Dust emission parameterization scheme over the MENA region: Sensitivity analysis to soil moisture and soil texture
2015; Wiley; Volume: 120; Issue: 20 Linguagem: Inglês
10.1002/2015jd023338
ISSN2169-8996
AutoresImen Gherboudj, S. Naseema Beegum, Béatrice Marticorena, Hosni Ghedira,
Tópico(s)Atmospheric chemistry and aerosols
ResumoJournal of Geophysical Research: AtmospheresVolume 120, Issue 20 p. 10,915-10,938 Research ArticleFree Access Dust emission parameterization scheme over the MENA region: Sensitivity analysis to soil moisture and soil texture Imen Gherboudj, Corresponding Author Imen Gherboudj Earth Observation and Hydro-Climatology Lab, MASDAR Institute of Science and Technology, Abu Dhabi, UAE Correspondence to: I. Gherboudj, igherboudj@masdar.ac.aeSearch for more papers by this authorS. Naseema Beegum, S. Naseema Beegum Earth Observation and Hydro-Climatology Lab, MASDAR Institute of Science and Technology, Abu Dhabi, UAESearch for more papers by this authorBeatrice Marticorena, Beatrice Marticorena Laboratoire Interuniversitaire des Systèmes Atmosphériques, UMR CNRS 7583, Universités Paris Est Créteil-Paris Diderot, IPSL, Créteil, FranceSearch for more papers by this authorHosni Ghedira, Hosni Ghedira Earth Observation and Hydro-Climatology Lab, MASDAR Institute of Science and Technology, Abu Dhabi, UAESearch for more papers by this author Imen Gherboudj, Corresponding Author Imen Gherboudj Earth Observation and Hydro-Climatology Lab, MASDAR Institute of Science and Technology, Abu Dhabi, UAE Correspondence to: I. Gherboudj, igherboudj@masdar.ac.aeSearch for more papers by this authorS. Naseema Beegum, S. Naseema Beegum Earth Observation and Hydro-Climatology Lab, MASDAR Institute of Science and Technology, Abu Dhabi, UAESearch for more papers by this authorBeatrice Marticorena, Beatrice Marticorena Laboratoire Interuniversitaire des Systèmes Atmosphériques, UMR CNRS 7583, Universités Paris Est Créteil-Paris Diderot, IPSL, Créteil, FranceSearch for more papers by this authorHosni Ghedira, Hosni Ghedira Earth Observation and Hydro-Climatology Lab, MASDAR Institute of Science and Technology, Abu Dhabi, UAESearch for more papers by this author First published: 28 September 2015 https://doi.org/10.1002/2015JD023338Citations: 18AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract The mineral dust emissions from arid/semiarid soils were simulated over the MENA (Middle East and North Africa) region using the dust parameterization scheme proposed by Alfaro and Gomes (2001), to quantify the effect of the soil moisture and clay fraction in the emissions. For this purpose, an extensive data set of Soil Moisture and Ocean Salinity soil moisture, European Centre for Medium-Range Weather Forecasting wind speed at 10 m height, Food Agricultural Organization soil texture maps, MODIS (Moderate Resolution Imaging Spectroradiometer) Normalized Difference Vegetation Index, and erodibility of the soil surface were collected for the a period of 3 years, from 2010 to 2013. Though the considered data sets have different temporal and spatial resolution, efforts have been made to make them consistent in time and space. At first, the simulated sandblasting flux over the region were validated qualitatively using MODIS Deep Blue aerosol optical depth and EUMETSAT MSG (Meteosat Seciond Generation) dust product from SEVIRI (Meteosat Spinning Enhanced Visible and Infrared Imager) and quantitatively based on the available ground-based measurements of near-surface particulate mass concentrations (PM10) collected over four stations in the MENA region. Sensitivity analyses were performed to investigate the effect of soil moisture and clay fraction on the emissions flux. The results showed that soil moisture and soil texture have significant roles in the dust emissions over the MENA region, particularly over the Arabian Peninsula. An inversely proportional dependency is observed between the soil moisture and the sandblasting flux, where a steep reduction in flux is observed at low friction velocity and a gradual reduction is observed at high friction velocity. Conversely, a directly proportional dependency is observed between the soil clay fraction and the sandblasting flux where a steep increase in flux is observed at low friction velocity and a gradual increase is observed at high friction velocity. The magnitude of the percentage reduction/increase in the sandblasting flux decreases with the increase of the friction velocity for both soil moisture and soil clay fraction. Furthermore, these variables are interdependent leading to a gradual decrease in the percentage increase in the sandblasting flux for higher soil moisture values. Key Points Alfaro and Gomes (2001) parameterization scheme well represent dust emission of MENA region Soil moisture and soil texture have significant roles in the dust emissions of MENA region Soil moisture effect on dust emission