Modeling and Assessing Potential Soil Erosion Hazards Using USLE and Wind Erosion Models in Integration with GIS Techniques: Dakhla Oasis, Egypt

Soil erosion modeling is becoming more significant in the development and implementation of soil management and conservation policies. For a better understanding of the geographical distribution of soil erosion, spatial-based models of soil erosion are required. The current study proposed a spatial-based model that integrated geographic information systems (GIS) techniques with both the universal soil loss equation (USLE) model and the Index of Land Susceptibility to Wind Erosion (ILSWE). The proposed Spatial Soil Loss Model (SSLM) was designed to generate the potential soil erosion maps based on water erosion and wind erosion by integrating factors of the USLE and ILSWE models into the GIS environment. Hence, the main objective of this study is to predict, quantify, and assess the soil erosion hazards using the SSLM in the Dakhla Oasis as a case study. The water soil loss values were computed by overlaying the values of five factors: the rainfall factor (R-Factor), soil erodibility (K-Factor), topography (LS-Factor), crop types (C-Factor), and conservation practice (P-Factor). The severity of wind-driven soil loss was calculated by overlaying the values of five factors: climatic erosivity (CE-Factor), soil erodibility (E-Factor), soil crust (SCFactor), vegetation cover (VC-Factor), and surface roughness (SR-Factor). The proposed model was statistically validated by comparing its outputs to the results of USLE and ILSWE models. Soil loss values based on USLE and SSLM varied from 0.26 to 3.51 t ha−1 yr−1 with an average of 1.30 t ha−1 yr−1 and from 0.26 to 3.09 t ha−1 yr−1 with a mean of 1.33 t ha−1 yr−1, respectively. As a result, and according to the assessment of both the USLE and the SSLM, one soil erosion class, the very low class (<6.7 t ha−1 yr−1), has been reported to be the prevalent erosion class in the study area. These findings indicate that the Dakhla Oasis is slightly eroded and more tolerable against water erosion factors under current management conditions. Furthermore, the study area was classified into four classes of wind erosion severity: very slight, slight, moderate, and high, representing 1.0%, 25.2%, 41.5%, and 32.3% of the total study area, respectively, based on the ILSWE model and 0.9%, 25.4%, 43.9%, and 29.9%, respectively, according to the SSLM. Consequently, the Dakhla Oasis is qualified as a promising area for sustainable agriculture when appropriate management is applied. The USLE and ILSWE model rates had a strong positive correlation (r = 0.97 and 0.98, respectively), with the SSLM rates, as well as a strong relationship based on the average linear regression (R2 = 0.94 and 0.97, respectively). The present study is an attempt to adopt a spatial-based model to compute and map the potential soil erosion. It also pointed out that designing soil erosion spatial models using available data sources and the integration of USLE and ILSWE with GIS techniques is a viable option for calculating soil loss rates. Therefore, the proposed soil erosion spatial model is fit for calculating and assessing soil loss rates under this study and is valid for use in other studies under arid regions with the same conditions. Agriculture 2021, 11, 1124. https://doi.org/10.3390/agriculture11111124 https://www.mdpi.com/journal/agriculture Agriculture 2021, 11, 1124 2 of 29

