Study of Aerosols ’ Characteristics and Dynamics over the Kingdom of Saudi Arabia Using a Multisensor Approach Combined with Ground Observations

1Department of Prep Year Physics, College of Applied and Supporting Studies, King Fahd University of Petroleum and Minerals, Dhahran 31261, Saudi Arabia 2Department of Physics, Faculty of Science, Alexandria University, Moharam Beek, Alexandria 21522, Egypt 3Schmid College of Science and Technology, Chapman University, Orange, CA 92866, USA 4Center of Excellence in Earth Systems Modeling and Observations, Chapman University, Orange, CA 92866, USA 5Department of Environmental Science, Faculty of Science, Moharam Beek, Alexandria University, Alexandria 21522, Egypt 6Department of Earth Science, King Fahd University of Petroleum and Minerals, Dhahran 31261, Saudi Arabia


Introduction
Aerosols' impact on the regional climate and their ultimate connections to the Earth's global climate system and hence forcing has been observed through their optical and microphysical properties [1][2][3][4][5].Thus, improving our understanding of aerosol properties and characterization is imperative especially over mega and highly populated cities where aerosols have major impacts on human health [6][7][8][9].The kingdom of Saudi Arabia is one of the major sources of aerosols in the world, including natural and anthropogenic components [10,11].Yet, it still lacks for better characterization of its atmospheric aerosols properties with significant shortage of in situ observations.Over the Middle East, dust sources extend from the north of the Tigris-Euphrates basin to the coast of Oman, yet dust activity is quite complex and widely impacted by seasonal variability [12].Temporal and spatial characteristics of Saudi Arabian dust storms, with focus on associated air parcel trajectories, have been investigated using dust concentrations and geochemistry of aerosols, station, and gridded weather observations and remotely sensed aerosol observations [13][14][15].
Using PM 10 concentrations, recorded by the King Abdel Aziz City for Science & Technology (KACST) monitoring network during 2000-2003, nine local and four external dust sources have been identified, in association with seven types of dust storms, triggered by seasonal distribution of meteorological conditions, yet dominated by Haboob (∼42%) and Shamal (∼37%).A thorough investigation of one of the most intense dust storms experienced in Saudi Arabia over the last two decades that struck Riyadh and lasted several hours on March 10th, 2009 was conducted [16].Significant changes of aerosol and meteorological parameters were observed with air pressure rapidly increasing by 4 hPa, temperature decreasing by 6 ∘ C, relative humidly increasing from 10% to 30%, Aerosol Optical Depth (AOD) at 550 nm increasing from 0.396 to 1.71, and Angstrom Exponent (AE) rapidly decreasing from 0.192 to −0.078.On the other hand, seasonality and aerosols' concentration variability impacting cloud properties and serving as condensation nuclei were studied over the Arabian Sea using satellite and ground observations [17,18], and over other surrounding locations in Northern Africa and Gulf region [19][20][21][22][23][24][25][26][27][28][29].According to [18] seasonal variations are clearly found in the shape and magnitude of the volume size distribution of the coarse size mode due to dust emission.
A model for PM 10 dust emission was constructed by [30], using the concept of a threshold friction velocity which is dependent on surface roughness that in turn was correlated with geomorphology or soil properties.This was applied for Kuwait, Iraq, Syria, Saudi Arabia, the United Arab Emirates, and Oman.The model results agreed quantitatively with measurements at four locations in Saudi Arabia and one in Kuwait for one major dust event (>1000 g/m 3 ).On the other hand [31] incorporated a two-stream scattering scheme based on the delta-Eddington approximation into the Florida State University Limited Area Model for computing the shortwave radiative fluxes due to dust aerosols over the Saudi Arabian region and to study their impact on synoptic-scale systems and the diurnal cycle over the region.The average diameter of dust particles collected during a dust storm at various heights near Riyadh was tested by [32].It was found that most of the particle size distributions can be described by a lognormal or normal distribution depending on the storm condition and height.The average diameter of sand/dust particles decreases with the increase of height according to a power law.
In this paper, we present a detailed climatological analysis of the optical and microphysical aerosol properties using satellite observations over four main selected regions in Saudi Arabia; criteria of selection follow, highlighted in Figure 1.Furthermore, we discuss patterns of aerosol transport and characteristics during the significant March 10, 2009 dust storm as well as aerosol optical and physical characteristics as an extension of the work performed by [16].Data from two AErosol RObotic NETwork (AERONET) ground based stations, namely, the Solar Village (long: 46.397 lat.24.907) and MASDAR (long.54.617 lat.24.442) (indicated by 5 and 6, resp., in Figure 1), are also used for validation purposes.The two stations are yet not included in our four selected regions of interest but are seen of great importance for comparison between deserts and urban areas.Aerosols in general have regional behavior and long-range transport may be a factor; therefore, using the ground data is seen as a needed addition.
Region 1 represents the North Western (NW) cities of Saudi Arabia, where temperature ranges from ∼46 ∘ C during summer to ∼ −6 ∘ C during winter, with rain fall during the winter months from November to March.The NW region has small population of less than 600,000 and contains the Tabuk industrial city on an area of 1.4 million square meter and known for its petrochemical, plastic, aluminum, and steel industries.Region 2 represents the Western Province (WP) of Saudi Arabia that includes three major cities, namely, Jeddah, Mecca, and Madinah known for their high population especially during Muslims' pilgrimage season (Hajj), which is said to be the largest people gathering in the world with more than 2 million visits per year.Particulate pollution is a serious problem in the city of Mecca and in the neighboring regions due to particles emitted by traffic during the Hajj season [33,34].Moreover, the city of Mecca is located in a valley in approximately the middle of the Sarawat mountain range that produces large concentrations of airborne sand and anthropogenic aerosols.The WP retains its very warm temperature in winter ∼18 ∘ C increasing to ∼40 ∘ C during summer, with sparse rain falls between November and January.Region 3 represents the East province (EP) of Saudi Arabia, containing one of the largest industrial complexes in the Middle East including steel mill, oil refinery, world's largest desalination plants, electric power, and petrochemical products such as plastic and fertilizers.The climate of the EP is tropically hot and relatively humid with temperature ranging from ∼46 ∘ C during the summer to ∼13 ∘ C during the winter [35].
Finally, region 4 is located at the southern Saudi Arabia and represents the Rub al Khali Desert or Empty Quarter (EQ).
The EQ is a massive source of aerosols as it covers an area of ∼560,000 km 2 with daily maximum temperature reaching as high as 56 ∘ C and receiving annual rainfall of less than 30 mm.