is higher over semiarid region than arid region 1 Introduction Arid and semiarid areas, such as the MENA (Middle East and North Africa), constitute one of the major sources of mineral dust aerosols on Earth, because large amounts of soil-derived dust are emitted from these regions [Prospero et al., 2002; Reid et al., 2008]. The mass mean diameters of emitted aerosols lie between 0.1 µm and about 20 µm [Patterson and Gillette, 1977], and these may be transported thousands of kilometers from their sources of origin. Numerous studies have shown the manifold effects of atmospheric dust on (1) atmospheric chemistry by providing surface area for chemical reaction [Chen et al., 2011], (2) marine biogeochemistry by the deposition of nutrient rich mineral dust [Jickells et al., 2005], (3) energy balance of the Earth-atmospheric system which alters the atmospheric dynamics and climate [Satheesh and Moorthy, 2005], (4) air quality and visibility [Prospero, 1999], and (5) human health [Perez et al., 2008]. Production of mineral dust from loose soil in arid and semiarid areas involves two major physical processors such as saltation and sandblasting [Gomes et al., 1990; Shao et al., 1993]. These two processes strongly depend on the soil and surface properties. While saltation refers to the horizontal movement of a layer of soil by the action of prevailing low-level wind, sandblasting refers to the release of fine dust particles by the disaggregation of soil grains during the saltation process. However, dust emission is initiated only when a critical wind velocity, known as threshold friction velocity, which is a function of soil moisture, surface roughness, soil texture, and grain size distribution, is attained [Alfaro et al., 1998; Fécan et al., 1999]. By its direct contribution to the interparticle cohesion, soil moisture is considered as one of the most important variables affecting the dust emission mechanism. Soil moisture increases the erosion threshold, due to the reinforcement of soil cohesion rising from (i) molecular adsorption on the soil grain surface and (ii) capillary forces between the grains. Thus, the increase in soil moisture can suppress dust emission very effectively [Ishizuka et al., 2008]. Several models have been proposed to explain the relation between top layer soil moisture and threshold friction velocity for different soil types [Chepil, 1956; Bisal and Hsieh, 1966]. In one of the earliest works, [Chepil, 1956] studied the soil moisture and dust emission relationship in a wind tunnel by measuring soil loss rates from trays filled with erodible soil at different water content levels and found that as the soil water content increases, the soil loss rate decreases. It was found that clayey soil requires a higher moisture content to prevent wind erosion than sandy loam [Bisal and Hsieh, 1966]. It was also observed that a small amount of moisture is required to make soil resistant to wind erosion. Similarly, Selah and Fryrear [1995] compared five different soil types and found that a larger threshold friction velocity is needed to move particles of higher moisture content both for abrading and nonabrading conditions. However, a lower wind velocity is needed to initiate wind erosion under abraded conditions. In another study, Fécan et al. [1999] compared the theoretical and empirical relationship between threshold friction velocity and soil moisture and proposed new equations to reproduce the change in threshold wind friction velocity caused by soil moisture variation. Dust emission, however, depends also on soil erodibility, vegetation cover, and soil surface roughness. While the erodibility represents the efficiency of the soil for the dust generation at a given meteorological forcing [Zender et al., 2003], the vegetation cover limits dust emission by reducing the surface wind velocities and influencing the dust transport. However, it is important to highlight that dry lands having sparse vegetation can emit significant amount of dust from the gaps available within the vegetation coverage depending upon the annual precipitation pattern which enhances the vegetation growth and coverage [Urban et al., 2009; Okin et al., 2011]. The effect of the surface roughness is to decelerate wind erosion by covering a portion of vulnerable surface and reducing a portion of the wind's momentum [Wolfe and Nickling, 1993]. There are a few studies that attempted to reproduce the natural condition for dust emission by setting up experiments in an open field [Ishizuka et al., 2005]. Such experiments are difficult to undertake because of (i) the lack of in situ measurements, which have limited coverage to fully characterize the dust emission process at the regional scale, and (ii) the transported dust particles from the surrounding areas. Obviously, a better understanding of dust emission can be undertaken through modeling of this process. There are two