the world, including Peru, Tunisia, Niger, and others. The Dakhla Oasis has agricultural expansion potential due to high groundwater quantities and soil fertility, and it is the main oasis that supports a large human population [34]. Land degradation is a common serious issue in agricultural production, and many methodologies are available for appreciating soil development extensions. The quantification and statistical analysis of soil degradation assessments depend on a variety of soil characteristics. It requires different data forms and information on soil properties [35].
Quantitative and spatial assessments of environmental hazards are needed because the soil erosion, salinity, and sodicity risks have become serious problems in some areas of Egypt, where soil quality degradation has increased. With wide spatial diversity and data limitations, the USLE equation is used for predicting and calculating soil loss. The GIS database and analysis are critical for observing and assessing soil environmental hazards in the study area [36]. In the geographic information systems (GIS) modeling environment, the proposed spatial model was designed and developed for soil erosion assessment based on the USLE factors' rates and land characteristics. Therefore, by overlaying thematic layers, ratings, and weighing soil parameters' data, action can be taken to achieve more effective land use planning and management [37,38]. GIS is a modern technique for processing, displaying, analyzing, and storing large amounts of spatial data in a variety of formats and scales. Therefore, it provides accessible assessment, quantification, and performance of soil information. A GIS analysis provides continuous spatial datasets, including spatially referenced modeling, and covers large areas [39]. In Egypt's desert oases, soil erosion is a severe environmental hazard, with some areas vulnerable to catastrophic erosion loss if favorable conditions exist. This is due to its arid climate, scarce vegetation cover, bare lands, soil types, topographic features, and lack of soil and water conservation practices. The proposed model can be validated by comparing its outcomes to those of the recommended validated models. In the current study, this was accomplished by comparing soil erosion rates calculated using USLE and wind erosion models (validation-models) to rates calculated using the proposed model. So, the main objective of the present work was to assess the performance of a proposed spatial-based model designed based on USLE and wind erosion models in predicting, quantifying, and assessing the soil erosion hazards in Dakhla Oasis, Egypt.

Study Area Description
The Dakhla Oasis is one of seven oases in Egypt's Western Desert. It extends 80 km from east to west and 25 km from north to south and is located between 28 • 14 00" and 30 • 00 00" E and 25 • 09 00" and 25 • 58 00" N ( Figure 1). The study area of the Dakhla Oasis is made up of several communities connected by a string of sub oases, and it covers 4952.15 km 2 (495,215 ha). Sediments from the Upper Cretaceous to the Lower Eocene and Quaternary eras cover the study area [40].
The Dakhla Oasis has a hot desert climate, where "Hyperthermic" is the soil temperature regime, and "Torric" is the moisture regime [41]. The National Oceanic and Atmospheric Administration (NOAA) [42] was the source of climate data for the study area, as shown in Table 1. The floor of the Dakhla depression as a physiographic unit represents arable lands for agriculture and orchards in the study area. According to Soil Taxonomy (2014), the dominant soil orders in the study area are Aridisols, Inceptisols, and Entisols [43,44].
Advanced remote sensing techniques and DEM interpretation methods were used to identify physiographic units using digital satellite images and field studies that provide the ground observation reality. The study area was categorized into eight physiographic units: (1) depression, (2) Palaya basins, (3) mesas, (4) residual hills, (5) rocky lands, (6) sand dunes, (7) sand sheets, and (8) waterbodies. These physiographic units are illustrated in Figure 2 and their characteristics are shown in Table 2.

Digital Images Processing
Geometrically, the Landsat-8 image (Path 177, Row 42, and year 2020) of the investigated area and Shuttle Radar Topography Mission (SRTM) data were used to produce the digital elevation model (DEM, 90 m resolution) of the studied area. The image was draped over the DEM using the ENVI 5.1 software to identify and delineate the map of physiographic units [45]. The study area maps of physiographic features, soil erosion factors, and soil loss rates were created using ArcGIS 10.2.2 software [46].  tem. = temperature.

Field Survey and Laboratory Analyses
Twenty-five geo-referenced soil profiles using the Global Position System (GPS) w chosen based on a physiographical map ( Figure 2). Soil profiles were dug and descri according to the Soil Science Division Staff [47]. Following the standard Soil Survey S methods [48], soil samples were analyzed in the laboratory for particle size distribu (sand, silt, and clay), using the pipette method, soil organic matter content (OM Walkley-Black method, and calcium carbonate content employing the Calcim Scheibler's method.

