Climate, not conflict, explains extreme Middle East dust storm

The recent dust storm in the Middle East (Sepember 2015) was publicized in the media as a sign of an impending ‘Dust Bowl.’ Its severity, demonstrated by extreme aerosol optical depth in the atmosphere in the 99th percentile compared to historical data, was attributed to the ongoing regional conflict. However, surface meteorological and remote sensing data, as well as regional climate model simulations, support an alternative hypothesis: the historically unprecedented aridity played a more prominent role, as evidenced by unusual climatic and meteorological conditions prior to and during the storm. Remotely sensed normalized difference vegetation index demonstrates that vegetation cover was high in 2015 relative to the prior drought and conflict periods, suggesting that agricultural activity was not diminished during that year, thus negating the media narrative. Instead, meteorological simulations using the Weather Research and Forecasting (WRF) model show that the storm was associated with a cyclone and ‘Shamal’ winds, typical for dust storm generation in this region, that were immediately followed by an unusual wind reversal at low levels that spread dust west to the Mediterranean Coast. These unusual meteorological conditions were aided by a significant reduction in the critical shear stress due to extreme dry and hot conditions, thereby enhancing dust availability for erosion during this storm. Concluding, unusual aridity, combined with unique synoptic weather patterns, enhanced dust emission and westward long-range transport across the region, thus generating the extreme storm.


Introduction
Dust emission from the land surface plays a significant role in climate, biogeochemical cycling, and other earth system processes [1][2][3]; it can disrupt weather, public health, and economic activity. Sediment availability for emission is enhanced by aridity [4,5], reduced vegetation cover [6,7], and soil disturbance [8]; factors that may be linked to land use dynamics and human activity, including conflict. Dust emission is also enhanced by certain synoptic meteorological patterns, such as frontal and cyclonic systems known to generate dust storms in the Middle East [9][10][11]. It is becoming necessary to understand and attribute the relative influence of climate and human factors on dust emission.
In late August and early September 2015, an unusually severe dust storm occurred in the Middle East, the trajectory of which is illustrated in figure 1. Satellite images show that the dust storm emerged around 31 August, 2015 along the northeastern border between Syria and Iraq. The storm was delineated by a wall of dust and a cyclone clearly evident from the satellite imagery. Over the next two days, the cyclone moved southeast toward and across the Persian Gulf. Finally, the dust was spread westward, resulting in Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.
Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. large portions of Syria, Lebanon, and the eastern Mediterranean Sea being obscured from the satellite view.
The Middle East is a notable hotspot dust source during the summer months and dust storms in this region are often associated with the Shamal (Arabic: , 'north') winds and cyclones [2]. The greatest frequency and longest persistence of cyclones in this region is observed over Syria and Iraq [9]. Furthermore, the highest density of dust sources in the Middle East occurs in northern Iraq between the Tigris and Euphrates rivers and along the Syria-Iraq border [12], regions with extensive desert cover, low population densities, and sparse agriculture concentrated along river valleys. These previously characterized regions of frequent cyclogenesis and high density of dust sources coincide with the origination point of the storm (figure 1).
The September monthly average aerosol optical depth (AOD) in 2015 was elevated across the entire region, relative to the climatological average for the years 2000-2015 (figure 2(A)). Most strikingly, the Mediterranean coastal region of western Syria, Lebanon, and Israel exhibited September monthly average AOD three to four times the climatological average. Daily AOD data convey a similar story across the region with peak storm values on 8 September, 2015 in the 99th percentile of historical values ( figure 2(B)). According to the historical data, daily AOD in the coastal region is typically less than that in the eastern, more arid part of the region; but was nearly twice as high during the peak of the storm ( figure 2(B)). Therefore, this dust storm was characterized by high atmospheric aerosol concentrations that were unusually and particularly intensified over the Mediterranean coastal region.
With scores of people being hospitalized, several media reports popularized the dust storm as a symbol of the ongoing conflict in the Middle East (table 1). These reports blamed land cover change attributed to conflict, including agricultural land desertion, reduced irrigation, and increased military vehicle traffic over unpaved surfaces for the unusually severe storm. Because these reports lack empirical support, we developed and tested alternative hypotheses to further our understanding of what caused the unexpected nature of this storm. First, we analyzed vegetation cover data as an indicator of possible agricultural land abandonment stemming from conflict. Second, we conducted simulations using the Weather Research and Forecasting (WRF) model to characterize the synoptic circulation patterns during the storm. Finally, we investigated surface air temperature, humidity, and wind speed measurements to estimate the impact of aridity on soil cohesion and erodibility.