broad categories of dust parameterization schemes in the literature [Zender et al., 2003], based on their complexities: (i) the simple category, termed the bulk mobilization scheme, in which the dust emission is estimated in terms of the third or fourth power of the wind friction speed and the emitted dust is redistributed empirically based on the predetermined size distribution [Tegen and Fung, 1994; Mahowald et al., 1999], and (ii) the complex category, termed the complete microphysical parameterization scheme, in which the complete microphysical specification of the erodible environment is used to predict the size-resolved saltation mass flux and resulting sandblasting dust [Marticorena et al., 1997; Alfaro and Gomes, 2001; Shao, 2001]. Even though the later scheme provides realistic estimations at a regional scale, a great deal of the input including the wind speed, soil moisture, surface texture, soil mineralogy, aerodynamic roughness length, smooth roughness length, and kinetic energy of the soil aggregates is required to perform these simulations. Obviously, the data sets estimated from remote sensing satellite measurements and weather forecast prediction models are highly recommended for the framework of dust emission, due to their enhanced temporal resolution and spatial coverage, even though the large uncertainty related to them would remain in the model results. Remote sensing has long been recognized to have high potential for measuring soil characteristics such as vegetation indices and soil moisture at large temporal and spatial scales. The vegetation indices are mainly retrieved from visible and near-infrared light reflections collected by several satellites (MODIS: Moderate Resolution Imaging Spectroradiometer; ASTER: Advanced Spaceborne Thermal Emission and Reflection Radiometer; and MERIS: Medium Resolution Imaging Spectrometer). These data have been considered for several applications such as the monitoring of the phenology/quantity of the crops, forest monitoring, and dust emission modeling. As for the soil moisture, a broad variety of passive sensors on board various polar-orbiting satellites have been launched to retrieve this parameter from the L band (SMOS: Soil Moisture and Ocean Salinity and SMAP: Soil Moisture Active/Passive) or C band (AMSR-E: Advanced Microwave Scanning Radiometer-EOS) microwave brightness temperature (TB) twice a day (6 A.M./6 P.M. or 1:30 A.M./1:30 P.M.) [Njoku et al., 2003; Kerr, 2007; Entekhabi et al., 2010]. Its retrieval method is physically based and accounts for most of the major components in radiative transfer theory, where the measured TB is a function of soil moisture, and other soil/vegetation parameters such as soil texture, surface temperature, vegetation opacity, vegetation albedo, and roughness length [Njoku et al., 2003; De Jau et al., 2008; Owe et al., 2008]. However, these satellites-based estimates are less considered for the dust emission applications [Parajuli et al., 2013; Kim and Choi, 2015]. In fact, for dust emission modeling, the soil moisture is generally estimated, either statistically by establishing a relationship between the soil moisture and meteorological parameters [Reichle, 2002; Zhang and Qui, 2004] or based on the equilibrium equation of soil moisture for hydrostatic equation [Zhang et al., 2006; Shang et al., 2007]. Such efforts to estimate the soil moisture are generally limited to the local area. Conversely, the global atmospheric reanalysis data sets, simulated using numerical weather prediction models and assimilated with archives of in situ and satellite observations, are widely used in several applications (meteorology, oceanography, climatology, and weather/atmospheric composition modeling). The ECMWF (European Centre for Medium-Range Weather Forecasting [Uppala et al., 2005]) ERA-Interim and NCEP (National Center for Environmental Prediction [Kalnay et al., 1996]) reanalysis data sets have been widely used for such purposes. The accuracy of these data sets have been improved by assimilating data from many sources of observations, which is generally higher at global scale than at local scale, because they are usually coarse gridded at approximately 1° [Uppala et al., 2005; Simmons et al., 2007]. There are many previous studies on the comparison between both the ERA-Interim and NCEP meteorological data sets [Molero et al., 2005; Lejiang et al., 2010; Jakobson et al., 2012]. Specifically, the impact of the 10 m wind speed on two dust emission parameterization schemes of Marticorena and Bergametti [1995] and Alfaro and Gomes [2001] were examined over the MENA region [Menut, 2008]. The results obtained from this study prove that both data sets are in good agreement, although the ERA-Interim wind values are higher than the NCEP values. A recent study has shown that the ERA-Interim wind speed is better correlated with MODIS Deep Blue aerosol optical depth than