Data Sets and Methods
In this work we utilized aerosol measurements from a number of sensors to leverage the wide variety of aerosol  [37].The abovementioned data will be used to characterize the apparent complex dynamics of the aerosols over the Kingdom that are dominated by coarse mode dust aerosols from the desert, modified and mixed with fine anthropogenic aerosol.
2.1.MODIS/Terra and MODIS/Aqua.MODIS/Terra and MODIS/Aqua instruments provide a number of validated aerosol products that include AOD at 10 km resolution over water and over land [38,39].The AOD uncertainty is ±0.05 (±0.15 * AOD) over land and ±0.03 (±0.05 * AOD) over ocean.Uncertainty in the monthly average fine fraction aerosol optical depth is ±20% [38].Over the desert region we use the Deep Blue product rather than the standard AOD product because the latter is retrieved using the dark-target approach [40] at near-infrared wavelengths (2.1 and 3.8 m) [41].The two blue channels (0.412 and 0.470 mm) allow the Deep Blue algorithm to be more sensitive to aerosols over bright surfaces for which surface reflectance is relatively small to infer aerosol properties [42].The MODIS Deep Blue algorithm primarily uses the UV channels to provide aerosol retrievals over deserts and other areas where the operational cannot.The uncertainties of the Deep Blue product are reported to be around 25-30% [43].The recent release of a new 3-km resolution aerosol product in the MODIS Collection 6 dataset [44] is highly advantageous for studying small regions such as those highlighted in the present study.