Methods
Methods include those for AOD, vegetation cover, WRF model simulations, and estimates of critical shear stress from surface temperature and relative humidity data. The impact of increased military traffic on dust emission could not be quantified due to a lack of data.

Aerosol optical depth
To evaluate the severity of the 2015 dust storm, Moderate Resolution Imaging Spectroradiometer (MODIS) AOD data were obtained. The AOD data are daily and monthly values computed by the Deep Blue algorithm over land. Relative AOD anomalies were computed for the September monthly average in 2015   Ground meteorological data, soil matric potential, and wind erosion threshold The physics of wind erosion are complex, as they involve atmospheric, soil, and land surface processes [13]. Wind erosion occurs when wind speed, u, exceeds a certain threshold value, u t [14]. Several factors affect u , t including field surface conditions, surface roughness, stability of the soil aggregates, size and shape of soil particles, clay content, vegetation cover, and near-surface soil moisture [15,16]. Theoretical and empirical expressions have been proposed to relate u t to soil surface moisture [15][16][17][18][19][20][21][22][23].
Studies describing the impact of mean wind on erosion usually use friction velocities, u⁎, because they are directly related to the surface shear stress, τ, exerted on the soil surface (τ=ρ u⁎ 2 , ρ is the air density). Humidity affects the threshold u⁎ t and is known to vary on a global scale, as do dust emissions [5,24]. This threshold evolves from the effect of inter-particle capillary forces (liquid bridge bonding) and hygroscopic forces (adsorbed layer bonding). The key macroscopic variable shaping these inter-particle bonding forces is the soil water matric potential, ψ [5,16,19,[25][26][27]. The Kelvin equation describes the dependence of ψ on relative humidity, RH, and surface temperature, , T is the surface temperature expressed in K, and V w is the molar volume of water (=1.8 10 −5 m 3 mol −1 ). Under arid conditions, ψ is negative relative to atmospheric pressure. A model to relate relative humidity and temperature to the critical shear stress for particle motion, τ * t , is given as

WRF model simulations
An 11-day regional-scale simulation from 30 August to 10 September, 2015 was conducted using the Advanced Research Weather Research and Forecasting (WRF-ARW) Model over the Middle East. The simulation consists of two nested domains (d01 and d02), whose resolutions are 9 km and 3 km, respectively. Domain 1 (d01) covered a much larger area while domain 2 (d02) focused on the area where the dust storm was observed (figure S1). Domain 1 includes 300 (west-east) by 239 (north-south) grid cells in the horizontal directions and Domain 2 includes 451 (west-east) by 394 (north-south) grid cells. The top of the domain was set to 50 hPa and 35 vertical levels were used for both d01 and d02.
The WRF simulation was initialized using National Centers for Environmental Prediction FNL (Final) Operational Global Analysis data at 00 UTC on 30 August, 2015 and was forced by the same data set every six hours at the lateral boundaries of d01. Oneway nesting was used (i.e., d02 was forced by d01 but did not provide any feedback to influence d01). The time steps were 30 s and 10 s for d01 and d02, respectively.
The physical parameterization schemes were chosen based on previous studies [28], including (1) the Dudhia scheme for shortwave radiation, (2) the Rapid Radiative Transfer Model scheme for longwave radiation, (3) the WRF Single-Moment 6-Class microphysics scheme, (4) the Yonsei University planetary boundary layer scheme, and (5) the Noah land surface model with the single-layer urban canopy model for cities. No cumulus parameterization was used due to the fact that the resolutions of the domains were both below 10 km. The radiation time step was chosen to be 10 min. The 30 s resolution MODIS land-cover data set was used, which includes 20 land-use categories. The resolutions of other static data (e.g., topography, soil type) were 5 min and 2 min for d01 and d02, respectively, to be consistent with the resolutions of the grids. The WRF simulated results agree favorably  figure S2 (supplementary material).

