ourinhos region, with AOD values of up to about 1,0. In the Ourinhos region, the AOD increased during the period 25-26 August, from about 0,2 to about 0,6 (Figure 20a), which is in agreement with the early afternoon lidar measurements (Figure 20b), which provided an AOD value of 0,380 during the period from 13:00 to 14:00 LT.

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 301

**Figure 20. (a)** AOD image from MODIS-AQUA on 26 August 2010, 16:40-16:45 UT (13:40-13:45 LT). The

(b)

(a)

**(b)** Lidar measurements on 26 August 2010, 12:01-15:01 LT. The time of the MODIS-AQUA overpass is

inset shows a simultaneous visual image of the Ourinhos region.

indicated by the dashed white line.

**Figure 19. (a)** Lidar signal (arbitrary units) visualized for 00:00-06:00 LT, up to 3 km AGL. **(b)** Backscatter Profile at 532 nm for the hourly mean period 03:00-03:59 LT on 26 August 2010.

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 301

300 Atmospheric Aerosols – Regional Characteristics – Chemistry and Physics

14:00 LT.

Visual images from overpasses of the MODIS-AQUA satellite on 25 and 26 August 2010 (at 17:35 and 16:40 UT, respectively; 14:35 and 13:40 LT) showed intense smoke plumes to the west and south of the Ourinhos region, with AOD values of up to about 1,0. In the Ourinhos region, the AOD increased during the period 25-26 August, from about 0,2 to about 0,6 (Figure 20a), which is in agreement with the early afternoon lidar measurements (Figure 20b), which provided an AOD value of 0,380 during the period from 13:00 to

**Figure 19. (a)** Lidar signal (arbitrary units) visualized for 00:00-06:00 LT, up to 3 km AGL. **(b)** Backscatter Profile at 532 nm for the hourly mean period 03:00-03:59 LT on 26 August 2010.

(a)

(a)

**Figure 20. (a)** AOD image from MODIS-AQUA on 26 August 2010, 16:40-16:45 UT (13:40-13:45 LT). The inset shows a simultaneous visual image of the Ourinhos region.

(b)

**(b)** Lidar measurements on 26 August 2010, 12:01-15:01 LT. The time of the MODIS-AQUA overpass is indicated by the dashed white line.

Aerosols collected during daytime and nighttime periods at the lidar site [85-87, 92], using low-volume filter samplers, were chemically characterized by means of ion chromatography. A higher concentration of K+ during the period from 22:00 on 25 August to 16:00 on 26 August 2010 indicated the presence of biomass-burning material (Figure 21), since K+ is a plant macronutrient released during the combustion process. Levoglucosan, a very specific chemical marker of biomass combustion, was well above average concentration during day sampling on 26 August and even higher during the following night, indicating a strong presence of biomass smoke on both days.

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 303

Further evidence of the impact on the Ourinhos region of emissions from sugar cane fires was obtained by comparing the concentrations of organic compounds in aerosol particles collected on 26 August with those collected one day earlier. Ambient levels of polycyclic aromatic hydrocarbons (PAH), as well as PAH derivatives, such as oxy-PAH, were significantly higher on 26 August 2010 than on the previous day, confirming that emissions

An elastic backscatter lidar system, with similar characteristics to the mobile lidar, was installed in 2011 at CEPEMA-USP (*Centro de Pesquisas em Meio Ambiente,* a Center for Environmental Research and Training, under the responsibility of the Universidade de São Paulo) in the Cubatão industrial area, with the ultimate goal of remotely monitoring industrial emissions. It also uses a commercial pulsed Nd:YAG laser, operating at three wavelengths (355, 532 and 1064 nm) with pulse energies of 100, 200 and 400 mJ, respectively, at 20 Hz PRF. A detailed description of the system and its location is found in [33]. The system allows the determination of the optical properties of the atmosphere, including aerosol backscatter and extinction coefficients, as well as an indication of the type of aerosol present, based on the Lidar Ratio. The