The Universal Soil Loss Equation (USLE) Model
Wischmeier and Smith [13] proposed the USLE, which has been widely used, still is used in many areas of the world. The USLE is an empirical-based model desig to predict, quantify, and assess the annual average soil loss rate for a given area un particular management conditions. This equation is calculated based on certain study

Digital Images Processing
Geometrically, the Landsat-8 image (Path 177, Row 42, and year 2020) of the investigated area and Shuttle Radar Topography Mission (SRTM) data were used to produce the digital elevation model (DEM, 90 m resolution) of the studied area. The image was draped over the DEM using the ENVI 5.1 software to identify and delineate the map of physiographic units [45]. The study area maps of physiographic features, soil erosion factors, and soil loss rates were created using ArcGIS 10.2.2 software [46].

Field Survey and Laboratory Analyses
Twenty-five geo-referenced soil profiles using the Global Position System (GPS) were chosen based on a physiographical map ( Figure 2). Soil profiles were dug and described according to the Soil Science Division Staff [47]. Following the standard Soil Survey Staff methods [48], soil samples were analyzed in the laboratory for particle size distribution (sand, silt, and clay), using the pipette method, soil organic matter content (OM) by Walkley-Black method, and calcium carbonate content employing the Calcimeter Scheibler's method.

The Universal Soil Loss Equation (USLE) Model
Wischmeier and Smith [13] proposed the USLE, which has been widely used, and still is used in many areas of the world. The USLE is an empirical-based model designed to predict, quantify, and assess the annual average soil loss rate for a given area under particular management conditions. This equation is calculated based on certain study area factors, such as slope, rainfall erosivity, soil type, topography elements, crop system, and management practices. These factors are a numerical representation of the particular conditions that influence the soil erosion risk in the region under study [13]. According to Robert and Hilborn [49], soil erosion values and rates reveal that soil factors can largely be  Tables 3 and 4 and Figure 3. The soil loss rate is expressed in the USLE model as follows: where A is the predicted annual mean soil loss (t ha −1 yr −1 ), R is the rainfall erosivity factor (MJ mm ha −1 h −1 yr −1 ), K is the soil erodibility factor (t ha h ha −1 MJ −1 mm −1 ), LS is the topographic factor (dimensionless), C is the land cover and management factor (crop types and tillage) (dimensionless), and P is the conservation practice factor (dimensionless). Table 3. Management strategies for reducing soil erosion.

R
The R factor for a field cannot be change.
The K factor for a field cannot be change. -

LS
Terraces may be constructed to reduce the slope length resulting in lower soil losses.
Terracing requires additional investment and will cause some inconvenience in farming. Investigate other soil conservation practices first.

C
Crop types and tillage methods that result in the lowest possible C factor will result in less soil erosion.
Cropping systems that will provide maximum protection for the soil. Use minimum tillage systems where possible.
P Support practice that has the lowest possible factor associated with it will result in lower soil losses.
Use support practices such as cross-slope farming that will cause deposition of sediment to occur close to the source. R = rainfall erosivity factor; K = soil erodibility factor; LS = topographic factor; C = land cover and management factor; P = conservation practice factor. The R-Factor quantifies soil erosivity based on the aggressiveness of runoff sediment deposits in a specific area over a given time period. Therefore, the rainfall erosivity factor represents the amount and rate of runoff caused by precipitation events [50]. Based on the annual maximum precipitation in 24 h during the study period, the R-Factor was calculated using the empirical Equation (2) developed by Chaudhry and Nayak [51].
where R a is the rainfall erosivity factor (MJ mm ha −1 h −1 yr −1 ) and X a is the annual precipitation average (mm). The R-Factor was estimated using the climatological data from the NOAA [42]. The R-Factor map was digitized and interpolated using the point-to-raster interpolation tool in the ArcGIS software (geostatistic analysis).

Rainfall Erosivity (R-Factor)
The R-Factor quantifies soil erosivity based on the aggressiveness of runoff sediment deposits in a specific area over a given time period. Therefore, the rainfall erosivity factor represents the amount and rate of runoff caused by precipitation events [50]. Based on the annual maximum precipitation in 24 h during the study period, the R-Factor was calculated using the empirical Equation (2) developed by Chaudhry and Nayak [51].
= 79 + (0.363 * )) + 89.7 (2) where Ra is the rainfall erosivity factor (MJ mm ha −1 h −1 yr −1 ) and Xa is the annual precipitation average (mm). The R-Factor was estimated using the climatological data from the NOAA [42]. The R-Factor map was digitized and interpolated using the point-to-raster interpolation tool in the ArcGIS software (geostatistic analysis).