MISR/Terra
. MISR/Terra operational aerosol retrievals performed at 17.6 km horizontal resolution include some information about particle size, shape, and single-scattering albedo, in addition to AOD which will be used in this study [45][46][47][48].A global comparison of coincident MISR and AERONET sunphotometer data showed that overall, about 70% to 75% of MISR AOD retrievals fall within 0.05 or 20% * AOD, and about 50% to 55% are within 0.03 or 10% * AOD, except at sites where dust or mixed dust and smoke are commonly found [48,49].Here, we have analyzed the Level 3 monthly aerosol product averaging select Level 1 and Level 2 parameters over daily, monthly, seasonal, and annual time periods from January 2003 to August 2009 to study aerosols shape and size.

CALIPSO.
CALIPSO is a Franco-American mission that supplies a unique dataset of atmospheric vertical profiles measured by CALIOP on-board the satellite with a 30-m vertical resolution.CALIPSO data releases began in mid-June 2006 and include Level 1 radiances, a Level 2 vertical feature mask, and cloud and aerosol layer products [50].
2.4.OMI/Aura.OMI/Aura provides the (UltraViolet Aerosol Index) UVAI data [51][52][53].The UVAI is a product of the standard OMI aerosol retrieval, which indicates the UV absorbing aerosols [53].Absorbing aerosols such as carbonaceous aerosols, desert dust, and volcanic ash above the boundary layer yield positive UVAI values (>1), whereas nonabsorbing small particle aerosols yield small negative values [53].Absorbing aerosols in the boundary layer may produce small UVAI values (<0.5) that make it difficult to separate their signal from the background noise.Given the large size (13 × 24 km at nadir) of the OMI pixels, subpixel cloud contamination is a persistent problem resulting in the overestimation of AOD and underestimation of the singlescattering coalbedo [53].
2.5.AERONET.AERONET federation of Cimel sunphotometers provide daytime AOD measurements every 15-min on average covering the 340-1600 nm wavelength range with typical AOD uncertainties of ±0.015 [36,54,55].AERONET AODs are derived from direct-beam solar measurements, and some information about particle size and indices of refraction are derived from sky-scans which will be used in this study from the stations mentioned in Section 1.  shows a single peak pattern annually taking place during the spring season during which dust events are dominant.

Results and Discussion
Table 1 shows the average AOD with the standard errors over regions 1-4.To examine the AOD background level over the Kingdom, monthly means climatology over regions 3 and 4 using MODIS standard AOD and Deep Blue, respectively, at 550 nm is calculated and represented (Figures 3(a The monthly MODIS climatological means shows a cycle with minimum AOD during autumn and winter months and then peaks during spring and summer months.The maximum annual mean AOD 550 from Aqua and Terra is ∼0.82 ± 0.06, ∼0.72 ± 0.05, respectively, while the minimum annual mean is ∼0.28 ± 0.06, ∼0.28 ± 0.05, respectively.It is clear that there is still a relatively high AOD background observed over the selected regions which can be associated with the industrial pollution and petrochemical industry active in region, which is typical of industrial cities in developing countries [7,56,57].Such high background AOD levels would have an indirect impact on the local climate and precipitation levels.Hence, the relation between the mean precipitation rate and AOD levels is investigated and as a result, monthly climatology of the mean precipitation rate over region 3 is additionally plotted to (Figure 3(a)) during the period 2003-2013.Minimum precipitation rates during the summer and maximum during the winter are observed.During March-April and September-November we find that the effect of precipitation on aerosols concentration is not clear owing to the indirect effects as compared with other studies over the Mediterranean basin where precipitation rate was shown as one of the strongest removal processes for atmospheric aerosols in winter [58].As presented in Figure 3(a) the precipitation rate is very low ranging from 0.02 to 0.04 mm/month; hence, precipitation might have a lower contribution to the wet deposition of aerosols.The possible cause for the lower AOD values   can be precipitation over the source regions or precipitation during the aerosol travel from source to the observation site.If this is not the case, the most probable cause is the seasonal change of the wind stream pattern, which advects the aerosols from the source regions to the observation site as well as higher summer AODs due to accumulation of fine aerosol particles in summer [58].It is noteworthy that annual dust and anthropogenic pollutants are the main emission sources during March-April and September-November periods.Despite the fact that the available precipitation data suffer from errors, they still can be used for long time averaging [59].
Terra mean AOD results were higher than Aqua during May-August; however the difference is not significantly high, consistent with [38] where they showed that there is no significant difference between Terra and Aqua results.The monthly climatology means over Rub al Khali Desert (Region 4) displays a maximum AOD 550 value of 0.5035 ± 0.03 and a minimum of 0.1469 ± 0.03 representing lower monthly AOD as compared to that over region 3.This is attributed to the industrial activities and local air pollution over region 3.