During May 2011, the system was deployed in a vertical pointing mode during an intensive field campaign. A 24-hour period was selected that demonstrated the complexity of the local situation, which is dominated by topographical effects and prevailing meteorological conditions [33]. Vertical profiles of the Backscatter Coefficient (BSC) and the Colour Ratio were calculated for 30-minute periods from 17:30 – 19:59 and 21:42 – 23:36 LT. The BSC was highest for all frequencies between 19:30 and 19:59 LT (Figure 22a), indicating a strong inflow of aerosols, while after 21:42 LT the BSC showed much lower values (Figure 22b), representing a relatively clean air mass. At the same time, the Colour Ratio between all frequencies increased significantly, indicating the presence of small particles, especially between 0,8 and 1,3 km AGL [33]. Ground-level observations of PM10 and PM2.5 for the 24-hour period indicate that PM10 concentrations were almost twice as high as those of PM2.5 until about 18:00 LT (Figure 23). During the same period, the sodar observed extremely low wind speeds from varying directions. However, this resulted in very stable PBL conditions, and a temperature inversion began to develop from 18:30 onwards, reaching its greatest depth and intensity at 21:30. Thereafter, it gradually dropped in height and began to erode, as the air flow from the interior intensified, until it totally dissipated by 01:00 LT [33], due to the katabatic warming of the descending northerly airflow, which then also reduced the aerosol concentrations at ground level (Figure 23). Figure 24a shows the development of the surface inversion at 20:00 LT, overlaid by warm air flowing from the interior, with simultaneous downward motion below 240 m AGL (Figure 24b), highlighting the complex interaction of meteorology and topography in this region. This situation clearly demonstrates the need for solid environmental impact studies *before* locating industrial developments, in order to avoid any negative health impacts in the local population due to

lidar is co-located with a sodar / RASS system and an air quality monitoring station.

from sugar cane fires affected the urban atmosphere of Ourinhos.

**5.3. Scanning lidar in Cubatão** 

the accumulation of pollutants.

In the study region, ions such as magnesium (Mg2+) and calcium (Ca2+) are associated with the re-suspension of soil dust, which often accompanies biomass fires due to the intense updrafts created. On 26 August, concentrations of these species were higher during the daytime, due to the increased emissions from barren fields and unsealed roads associated with higher wind speeds (Figure 21).

**Figure 21.** Soluble major cation concentrations for the period 25-27 August 2010 (sampling periods are indicated in local time; after [85]).

Further evidence of the impact on the Ourinhos region of emissions from sugar cane fires was obtained by comparing the concentrations of organic compounds in aerosol particles collected on 26 August with those collected one day earlier. Ambient levels of polycyclic aromatic hydrocarbons (PAH), as well as PAH derivatives, such as oxy-PAH, were significantly higher on 26 August 2010 than on the previous day, confirming that emissions from sugar cane fires affected the urban atmosphere of Ourinhos.

## **5.3. Scanning lidar in Cubatão**

302 Atmospheric Aerosols – Regional Characteristics – Chemistry and Physics

strong presence of biomass smoke on both days.

with higher wind speeds (Figure 21).

indicated in local time; after [85]).

Aerosols collected during daytime and nighttime periods at the lidar site [85-87, 92], using low-volume filter samplers, were chemically characterized by means of ion chromatography. A higher concentration of K+ during the period from 22:00 on 25 August to 16:00 on 26 August 2010 indicated the presence of biomass-burning material (Figure 21), since K+ is a plant macronutrient released during the combustion process. Levoglucosan, a very specific chemical marker of biomass combustion, was well above average concentration during day sampling on 26 August and even higher during the following night, indicating a

In the study region, ions such as magnesium (Mg2+) and calcium (Ca2+) are associated with the re-suspension of soil dust, which often accompanies biomass fires due to the intense updrafts created. On 26 August, concentrations of these species were higher during the daytime, due to the increased emissions from barren fields and unsealed roads associated