Soil Erodibility (K-Factor)
One of the most significant barriers for soil erosion modeling on large spatial scales is the insufficient data on soil attributes [52,53]. The K-Factor describes the predisposition of soil particles to be transported and separated by runoff. The K-Factor is rated primarily on a scale of 0 to 1, where 0 is low predisposition for soil erosion and 1 is highly predisposed. The K-Factor rate values depend on soil texture classes, soil structure stability, soil

Soil Erodibility (K-Factor)
One of the most significant barriers for soil erosion modeling on large spatial scales is the insufficient data on soil attributes [52,53]. The K-Factor describes the predisposition of soil particles to be transported and separated by runoff. The K-Factor is rated primarily on a scale of 0 to 1, where 0 is low predisposition for soil erosion and 1 is highly predisposed. The K-Factor rate values depend on soil texture classes, soil structure stability, soil mineralogy, organic matter content, soil permeability, and infiltration. Wischmeier et al. [54] developed an equation to rate K-Factor using the empirical equation, and the K-Factor values for different soil textures were estimated using Equation (3).
where K is the soil erodibility factor, M is the soil particle size distribution parameter calculated as [very fine sand (%) + silt (%)] × [100 − Clay (%)], a is the soil organic matter (%), b is the soil structure code, and c is the soil profile permeability class. Robert and Hilborn [49] clarified the K-Factor rates for various soil textures, as shown in Table 5.

Slope Length and Slope Gradient (LS-Factor)
The LS-Factor is expressed as the ratio extracted from the DEM image and the computation using GIS-based methods. The LS-Factor was used in the USLE to assess the local topography effect on soil erosion per unit zone. Based on topographic effects on soil erosion and cutting rates, the runoff on longer slopes with the greatest accumulation increases its transport detachment and capacity. The LS-Factor can be used to express the soil erosion ratio as defined by Wischmeier and Smith [13] and expressed as Equation (4).
where L is the slope length (m) and S is the slope gradient (%).
The LS-Factor rates for topography effects were explained by Robert and Hilborn [49] and are shown in Table 6. Table 6. Slope length and slope steepness factors (LS-Factor) data.

Cropping Management (C-Factor)
The C-Factor is used to determine and reflect the impacts of both land cover and land management, on various vegetation types. The vegetation cover factor and the LS-Factor are very sensitive to potential tilled soil erosion values under defined conditions and, thus, differ depending on the adopted soil conservation practices. In this study, the C-Factor was assessed using the Normalized Difference Vegetation Index (NDVI), which is expressed by Equation (5) [55].
where C is crop management (C-Factor), and α and β are the parameters that determine the shape of the curve relating to NDVI. The C-Factor rates for crop management were explained by Robert and Hilborn [49] and are shown in Tables 7 and 8.

Conservation Practice (P-Factor)
The P-Factor is a support practice factor that represents the soil loss ratio of upand down slope tillage. It acts through altering the flow pattern, gradients, or direction of surface runoff with specific support and expressed land practice effects that reduce the amount of runoff and erosion [14]. The P-Factor is an important consideration in the USLE equation model for mapping conservation practices using classified images of land use/land cover and slope. These images include different types of agricultural management support practices, such as up-and-down sloping, cross sloping, strip cropping, contour farming, and terracing. The P-Factor rates for conservation support practices, as represented by Robert and Hilborn [49], are shown in Table 9.

