**4. Dust storms, their causes, effects, and attempts to forecast and model them**

This third and concluding section describes both the basics of dust storms and how forecasting and numerical simulation of dust storms are accomplished today. Dust and sandstorms afflict most arid and semi-arid regions of the world, cause serious problems in commercial air traffic and vehicular traffic, degrade building surfaces, lead to increased house and office cleaning costs, and adversely affect human respiratory health. Because of their ultra-high turbulence as they contact the land surface, and because their resultant particulates concentrations are extremely heterogeneous in both time and space, accurately simulating these phenomena remains an elusive goal. Despite these shortcomings in the simulations, weather forecasters are still faced with the necessity of predicting these storms' locations, durations, and severities. These predictions then allow authorities such as the National Weather Service or highway departments to broadcast near real-time warnings to the vehicular-driving public.

As the authors live in Arizona, the next remarks concern the landscapes and weather of this state. Arizona has three distinct physiographical provinces: (1) lowland deserts in the south and southwest, (2) rugged mountainous highlands in its north-central region, and (3) the Colorado Plateau -- a broad, high- elevation plain comprising its northern third. In the populated areas the elevations range from 43 m (140 feet) above sea level in the far southwest corner at Yuma

to about 2100 m (6,900 feet), an elevation that extends from the north-centrally located Flagstaff in a broad swath to the east-southeast, culminating in the far east-central region next to New Mexico. These substantial elevation differences lead to pronounced differences in weather. Extreme inclement weather often leads to vehicular crashes, whether it is heavy rain, thick fog, heavy snowfall, icy roads from wet winter rain or snow, or blowing dust. Except for fog, a rare phenomenon in Arizona, this marked spatial variation in weather leads to vehicular crashes from all these extreme weather conditions somewhere in the state. The following discussion is limited, however, to the lowland deserts and blowing dust. The focus here is on how the predictive capacity of the weather-forecasting and of the atmospheric science communities has been brought to bear on developing better dust predictions to reduce vehicular crashes and to reduce population exposure to unhealthful levels of airborne particulates. In recent years, a considerable body of work on this subject has been conducted by the National Weather Service, by the Arizona Department of Transportation, by other governmental agencies, and by academic researchers.

## **4.1 Dust storm and sandstorm basics**

Two lengthy reports provide information for this discussion, first, a global report assembled by multiple researchers for the United Nations Environmental Programme [1], and second, a comparable report limited to Arizona [28]. Both reports thoroughly discuss the atmospheric physics and meteorology of dust storms and sandstorms. Both estimate the economic damage wrought by dust storms. Both consider mitigation efforts to reduce the flux of anthropogenic dust. To the interested reader, both are worth obtaining and studying.

Dust storms occur whenever strong winds encounter dry, erodible land surfaces. Entrainment of particles occurs when the wind shear stress exceeds the ability of the surface material to resist detachment or transport. Wind erosivity is a product of wind velocity and wind flow characteristics, especially turbulence near the ground. In addition to ambient wind speed, vegetation and land-form characteristics of surface roughness play a large role in determining wind erosivity. Local wind conditions are also influenced by wind systems generated over larger areas, and thus may depend on land use and other physical characteristics in neighboring regions. Consider the "dry, erodible land surfaces", which, according to [1], consist of 75% natural landscapes, such as the Sahara or Gobi Deserts and dry lake beds, and of 25% anthropogenic land surfaces, such as active or abandoned agricultural fields, unpaved roads, large mining and construction sites, and so forth. Mitigating dust emissions can only be directed to the anthropogenic dry land surfaces, so mitigation discussions are limited to this human-caused one fourth of the problem. The global report [1] presents their Table 2.2, p. 10, that gives the different types of land surfaces that can or cannot produce dust in high winds (**Table 3**).

Although many of these land surfaces have low or moderate dust or sand potential, the more important ones are (1) lakes that are ephemeral or are of dry, nonconsolidated surfaces; (2) high-relief alluvial deposits that are both unarmored and unincised; and eolian sand dunes. Most agricultural soils for growing crops have been accumulated through alluvial processes, oftentimes augmented by the deposition of loess, so these fields when in between crops or when fallow or abandoned have high potential for dust emissions. Lakes in this list of dust potential actors are sometimes shallow water bodies constructed in large irrigation projects, but for various reasons the upstream waters that could be diverted to fill them become unavailable, leaving expansive dry lake beds prone to heavy dust emissions. In contrast with the eolian sand dunes and natural dry lake beds, these two categories of dust producers are amenable to mitigation.