**Figure 21.** Soluble major cation concentrations for the period 25-27 August 2010 (sampling periods are

An elastic backscatter lidar system, with similar characteristics to the mobile lidar, was installed in 2011 at CEPEMA-USP (*Centro de Pesquisas em Meio Ambiente,* a Center for Environmental Research and Training, under the responsibility of the Universidade de São Paulo) in the Cubatão industrial area, with the ultimate goal of remotely monitoring industrial emissions. It also uses a commercial pulsed Nd:YAG laser, operating at three wavelengths (355, 532 and 1064 nm) with pulse energies of 100, 200 and 400 mJ, respectively, at 20 Hz PRF. A detailed description of the system and its location is found in [33]. The system allows the determination of the optical properties of the atmosphere, including aerosol backscatter and extinction coefficients, as well as an indication of the type of aerosol present, based on the Lidar Ratio. The lidar is co-located with a sodar / RASS system and an air quality monitoring station.

During May 2011, the system was deployed in a vertical pointing mode during an intensive field campaign. A 24-hour period was selected that demonstrated the complexity of the local situation, which is dominated by topographical effects and prevailing meteorological conditions [33]. Vertical profiles of the Backscatter Coefficient (BSC) and the Colour Ratio were calculated for 30-minute periods from 17:30 – 19:59 and 21:42 – 23:36 LT. The BSC was highest for all frequencies between 19:30 and 19:59 LT (Figure 22a), indicating a strong inflow of aerosols, while after 21:42 LT the BSC showed much lower values (Figure 22b), representing a relatively clean air mass. At the same time, the Colour Ratio between all frequencies increased significantly, indicating the presence of small particles, especially between 0,8 and 1,3 km AGL [33]. Ground-level observations of PM10 and PM2.5 for the 24-hour period indicate that PM10 concentrations were almost twice as high as those of PM2.5 until about 18:00 LT (Figure 23). During the same period, the sodar observed extremely low wind speeds from varying directions. However, this resulted in very stable PBL conditions, and a temperature inversion began to develop from 18:30 onwards, reaching its greatest depth and intensity at 21:30. Thereafter, it gradually dropped in height and began to erode, as the air flow from the interior intensified, until it totally dissipated by 01:00 LT [33], due to the katabatic warming of the descending northerly airflow, which then also reduced the aerosol concentrations at ground level (Figure 23). Figure 24a shows the development of the surface inversion at 20:00 LT, overlaid by warm air flowing from the interior, with simultaneous downward motion below 240 m AGL (Figure 24b), highlighting the complex interaction of meteorology and topography in this region. This situation clearly demonstrates the need for solid environmental impact studies *before* locating industrial developments, in order to avoid any negative health impacts in the local population due to the accumulation of pollutants.

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 305

**Figure 24.** Sodar/RASS measurements on 25 May 2011, mean profiles for 19:30-20:00 LT. **(a)** Vertical

(b)

(a)

profile of temperature; **(b)** Vertical profile of vertical wind velocity.

**Figure 22. (a)** Vertical profile of Backscatter Coefficient (BSC) at 355 nm; **(b)** Colour Ratio 532/355 nm. (After [33]).

**Figure 23.** Concentrations of PM10 and PM2.5 from 10:00 LT on 25 May to 10:00 LT on 26 May 2011. (After [33]).

(After [33]).

(After [33]).

**Figure 22. (a)** Vertical profile of Backscatter Coefficient (BSC) at 355 nm; **(b)** Colour Ratio 532/355 nm.

(b)

(a)

**Figure 23.** Concentrations of PM10 and PM2.5 from 10:00 LT on 25 May to 10:00 LT on 26 May 2011.

Review of Aerosol Observations by Lidar and Chemical Analysis in the State of São Paulo, Brazil 305

(a)

**Figure 24.** Sodar/RASS measurements on 25 May 2011, mean profiles for 19:30-20:00 LT. **(a)** Vertical profile of temperature; **(b)** Vertical profile of vertical wind velocity.