Index of Land Susceptibility to Wind Erosion (ILSWE) Model
Wind erosion is a complex process with numerous factors and indicators, making it difficult to define comprehensive criteria for assessing factors at regional scales. As a result, studies of wind erosion at the regional scale employ a complexity reduction strategy while retaining the main wind erosion factors [56]. Therefore, Fenta et al. [32] developed the Index of Land Susceptibility to Wind Erosion (ILSWE), which was used in this study to estimate and assess soil erosion severity and risks by combining layers describing the major factors: climatic erosivity, soil erodibility, soil crust, vegetation cover, and surface roughness. The ILSWE index outcomes were classified into five classes using the variance minimization classification scheme: very slight (0-2), slight (2-10), moderate (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20), high , and very high (>50) erosion severity. The general methodological framework used to assess ILSWE is described in Equation (6).
where CE is the climatic erosivity factor, EF is the wind-erodible fraction factor; SC is the soil crust factor, VC is the vegetation cover factor, and SR is the surface roughness factor.

Climatic Erosivity (CE-Factor)
Climatic erosivity (CE-Factor) is calculated according to climatic data on output conditions that are instrumental in wind erosion estimating [57]. The CE-Factor was calculated based on FAO [58] and developed in Equation (7) as follows: where W i is the mean monthly wind speed (ms −1 ) at 2 m height in month i, PET i is the potential evapotranspiration (mm) in month I, P i is the precipitation (mm) in month i, and d i is total number of days in month i.

Wind-Erodible Fraction (EF-Factor)
The wind erodible fraction (EF-Factor) describes the relationship between wind erosion and soil properties. The EF-Factor was calculated using the multiple regression equation (Equation (8)) proposed by Fryrear et al. [59] based on some soil properties.
where EF is the wind erodible fraction factor, SA is the sand content, SI is the silt content, CL is the clay content, OM is the organic matter content, and CaCO 3 is the calcium carbonate content.

Soil Crust (SC-Factor)
In arid and semi-arid regions, wind erosion often dominates water erosion. Therefore, soil crust plays a vital role in maintaining soil resources [60]. TheSC-Factor was calculated to estimate wind erosion susceptibility based on soil crust influence, which was regressed based on the clay and organic matter content coefficient of friction [61], and is determined as the following Equation (9):

Vegetation Cover Factor (VC-Factor)
The fractional vegetation cover that affects wind erosion by dissipating wind erosivity and providing an efficient shelter effect is defined as the VC-Factor. It was calculated using Equation (10) based on NDVI values of highly dense vegetation (NDVIv) and bare soil (NDVIs) derived from 2020 Landsat-8 images.

Surface Roughness Factor (SR-Factor)
The SR-Factor was calculated based on the ridge height to ridge spacing ratio according to the digital elevation model (DEM) maximum, minimum, and mean using the focal statics tools within ArcGIS 10.2.2 software as the following equation:

Spatial Soil Loss Model (SSlM) Designing
According to Panagos et al. [52] and Efthimiou [53], a lack of sufficient data on soil characteristics is one of the greatest obstacles for soil erosion modeling on larger spatial scales. The integration of the USLE and the GIS techniques is globally popular for soil erosion prediction and assessment purposes [24][25][26]. In the ArcGIS environment software, the proposed SSLM was designed and implemented [46]. The SSLM was performed by matching tables mathematically with the modeling environment using relational database fields through the multiplication of input rasters. The computed rates of the USLE factors (R, K, LS, C, and P) were spatially distributed using geostatistical analysis techniques (Kriging models). Then, the surface maps of soil erosion factors in a spatial zone were extracted with a unique value parameter [44]. Modeling the land susceptibility to wind erosion implied five steps: (1) selecting the criteria, (2) assigning a rating for each criterion, (3) calculating a weight for each criterion, (4) developing five factors (CE, EF, SC, VC, and SR), and (5) generating the final land susceptibility to wind erosion severity (ILSWE) map using ArcGIS software.
In this model, the final potential soil erosion spatial distribution rates and class overlay maps were created using relational database fields and the mathematical multiplication of input rasters. The thematic maps of the five factors of USLE model (R, K, LS, C, and P) and the five factors of ILSWE model (CE, EF, SC, VC, and SR) were then used to integrate them into a single soil loss rate. The soil erosion map was created by reclassifying the factor rates of the USLE and ILSWE in the GIS environment following interpolation techniques. Spatial analysis processes and overlay processes based on subclass weightage values were performed with the different thematic maps. The final maps resulted from the raster-processed weighted overlay values that provided the value of each subclass. The concept flowcharts of the SSLM for USLE and ILSWE models are shown in Figures 4 and 5, respectively.