Absorbing Aerosol Characteristics, Vertical
Structure, and Subtypes.Absorbing aerosols, originating from arid areas, have long-range transport capability, during which they usually interact with urban/industrial pollutants and other aerosols under various meteorological conditions.Such transport results in a complex mixing scenario among different aerosols, affecting the climate system radiative balance directly and indirectly [2,60].To detect and quantify these aerosols, such as carbonaceous and mineral dust, the near-UV aerosol sensing is adequate since they absorb ultraviolet (UV) radiation.
The UVAI data used in this study are sensitive to dust vertical distribution, wind-transport, and the dust size distributions, which is a reasonable approach for arid environments [61].Figures 4(a) and 4(b) present the daily OMI UVAI variations for 2012-2013 and monthly climatology means of the two-year period over region 3 (EP).We observe perennial positive values indicating the existence of absorbing aerosols all year round [62].The daily UVAI observations highlight a monthly component during the 2-year period with a clear spike during late spring and summer seasons which is evident from the UVAI monthly climatology showing maxima of 2.1 ± 0.13 in May (Figure 4(b)).The high UVAI values during May are consistent with large AOD values observed from the MODIS data (Figure 3(a)) as compared to those observed over region 4.This leads us to believe in the presence of higher air pollution levels over region 3, representing the oil rich eastern province of Saudi Arabia.It is evident that MODIS AOD and UVAI show a similar pattern in aerosol variability (Figures 3 and 4) yet with sharper AOD peaks during summer.This is owed to the UVAI sensitivity to aerosol layer height [62], where the dust aerosols exist in a wider range of altitudes than possible anthropogenic aerosols, which are confined to the planetary boundary layer (PBL) [26,28], where UVAI tends to be insensitive to these boundary layer aerosols [63].This mixing scenario between aerosols of natural and anthropogenic origin is observed from the aerosols vertical backscatter and subtypes obtained from CALIPSO which is quite useful here for observing weak aerosol layers and thin clouds by detecting optical depths of 0.01 or less [64].In Figure 5, the vertical profiles of the atmosphere up to 20 km, represented by total attenuated backscatter at 532 nm, are shown as CALIPSO overpasses the Eastern part of inland Saudi Arabia on 11 March (Figure 5