the NCEP wind speed over Bodele, Chad [Parajuli et al., 2014]. This work is an attempt to understand the effect of the soil moisture and soil texture on the dust emission over the MENA region. To reach this objective, the dust emission parameterization scheme proposed by Alfaro and Gomes [2001] is considered to simulate the sandblasting dust flux based on the remotely sensed and forecast data sets, such as SMOS soil moisture, ERA-Interim wind speed at 10 m height, FAO (Food Agricultural Organization) soil texture maps, MODIS Normalized Difference Vegetation Index (NDVI), ERA-Interim surface roughness length, and erodibility of the soil. Since these latter data sets have different temporal resolution, the sandblasting flux simulations were carried out twice a day at 6 A.M. and 6 P.M. local solar time from June 2010 to December 2013. The effect of soil moisture on dust emission has been evaluated by (i) calculating the effect of the soil moisture on the threshold friction velocity, (ii) correlating the near-surface particulate mass concentrations (PM10) observed at four stations in MENA region (Banizombou in Niger, Cinzana in Mali, M'Bour in Senegal, and Liwa in United Arab Emirates)[Marticorena et al., 2010] to the soil moisture, wind speed, and NDVI vegetation index data sets, (iii) analyzing the effect of the temporal evolution of the SMOS soil moisture on the simulated sandblasting fluxes over three locations in the MENA region (Cinzana, Tamanrasset, and Liwa), after validating the fluxes using the satellite-based dust products (MODIS Deep Blue and EUMETSAT MSG dust products) and the available ground-based PM10 measurements, and finally, (iv) performing sensitivity analysis of the simulated sandblasting flux to the soil moisture and soil clay fraction. 2 Dust Emission Parametrization (Saltation and Sandblasting) For the current study, we have used the complete dust emission parameterization scheme, in which the emission process is described as a balance between the kinetic energy provided by the saltation soil grains and the cohesion energy of dust particles. The model is a coupling between the saltation flux developed by Marticorena and Bergametti [1995] and sandblasting theory by Alfaro and Gomes [2001]. A brief description of the model is detailed below. The particle size-dependent (Dp, in meters) threshold friction velocity for smooth and dry surface (u* ts) is parameterized as follows [Shao and Lu, 2000]: (1)where ρp is the soil particle density equal to 2.65 × 103 kg m−3, ρa is equal to 1.12 kg m−3, g is acceleration due to gravity 9.81 m s−2, an is constant approximated to 0.0123, and Γ is constant approximated to 300 kg m−2. The threshold friction velocity of the real surfaces can be obtained by correcting equation 1 for drag partitioning (feff) and the factor accounting for the effect of soil moisture content (fw) as shown below: (2) Following King et al. [2005], the drag partitioning (feff) is as below (3)where z0 is the aeolian roughness length (in meters), the smooth roughness length (z0s, in meters) is calculated based on the median diameter (Dmed) of the coarsest population of the soil size distribution as follows: (4) Though there are many values for x, suggested by several investigators (24 [Schlichting and Gersten, 2000] and 30 [Bagnold, 1941; Marticorena et al., 1997]), we have used the value 10, by following the experimental study by Andreotti et al. [2006]. The term accounting for the gravimetric soil moisture (w, in percent) on threshold friction velocity (wt, in percent) was calculated using the formula proposed by [Fécan et al., 1999] as (5)where the threshold soil moisture is estimated based on the clay fraction (Mclay, in percent) using the following relationship: (6) The gravimetric soil moisture content (w, in percent) is calculated from the volumetric soil moisture (wv, m3 m−3) based on the soil bulk density (ρb,d, kg m−3) and density of the water (ρw = 1000 kg m−3) as follows: (7) The saltation fiction velocity (u*) is calculated based on the wind measurements at height z as follows: (8)Where k is the von Karman constant equal to 0.4 and Us is the wind speed at height z in m/s. Theoretically, the dust mobilization starts when the friction velocity (u*) exceeds the threshold friction velocity (u* t). Therefore, the horizontal saltation flux for each soil size bin Dp is estimated according to White [1979] as (9)where K is a constant equal to 1 for deserts surfaces [Marticorena et al., 1997], V is the vegetation fraction, and E is the soil erodibility factor, added to account for the mismatch between the observed dust sources in satellite images and the modeled dust sources. The total saltation flux is obtained by integrating Fh over the soil size distribution from to . The interval is chosen in order to cover the whole range of possible soil sizes. Thus, the total saltation flux can be written as (10)where Srel(Dp) is