Model Validation
Using the average linear regression method, the results of the SSLM and ILSWE were compared to the USLE model results [62]. The correlation coefficient was computed to validate the relationship between the SSLM, ILSWE, and the USLE values. SPSS package software version 22.0 was employed for the statistical analysis.

Study Area Characteristics
The terrain data of the study area showed that land is flat to almost flat in the physiographic configuration, with a slope ranging from 0.1% to 2.2% and a slope length ranging from 60 to 981 m. This physiography was interrupted by residual relief in the form of isolated hills (Inselberg) and waterbodies. There were five soil texture grades: clay, loam, silty clay loam, sandy clay loam, and sandy loam. As organic residue was lacking, the organic matter (OM) content of these soils ranged from 1.02 to 20.4 g kg −1 . The annual average precipitation rate (rainfall rate) was very low (1 mm yr −1 ). Agricultural lands, crops and pastures, rocky outcrops, and bare lands were the land use/land cover in the investigated region, as defined by satellite image classification and the NDVI [63]. Ridge tillage, fall plow, mulch tillage, and no-till systems were the tillage systems employed in the study area. Agricultural management conservation support and land practices included contour farming, strip cropping, cross slopes, and up-and-down slopes. Table 10 shows the support practices employing the land use/land cover and slope-classified images, which include different types of agricultural management practices based on altering flow patterns, gradients, or the direction of surface runoff with specific support and land practice effects. With minor restrictions on the soil quality grade, approximately 80% to 85% of the total study area was identified as suitable soil for agricultural use [64].  Table 11 shows the descriptive statistical data on the rates of the various factors used in the USLE calculation. In addition, the spatial distribution maps of those factors across the study area are illustrated in Figures 4-9. The variability of USLE factors was classified into three categories based on the coefficient of variation (CV): low variation (CV < 15%) for the R-Factor; moderate variation (15 ≤ CV ≤ 35%) for the K-Factor, C1-Factor, and P-Factor; and high variation (CV > 35%) for the C2-Factor and LS-Factor [65]. The USLE model factors' mean values were ranked as follows: R > C > P > K > LS (Table 11).  Table 11 shows the descriptive statistical data on the rates of the various factors used in the USLE calculation. In addition, the spatial distribution maps of those factors across the study area are illustrated in Figures 4-9. The variability of USLE factors was classified into three categories based on the coefficient of variation (CV): low variation (CV < 15%) for the R-Factor; moderate variation (15 ≤ CV ≤ 35%) for the K-Factor, C1-Factor, and P-Factor; and high variation (CV > 35%) for the C2-Factor and LS-Factor [65]. The USLE model factors' mean values were ranked as follows: R > C > P > K > LS (Table 11).

Rainfall Erosivity Factor (R-Factor)
The rainfall erosivity factor (R-Factor) refers to the aggressiveness of runoff sediment deposits due to motion action. The motion action aggressiveness of the runoff sediment deposits in the study area was measured by the rainfall erosivity factor. According to the spatial distribution determined using the kriging interpolation method and the estimated R-Factor values, the R-Factor rates in the study area ranged from 79.361 to 79.365 MJ mm ha -1 h -1 yr -1 with an average of 79.363 MJ mm ha -1 h -1 yr -1 . These results are in agreement with those found by Xu et al. [50] in a study that adapted the RUSLE and GIS to model soil erosion risk in the mountain karst watershed in Guizhou Province, China. The R-Factor spatial distribution map revealed no changes across the study area, indicating that the region received the same low amount of rainfall as shown in Figure 6. Furthermore, as shown in Table 11, the R-Factor values were very close to each other, with the standard deviation, coefficient of variation (%), and standard error all equaling zero. According to these findings, the intensity of the rainfall erosivity factor in the soil loss rate is very low. Soil loss is closely associated with rainfall, notably due to the detachment power and runoff of raindrops striking the soil surface [66]. As a result, soil loss was very low in the study area, which is typical of an arid desert [40,43].  Table 11. Descriptive data on the rates of USLE factors, the USLE model, and SSLM.