Microphysical Properties.
AERONET ground-based measurements of AOD data at 440 nm and 1020 nm are analyzed for the available periods of April 2012-December 2012 for Solar Village and from June 2012 till May 2103 for MASDAR to verify satellite data as discussed in Sections 1 and 2.5.As previously discussed we considered the Solar Village location as an AOD representative over desert, while the MASDAR one to represent the AOD over the EP of Saudi Arabia.Moreover, the HYSPLIT backward trajectories shown in Figures 6(a) and 6(b) emphasize the fact that regions 3 and 4 are one of the possible sources of aerosols received at the Solar Village and MASDAR, respectively.The observations from both stations at 440 nm show maxima of 0.71 in July and a minimum of 0.27 in March over MASDAR and a maxima of 0.08 in November and a minimum of 0.05 in August over Solar Village (Figure 7).The high July AOD value (0.71) over MASADR showed a close agreement with the July AOD value (0.76) retrieved from Aqua over the EP, which confirms a general agreement between satellite and MASDAR AOD observations.However, such agreement was not established between the August AOD value (0.39) from Aqua over region 4 as compared to the AOD low value (0.05) recorded by the Solar Village station.This is expected owing to aerosols deposition taking place as they move from Rub al Khali to the Solar Village for a long distance (Figure 6(a)).It is clear that the AOD values at 440 nm are higher than those at 1020 nm since the longer wavelength is greatly affected by water-vapor absorption and hence is used to derive the total water-vapor column [24].We use the AERONET AE (440/870) to identify the aerosol origin being natural, anthropogenic, or mixed over regions 3 and 4 of Saudi Arabia where small AE values are associated with large dust particles (natural origin) as compared to large AE values representing small size aerosols (anthropogenic/mixed) (Figure 8).The results show small AE values during spring (mostly desert dust) and larger AE values are observed during July-December mainly associated with smaller particles over MASDAR and Solar Village reflecting a more mixed scenario over both locations.This is attributed to the fact that pollutants as well as mixed aerosols with dust are quite common over Arabia.This is further   show that dust originated from the North West of Saudi Arabia and the east of Egypt where aerosols showed high dominancy at ∼3000 m (Figure 10).However, the middle of the desert where Solar Village is located exhibits mixed particles at 3000 (black) and 6000 (blue) m showing higher deposition as aerosols approach the EP and agreeing with the small AERONET AE values around the springtime favoring dust presence (Figure 10).MISR AOD, size, and shape fractions data was used to characterize the types of aerosol based on their physical and  It is also noticed that spherical particles are higher than nonspherical and show a more stable value all year round indicating high pollution levels.However, the nonspherical particles are significantly higher during March-July than the rest of the year (Figure 11(b)) which is consistent with the dust storms season extending from March till late summer [14].The AOD of nonspherical particles reaches a maximum in June with a value of ∼0.3517 ± 0.01, where the maximum of the spherical fraction AOD occurs in July with a value of ∼0.4867 ± 0.01.The large AOD fractions are higher during March-May with a significant contribution from the desert dust, while during the rainy season, November-February, the AOD of small, medium and large particles significantly drops as expected.
During spring and summer seasons the MISR AOD generally shows higher values for large and small size particles as compared to the medium ones shown in Table 2.These high values during these seasons disclose that most particles are of large sizes with a radius larger than 1 m especially for April and May.However, the October distribution shows that both small and large particles exists with more domination of large particles as can be revealed from the two peaks at ∼0.1 and ∼4 m (Figure 12).The above results related to AOD, AE, and size distribution during spring and other seasons agrees with the spring maximum local desert-dust activity as well as mixed scenarios during other seasons [66].

Conclusions
Sand-dust storms (SDSs) over the gulf region affect human life in various aspects including health and economy.They can be associated with climate change and result in different feedback mechanisms as well as with air pollution.Therefore, early detection and monitoring of SDSs is a matter of the utmost importance for possible risk reduction caused by these hazardous phenomena.Generally, satellite sensing using various sensors equipped for different purposes, provides synthetic ways of monitoring SDSs most effectively.In this study, multisensor approach is applied to severe SDS cases, which originated in Arabia and from the surroundings, to investigate characteristics of the case storm and to explore the capability of such approaches for its detection and monitoring.For this purpose, we employ several satellite sensors, namely, MODIS, OMI, and the lidar instrument and visible-infrared imagers on-board CALIPSO and data from the AERONET ground stations.Information on aerosol loading is provided by AOD retrieved from the Terra/Aqua MODIS.For retrieving AOD over bright land surface, DB algorithm was adopted.For the dust storm event, information on dust clouds was obtained by the standard AOD except over the bright surface areas around Rub al Khali (Empty Quarter) Desert, which were covered by DB AOD.This indicates that combined aerosol loading information available from other algorithms (and sensors) provide more continuous and complete detection of severe SDSs.
Overall the multisensor approach using satellite remote sensing provides useful information on various properties of the dust storm event, including location and magnitude of aerosol loading, size mode of aerosols, the behavior of long-lived absorbing aerosols, and the vertical structure.This enables us to make more continuous and complete detection of severe SDSs.In addition, our approach provides information on anthropogenic aerosols and pollutants, not just dust particles.Early detection of such aerosols and pollutants is especially very useful and important for health protection in the areas located downstream from the source region.