*Bowing Sand, Dust, and Dunes, Then and Now–A North American Perspective DOI: http://dx.doi.org/10.5772/intechopen.98337*


#### **Table 3.**

*Different land surface types that can (or cannot) produce blowing sand and dust, (Table 2.2 of [1]).*

The literature on landscape characteristics of dust potential, and on the atmospheric physics and meteorology of the formation, transport, and eventual dissipation of dust storms is both voluminous and can be highly technical. For greater detail, the reader is referred to the two already cited reports or other textbooks or journal articles. Both long reports would be comprehensible for the average reader.

To summarize dust storms, their *formation* comes about from extremely high surface winds, produced either from massive thunderstorms or from synoptic weather fronts. Their direction and distance of *transport* is determined by the continuing influence of these winds, in conjunction with their continued ability to contact erodible land surfaces. Their *dissipation* occurs as the wind speeds decrease with the weakening of either the thunderstorm activity or the large-scale frontal movements. The global report presents in their Figure 2.2 and Table 2.1 a clear conceptualization of the phenomenon, along with the influential physical characteristics of weather variables, of soil surfaces, of vegetation, and of landforms, shown as **Figure 9** and **Table 4**.

#### **Figure 9.**



*(+) indicates that the factor reenforces wind erosion; (*−*) indicates that the factor has a protective effect, reducing wind erosion; (+/*−*) indicates that the effect can be positive or negative depending on the processes involved.*

#### **Table 4.**

*Key physical factors influencing wind erosion (Table 2.1 of [1]).*

In the Arizona report [28] the authors state that in Phoenix from 1948 to 2015 the number of summer dust storms ranges from three to five in the earlier years down to one to three in the latter years. They speculate that the decrease in frequency may stem from the expanding Phoenix urban area, in which formerly outlying agricultural lands with considerable dust potential have been converted into residential and commercial buildings, into landscaping that includes parks and lawns, and into what generally is called the "built environment". They present one photograph of a dust storm, shown as **Figure 10**.

**Figure 10.** *Dust storm of 5 July 2011, as it approaches the National Weather Service office at Sky Harbor airport in Phoenix: (Figure 36 of [28]).*

As for the observational tools and methods of predicting these dust storms, the authors offer the following, summarized in **Table 5**.

That concludes the summaries of the lengthy global and Arizona reports on all aspects of dust storms. This review paper now continues, and concludes, with a discussion of how these storms can be forecast or simulated.

#### **4.2 Forecasting and simulating dust storms**

The narrative immediately above covers the day-to-day observational tools and predictive systems for dust storms. At least two questions remain: (1) how are these dust storms studied by numerical simulations, and (2) how well do these simulations match the various observations such as satellite observations, Doppler radar images, and ground-based measurements of particulates concentrations? What follows are two examples of recent research on dust storm simulations. Both examples are highly technical papers unsuited for the non-technical reader.

One instructive example of the difficulties in performing these simulations and of what improvements might be forthcoming, can be found in the work of [12]. The authors assert that "regional-to-global models generally do not accurately simulate these storms", for two reasons: "(1) using a single mean value for wind speed per grid box, i.e., not accounting for subgrid wind variability and (2) using convective parametrizations that poorly simulate cold pool outflows". Their remedies take two forms. First, they "incorporate a probability distribution function for surface wind in each grid box to account for subgrid wind variability due to dry and moist convection." Second, they use "lightning assimilation to increase the accuracy of the convective parameterization to better simulate cold pool outflows".

These researchers built the subgrid wind variability and lightning assimilation into two different physico-chemical models: the Weather Research and Forecasting


#### **Table 5.**

*Observational tools, warnings, and prediction systems for dust storms.*

Model (WRF) and the Community Multiscale Air Quality model (CMAQ ). The windblown dust emissions parameterizations employed incorporate saltation bombardment (sandblasting) and a novel dynamic relation for the surface roughness length. To better estimate vegetative cover, these researchers used the fraction of absorbed photosynthetically available radiation (fPAR) from the Moderate Resolution Imaging Spectroradiometer (MODIS), which is a satellite-based instrument. Earlier work showed that the modeled airborne soil concentrations agreed quite well with observations in the spring, but that it underestimated these concentrations in summer, when convective dust storms are most frequent and most severe.

As for improving convection through lightning assimilation, the authors used the Kain-Fritsch convective scheme in WRF by activating its deep convection where lightning is observed and suppressing it where lightning is absent.