the relative surface distribution covered by particles with a mass median diameter Dp. The vertical sandblasting flux is obtained by summing over the three aerosol modes as given below: (11)where Nclass is the number of intervals discrediting the soil size distribution in the entire soil size distribution, dm,i is the mass median diameter of the ith mode, Fv,i is the upward vertical mass flux of aerosols in mode i (kg m−2 s−1), β is proportionality constant equal to 163 m s−2, pi is the fraction of kinetic energy of a soil aggregation required to release aerosol particles of mode i, dm,i is the mass median diameter of the ith mode, and ei is the binding energy of aerosol particles for mode i. Finally, the three mass fluxes are redistributed into 100 aerosol bins from 10−7 m to 10−4 m in diameter, following [Menut et al., 2012]. 3 Description of the Input Data In this paper, long-term remotely sensed and forecasting data sets between June 2010 and December 2013 are considered as input data for the dust emission parameterization scheme described in section 2. The description of these data sets is presented as follows. 3.1 SMOS Soil Moisture The SMOS satellite has the ability to see through the clouds and to provide global coverage of the surface soil moisture measurements (SM), revisiting the equator every 3 days with the morning ascending pass at 06 A.M. local solar time and an afternoon descending pass at 06 P.M. local solar time. The SMOS mission aims to monitor the soil moisture at a depth of ~5 cm with an accuracy of 0.04 m3 m−3 [Kerr, 2007]. The SMOS soil moisture is well validated with ground-based measurements across the globe, except the arid and semiarid areas [Dall'Amico et al., 2012; Gherboudj et al., 2012; Sánchez et al., 2012; Pierdicca et al., 2013]. Despite the use of lower frequency at L band, the first results proved that the validation of the SMOS soil moisture data retrieved from the earlier processing prototypes was site dependent. However, since its launch in late 2009, SMOS products have gained in maturity and all the acquired data were reprocessed with the latest processing prototypes that consider the multiorbit retrievals of the soil moisture estimates at both ascending and descending overpasses to reach the mission target accuracy of 0.04 m3 m−3. The final reprocessed global Level 3 soil moisture product (referred to as SMOSL3) has been recently released by CATDS (Centre Aval de Traitement de Données SMOS). Over the arid and semiarid areas, the accuracy of the SMOSL3 product has been investigated through intercomparisons between SMOSL3 and other satellites (ASCAT: Advanced SCATterometer and AMSRE) or Land Data Assimilation System (ECMWF SM-DAS-2) products [Al-Yaari et al., 2014]. Studies showed that the AMSR-E soil moisture data are in agreement with the gravimetric soil moisture, and the error is within the AMSR-E error range of 0.03 m3 m−3 [Wang et al., 2009; Al-Jassar and Rao, 2010]. However, even though the SMOSL3 and AMSRE soil moisture are well correlated over the arid (Sahara, Arabian Peninsula, South Africa, and central Australia) and semiarid (Indian subcontinent, Great Plains of North America, Sahel, eastern Australia, and southeast of Brazil) areas [Al-Yaari et al., 2014], both SMOSL3 and AMSRE products exhibited weak correlation with SM-DAS-2, especially over the desert regions. However, while good accuracy is attributed to the SM-DAS-2 products, they cannot represent the ground truth. This mismatch would arise from their low variability, which corresponds to the SM accuracy of the targeted satellites [Al-Yaari et al., 2014]. The errors associated to the SMOSL3 might be caused by (1) the uncertainties attributed to the number of multiangular observations, brightness temperature accuracy, and surface conditions (dry or wet soil conditions, dense or sparse vegetation cover, etc.) and/or (2) the radio frequency interference (RFI) generated by the interference of electromagnetic radiations from several sources (satellite transmissions, aircraft communications, TV-radio links, and FM broadcast), which disturbs the natural microwave emission emitted by the Earth surface and is measured by passive microwave systems [Kerr et al., 2012]. Figure 1 presents the spatial pattern of the averaged probability of RFI occurrences (RFIfrac) on SMOS observations between June 2010 and December 2013 over the MENA region for both ascending (6 A.M.) and descending (6 P.M.) overpasses. In general, higher RFI is observed more frequently over the eastern side of the MENA region than the western side. While the averaged probability of RFI occurrences reaches ~ 75% over the Arabian Peninsula desert, it is less than 20% over North Africa. The most significant errors in SMOSL3 data are associated with these effects and are filtered out using the radio frequency interferences (RFI) flag. For the current study, the SMOSL3 values which