Figure 1 :
Figure 1: Map of Saudi Arabia including the four regions considered in this study represented by the cities of Tabuk, Mecca, and Dhahran and Rub al Khali (Empty Quarter) Desert.AERONET Solar Village and MASDAR locations are indicated by 5 and 6.

Figure 2 :
Figure 2: Time series plot of MODIS AOD at 550 nm using Terra and Aqua over (a) Tabuk, (b) Mecca, and (c) Dhahran and Aqua Deep Blue data over (d) Rub al Khali (Empty Quarter) Desert.
) and 3(b)) with standard vertical error bars of the monthly mean values.

Figure 3 :
Figure 3: Climatology MODIS monthly AOD means over (a) region 3 (EP) from Terra and Aqua (2003-2013), the right axis is for the precipitation rate of the EP for the same period, (b) region 4 (Rub al Khali) with standard deviations.

Figure 4 :
Figure 4: (a) The monthly climatology means of UVAI over region 3 (EP) for 2012-2013 and (b) is Daily OMI UVAI for 2012-2013.The vertical bars represent standard errors.

Figure 5 :
Figure 5: CALIPSO (a) total attenuated backscatter at 532 nm of a dust storm event as measured by the night-time CALIPSO overpass over the inland Saudi Arabia (at 2238 to 2251 UTC 11 March 2008) and (b) aerosol subtype over the same area.The surface elevation is shown as a thick black line.The inset map shows the path of CALIPSO overpass over the globe (black line) and the study region (red line).

Figure 7 :
Figure 7: AERONET AOD observations at 440 and 1020 nm from April 2012 to May 2013 using Solar Village and MASDAR stations.The vertical bars represent standard errors.

Figure 8 :
Figure 8: AERONET AE observations using MASDAR from April 2012 to May 2013 using MASDAR station.The vertical bars represent standard errors.

Figure 10 :
Figure 10: HYSPLIT backward trajectory dispersion starting from region 3 on March 09, 2009 16:00 UTC for 3 days.Heights are indicated by the legend bar.
) and 11(b) present the monthly size and shape averages, from 2003 to 2013 over region 3 where the large and small size particles' fraction showed higher dominancy than the medium ones (Figure11(a)) with even more large size particles abundance during spring season.Such particle size occurrence agrees with the AE values with distinct dusty versus anthropogenic episodes as compared to mixed ones.

Table 1 :
Annual AOD means for Terra and Aqua over regions 1-3 and for Aqua Deep Blue over region 4 (desert).
[14]Analysis of Aerosol Optical Depth (AOD).In this study, the MODIS Terra and Aqua AOD data were acquired to investigate the aerosols load, characteristics, extent, and temporal variability over three Saudi Arabian subregions including Tabuk(34.70∘E to 36.21 ∘ E and 27.60 ∘ N to 28.99 ∘ N); Mecca (38.42 ∘ E to 40.82 ∘ E and 20.50 ∘ N to 22.28 ∘ N), and Dhahran (49.98 ∘ E to 50.19 ∘ E and 26.21 ∘ N to 26.56 ∘ N) (Figure1).These regions are selected to represent the NW, W, and E A sudden increase in the aerosol concentration is observed during 2009 over region 3 (EP) that can be attributed to the major March 2009 dust event in Saudi Arabia as well as to some industrial activities in that region[14].This significant observed increase in the AOD concentration average values over Dhahran (Figure2(c)) during 2008-2009 yet has

Table 2 :
MISR AOD monthly mean size fractions for the period 2003-2013.
optical properties to study aerosols microphysics.The MISR AOD at 558 nm data categorizes particles into small, medium, and large sizes as well as to spherical and nonspherical.Figures11(a