Results and discussion
Agriculture in northern Syria and Iraq was robust in 2015, despite a declining trend due to recent drought. It is estimated that nearly 1.5 million people migrated internally from rural agricultural areas to urban areas, initiated by a severe drought during 2007-2010 [29]. However, rainfall recovery following the most recent drought (2013 and 2014) resulted in greater than average vegetation density and wheat yields in 2015, both comparable to pre-drought conditions [30]. Across much of the region, vegetation cover in 2015 was nearly double the 2007-2010 average ( figure 3(A)) and was also greater than the 2001-2007 pre-drought average ( figure 3(B)) in many areas. Regions of vegetation decline were relatively sparse and concentrated near deserts. This finding indicates that land cover changes associated with the ongoing conflict were unlikely to be at the origin of increased erodibility of the soil surface and suggests meteorological conditions as the more probable driver for enhanced dust uplift and transport. As mentioned earlier, dust storms in the Middle East are often associated with the Shamal winds and cyclones [11,31]. To illustrate, figure 4 shows the low-level (700 hPa, figure 4(A)) and high-level (300 hPa, figure 4(C)) synoptic circulation patterns on 31 August, 2015 simulated using the WRF model (see the Material and Methods section). The Shamal was characterized by the northwesterly winds to the west of a low-pressure zone located slightly to the east of the northern Syria-Iraq border (figures 4(A) and (C)). However, the Shamal winds do not explain the extremely elevated dust concentrations over the Mediterranean coast on 7 and 8 September, as shown in figures 1 and 2. The WRF simulation results further reveal that on 6 September (figures 4(B) and (D)) the front moved to the north and persisted at 300 hPa, but not at 700 hPa. Interestingly, the low-level winds ( figure 4(B)) showed a reversal as compared to the high-level winds that remained northwesterly ( figure 4(D)). This surface wind reversal, subsequent to the cyclonic activity, likely both enhanced surface shear stress and transported previously suspended particles westward, thereby causing the extremely elevated dust concentrations along the Mediterranean coast. It is noted that westward dust transport in this region is not common [32].
In addition to the synoptic circulation patterns that favor dust storm genesis and development, surface air temperature and humidity measurements from the Har Kenaan meteorological station in Northern Israel show that the summer of 2015 was unusually dry and hot relative to the most recent 20 years (figures 5(B) and (C)). In fact, extreme high temperature and low humidity were more frequent during August and September 2015 than during the previous drought period 2007-2010. The soil matric potential, y, and related threshold shear stress required for wind erosion (see the Materials and Methods section), t * , t were lower during these dry and hot conditions (5), increasing the likelihood of dust emission. In the weeks preceding the dust storm, low humidity and high temperature extremes caused t * t to reach unusually low values. At times, t * t in August and September 2015 was approximately 75% relative to the average during 1994-2014 ( figure 5(D)). Low t * t relative to average conditions implies easier and more likely dislodging of large dry dust particles into the atmosphere. Even if military vehicle traffic was a contributor to this dislodging, which remains a possible but not necessary factor in the genesis of this dust storm, it would have been aided by reduced particle cohesion due to extreme aridity. It has been suggested that long-term changes in atmospheric circulation over the region have increased aridity [29]. Therefore, such trends could affect the critical shear stress required for wind erosion, and may cause extreme dust storms to become more frequent in the region in the future.
Typically, Saharan dust particles range in diameter from d p =0.1-10 μm and the associated terminal velocity for this d p range is 0.01-20 mm s −1 (3). Uplifting of dust particles to a minimum height of 1000 m by convective updrafts is reasonable. Given that the WRF predicted maximum boundary layer height exceeds 2 km, a minimum settling time of 14 h is plausible. The simulated easterly wind speed at 700 hPa on September 6 was approximately 30 km h −1 , giving a minimum travel distance ranging from 400 km (for d p =10 μm) to 800 000 km (for d p =0.1 μm) during such settling times. These travel distances are well within the domain area of interest and it can be surmised that particle size was not a limiting factor for long-range transport.

Conclusion
Climate, land use, and conflict must be simultaneously acknowledged as elements in a complex humanenvironment relationship that has been playing out in the Middle East as far back as 4000 years ago, following the collapse of the Akkadian empire [33,34]. Then and now, climate variability and control of scarce water resources have played a critical role in regional politics and conflicts [29] and their impacts on land use and land cover cannot be overlooked. Viewed from this perspective, it is understandable why the extreme 2015 dust storm was publicized as a sign of an impending 'Dust Bowl', and its severity attributed to the ongoing regional conflict. However, our analysis showed that aridity and synoptic meteorology played a prominent role in dust emission and transport for this episode. The storm occurred during a period of extreme aridity and was associated with a cyclone and 'Shamal' winds, typical for dust storm generation in this region. The Shamal was immediately followed by an unusual wind reversal at low levels that spread dust west to the Mediterranean Coast. This represents the key finding of this study and its importance is underscored by the reversibility of extreme dust events. Conflict-induced land cover change is reversible in the near term, potentially averting a persistent increase in dust storm frequency and severity; however, if such dust storms result from aridity and synoptic meteorology that could be significantly affected by climate change, then prolonged impacts in the Middle East may be unavoidable.