Cover Management Factor (C-Factor)
According to soil conservation practices, the C-Factor describes the effect of land cover and management practices on different types of crops and tillage methods. Based

Cover Management Factor (C-Factor)
According to soil conservation practices, the C-Factor describes the effect of land cover and management practices on different types of crops and tillage methods. Based grades), where these zones are classified as barren lands. The low values (blue color grades), on the other hand, were discovered in plain areas in the middle of the study area, which are characterized by the presence of agricultural activities (Figure 9). Concerning the spatial variability of the C2-Factor, it was in contrast to the C1-Factor, where the high values were found in agricultural areas due to the application of tillage methods, particularly fall plowing, while low values were found in barren areas, which were characterized by no-till methods ( Figure 10). Figure 9. Spatial distribution map of the crop-type factor (C1-Factor). Figure 9. Spatial distribution map of the crop-type factor (C1-Factor).

Rainfall Erosivity Factor (R-Factor)
The rainfall erosivity factor (R-Factor) refers to the aggressiveness of runoff sediment deposits due to motion action. The motion action aggressiveness of the runoff sediment deposits in the study area was measured by the rainfall erosivity factor. According to the spatial distribution determined using the kriging interpolation method and the estimated R-Factor values, the R-Factor rates in the study area ranged from 79.361 to 79.365 MJ mm ha −1 h −1 yr −1 with an average of 79.363 MJ mm ha −1 h −1 yr −1 . These results are in agreement with those found by Xu et al. [50] in a study that adapted the RUSLE and GIS to model soil erosion risk in the mountain karst watershed in Guizhou Province, China. The R-Factor spatial distribution map revealed no changes across the study area, indicating that the region received the same low amount of rainfall as shown in Figure 6. Furthermore, as shown in Table 11, the R-Factor values were very close to each other, with the standard deviation, coefficient of variation (%), and standard error all equaling zero. According to these findings, the intensity of the rainfall erosivity factor in the soil loss rate is very low. Soil loss is closely associated with rainfall, notably due to the detachment power and runoff of raindrops striking the soil surface [66]. As a result, soil loss was very low in the study area, which is typical of an arid desert [40,43].

Soil Erodibility Factor (K-Factor)
The K-Factor for the soil profiles was calculated using the average particle size, soil structure stability, soil permeability, and organic matter values. As shown in Figure 7 and Table 11, the soil erodibility values ranged between 0.29 and 0.85 t ha h ha −1 MJ −1 mm −1 with a mean of 0.48 t ha h ha −1 MJ −1 mm −1 . Moreover, with a CV of 32.46%, the K-Factor showed moderate variability across the study area (Table 11). According to the K-Factor spatial variation map, the K-Factor values gradually decreased from the depression's upper outer borders to its floor (Figure 7). The spatial variation indicated that the inherent erodibility of the soil at the upper oasis depression was high (red color grades), while that of the soil at the lower sites was low (blue color grades). This is because the soils on the depression floor are agriculturally active, which correlates with soil structure stability and increased organic matter, resulting in less soil susceptibility to detachment and increased infiltration. These findings are compatible with those reported by Panagos et al. [52], where they studied soil erodibility in Europe based on land use/cover to assess soil erosion. Furthermore, the upper outer zones and uplands soils (yellow and red color grades) of Dakhla Oasis have the highest K-Factor rates because they are barren lands with extremely low organic matter and coarse-textured soils. According to these findings, it was obvious that the soil texture is the most important factor influencing the K-Factor rate in the investigation area.