correspond to the percentage of radio frequency interference greater than 30% are rejected [Al-Yaari et al., 2014]. This process will definitely remove several SMOS soil moisture values, especially over the Middle East region. Figure 1Open in figure viewerPowerPoint Spatial patterns of the averaged probability of RFI occurrences on SMOS observations between June 2010 and December 2013: (a) 6 A.M. and (b) 6 P.M. After filtering out the misestimated values, the mean and standard deviation of the SMOS soil moisture at 6 A.M. and 6 P.M. are calculated based on daily observations between June 2010 and December 2013 (Figure 2). As presented in Figures 2a1 and 2b1, both SMOS overpasses reveal dry soil over the Sahara and Middle East regions, where the average soil moisture is less than 0.1 m3 m−3. Wet regions are detected in the Mediterranean region, banks of the Nile River, southern Sahel region, Iraq/Iran borders, and the eastern part of Pakistan. The average soil moisture over the region is 0.14 m3 m−3 and may occasionally reach the exceptionally high value of 0.3 m3 m−3. This soil wetness is mainly caused by the seasonal rainfall that occurs during the winter (summer) season for the northern region (Sahel region). In addition, a slight increase in the average SMOS soil moisture estimates is observed between 6 A.M. (ascending mode) and 6 P.M. (descending mode). This increase is ~ 0.02 m3 m−3 over the Sahara and western part of the Arabian Peninsula desert and is caused by the day warming, which reduces more the volumetric soil moisture than the night warming. Figure 2Open in figure viewerPowerPoint Mean and standard deviation of the SMOS soil moisture (m3 m−3) between June 2010 and December 2013: (a) 6 A.M. and (b) 6 P.M. The standard deviation of the SMOS soil moisture for both ascending and descending overpasses is displayed in Figures 2a2 and 2b2. As shown in these figures, both overpasses show lower standard deviations ( 0.07 m3 m−3) when the soil is wet. The standard deviation of the soil moisture that varies between 0.04 m3 m−3 and 0.06 m3 m−3 are mainly observed over the semiarid region such as the southern Sahel. In addition, a slight decrease in this parameter is observed between 6 A.M. and 6 P.M. Even though the mean and standard deviation have proven to be extremely useful measures of the average and the spread of the data, the temporal variability in the SMOS soil moisture data is not well described over the year. In fact, it was demonstrated through several research studies that the soil moisture exhibits a high degree of variability in time and space, which is governed by various processes such as rainfall, evaporation, percolation, infiltration, and runoff [Arkley and Ulrich, 1961]. In addition, this variability varies across a range of scales due to the influences of microtopography, soil texture, and vegetation [Famiglietti et al., 2008]. 3.2 ECMWF ERA-Interim Wind Speed The ERA-Interim Re-analysis data on near-surface wind speed (10 m height) are used for the present study [Uppala et al., 2005]. All 6-hourly data (00 UTC, 06 UTC, 12 UTC, and 18 UTC) are considered that cover both SMOS overpasses (i.e., 6 A.M. and 6 P.M. local solar time) and the time shift from the Atlantic Ocean to the Middle East. The choice of this data set to simulate the dust flux is based on (1) huge and continuous improvement of its global model skill over time and (2) its higher spatial resolution (0.25° × 0.25°) compared to NCEP/NCAR Re-analysis data (1° × 1°). The ERA-Interim long-term data set is available for the public [Uppala et al., 2005] and can be downloaded through the website (http://apps.ecmwf.int/datasets/). Previous studies on an intercomparison between the ERA-Interim wind data and ground-based measurements have been undertaken to assess their validity over different regions at different temporal scales [Weller and Anderson, 1998; Bozzano et al., 2004; Ruti et al., 2008]. The obtained results over various oceanic regions indicates that ERA-Interim provides realistic wind speeds against in situ measurements, with underestimation for higher winds and an overestimation for lower winds [Weller and Anderson, 1998; Bozzano et al., 2004; Ruti et al., 2008]. Obviously, the disparity between the observations and ERA-Interim Re-analysis data comes from low spatiotemporal resolution of the reanalysis data. Figure 3 shows the seasonal mean pattern of the synoptic wind vectors derived from the reanalysis data of ERA-Interim at 10 m height with spatial resolution of 1° from 2010 to 2013. The seasonal variations in the wind data (both magnitude and direction) are very distinct over the entire MENA region. Based on the meteorological conditions (wind, temperature, precipitation, etc.) prevailing over the region, the year can be broadly divided into four distinct seasons, namely, the winter (December–February, DJF), spring (March–May), summer
Referência(s)