The authors went on to test their modified model on the major dust storm of 5 July 2011, which began with late afternoon severe thunderstorms near Tucson, Arizona. Cold pool outflows associated with this region of large storms moved northwest toward Phoenix, bringing with them a wall of dust extending 160 km wide and 1.5–1.8 km high. Both modifications – to the winds within the subgrids and to the deep convection scheme employed when lightning was present – enabled *Bowing Sand, Dust, and Dunes, Then and Now–A North American Perspective DOI: http://dx.doi.org/10.5772/intechopen.98337*

the simulated particulates concentrations from CMAQ to better match the measured [PM10], as shown in **Figure 11**. [Note: "[PM10]" is read as "concentrations of PM10".]

The work just summarized is highly technical, even for atmospheric scientists who study these phenomena. Missing from this work is any explicitly numerical comparison of model-generated versus observed [PM10]. While the concentration maps of **Figure 11** are illustrative, they are far from definitive. The next (and last work) summarized, which does have these explicit comparisons, is also on the technical side, but perhaps is not quite as obtuse, an opinion better left to the interested reader.

This is the work of [13], in which researchers analyzed nine dust storms in south-central Arizona with the Weather Research and Forecasting model with chemistry (WRF-Chem) at 2 km resolution. The all-important windblown dust emission algorithm was the Air Force Weather Agency model [29]. In all simulations of air pollutant concentrations, it is essential to get the emissions quantified accurately both temporally and spatially. For windblown dust emissions this goal frequently proves to be elusive because the available coverages of soil moisture, surface roughness, and vegetative cover suffer from both insufficient resolution and from temporal delays between the observations of these variables and the event itself. In this highly dynamic environment, with rainfall stochastically distributed in localized pockets, and with soil surface texture varying widely even within small areas, the uncertainties of the emitted dust flux reach unreasonable proportions. Nonetheless, one proceeds with what information one has.

In comparison with ground-based [PM10] observations, this modeling system unevenly reproduces the dust-storm events. The model adequately estimates the location and timing of the events, but it is unable to precisely replicate the magnitude and timing of the elevated hourly [PM10]. Furthermore, the model underestimated [PM10] in highly agricultural Pinal County for two reasons. First, because it underestimated surface wind speeds and, second, because the model's erodible fractions of the land surface data were too coarse to effectively resolve the active and abandoned agricultural lands.

In Phoenix the model's performance depended on the event, with both underand over-estimations partly due to incorrect representation of urban features. Increasing the fraction of erodible surfaces in the Pinal County agricultural areas improved the simulation of [PM10] in that region. Both 24-hr and 1-hr measured

#### **Figure 11.**

*Simulated hourly PM10 surface concentrations (*μ*g m*<sup>−</sup> *3) at 06:00 UTC on 6 July 2011 (23,00 local time on 5 July 2011) from three runs (left to right) (1) (control --no lightning assimilation (LTGA) and no subgrid wind variability (SGWV), (2) with SGWV, and (3) with SGWV and LTGA), overlaid with the observations of 11 PM10 monitoring sites. (this is lowest panel from Figure 6 of [12]).*

**Figure 12.**

*Model static fields: (a) main land cover and land use type in south-Central Arizona, and (b) fraction of erodible surface (Figure 2 of [13]).*

[PM10] were, for the most part, and especially in Pinal County, extremely elevated, with the former exceeding the health standard by as much as 10-fold and the latter exceeding health-based guidelines by as much as 70-fold.

The authors present several graphics that depict, among other things, the landscape and the degree of erodible surface (**Figure 12**).

**Figure 13** is a sample of the model's inability to match the observations, in which each panel represents a different dust storm. The observations in each case came from eight to 13 continuous PM10 monitoring sites, all in Pinal County. In two storms the model grossly over-estimated the observed values; in the other four the model greatly underestimated the measured peak concentrations.

**Figure 13.**

*Comparison of averaged PM10 time series over Pinal County for different cases: (a) April 13–14, 2006 (total 8 sites), (b) July 7–18, 2009 (total 9 sites), (c) January 21–22, 2010 (total 12 sites), (d) July 21–22, 2012 (total 13 sites), (e) June 30–July 1, 2013 (total 18 sites), (f) July 3–4, 2014 (18 sites), (g) June 27–28, 2015 (total 18 sites), and (h) July 7–8, 2014 (total 18 sites) (Figure 3 of [13]).*

The authors conclude: "Given the severity and frequency of these dust storms and conceding that the modeling system applied did not produce the desired agreement between simulations and observations, additional research in both the windblown dust emissions model and the physico-chemical model is called for."

Thus, concludes the last part of this four-part review paper that has presented information on sandstone formation, on sand dune field formation and dynamics, on the 1930s Dust Bowl saga, and on dust storm and sandstorm basics and the forecasting and prediction thereof. The interested reader is encouraged to consult the references for a more in-depth look into these subjects.