Slope Length and Slope Gradient Factor (LS-Factor)
The effects of local topography on runoff per unit, cutting rates, and accumulation are explained by the LS-Factor. The topographic factor rates (slope length and slope gradient) in the study area varied from 0.09 in flat areas to 0.37 on steep slopes lands, with an average of 0.14 ( Figure 7). The distribution patterns of the SL-Factor were characterized by high variations, with a CV of 47.08% as presented in Table 11 and Figure 8. The LS-Factor spatial distribution map revealed that the highest LS-Factor values were found in the outer zones of the northern and southern parts (red, orange, yellow colors) of the study area's western half, where the land is steep. Low values were observed in the depression's eastern half and across its floor, where the land is flat in the plain areas ( Figure 8). Moreover, the high LS-Factor values decreased gradually from the outer (steep lands) to the inner (flat lands), which corresponds with the slope gradients in the study area. This means that in a three-dimensional complex terrain with flow convergence and divergence, the upslope contributing area, and not the slope length, controls soil loss [67]. Gelagay and Minale [67] reported that in the narrower steeper slope in the upper part of the watershed, very severe soil loss was observed at a rate that exceeded the tolerable soil loss limit. They attributed this to the steepness of the slope, the dominance of soil types that are inherently less resistant to the eroding power of rainfall, and the lack of supportive practice [67].

Cover Management Factor (C-Factor)
According to soil conservation practices, the C-Factor describes the effect of land cover and management practices on different types of crops and tillage methods. Based on the NDVI index extracted from the Landsat-8 satellite image of the study area, as well as field measurements, the C-Factor is effective in relating to biophysical properties when exported to thematic maps (crop type and tillage method). The crop-type factor (C1-Factor) for different types of crops ranged from 0.42 to 1 with a mean of 0.70, while the tillage system factor (C2-Factor) for no-till and fall plow lands was between 0.25 and 1, with an average of 0.73, as shown in Table 11. Furthermore, with a CV of 31.53%, the C1-Factor showed moderate variability, while the C2-Factor reviled high variability with a CV of 43.95% across the study area. These results were similar to those found by Kouli et al. [55] in a study that was conducted to predict soil erosion using the revised universal soil loss equation (RUSLE) in a GIS framework in Chania, northwestern Crete, Greece. They reported that the vegetation cover factor together with slope steepness and length factors is most sensitive to soil loss [55].
The spatial variability of the C-Factor based on the C1-Factor and according to the C2-Factor is shown in Figures 9 and 10, respectively. The high rates of C1-Factor were found in the study area's southern zones with some in the northern ones (red color grades), where these zones are classified as barren lands. The low values (blue color grades), on the other hand, were discovered in plain areas in the middle of the study area, which are characterized by the presence of agricultural activities (Figure 9). Concerning the spatial variability of the C2-Factor, it was in contrast to the C1-Factor, where the high values were found in agricultural areas due to the application of tillage methods, particularly fall plowing, while low values were found in barren areas, which were characterized by no-till methods ( Figure 10). Agriculture 2021, 11, x FOR PEER REVIEW 19 of 29 Figure 10. Spatial distribution map of the tillage method factor (C2-Factor).

Conservation Practice (P-Factor)
According to the land use/land cover and slope maps, the P-Factor represents upand-down slope tillage through surface runoff altering flow patterns, gradients, or directions with a specific support land practices' effect. The P-Factor values of the investigated area ranged from 0.37 to 1.00 with mean of 0.52 (Table 11). Furthermore, with a CV of 22.80%, the P-Factor was characterized by moderate variability throughout the study area. The selection of appropriate conservation measures such as support practice, contour farming, strip cropping, cross slope, and up-and-down slope, can help to reduce the soil erosion caused by soil loss. The obtained results are consistent with those found by Dabral et al. [14] in a study that used USLE, GIS, and remote sensing to assess soil erosion in a hilly catchment in northeastern India. According to this study, the highest and lowest conservation factor (P) values were found to be 1 and 0.28, respectively. Figure 11 shows a map of the spatial distribution of P-Factor values.

Conservation Practice (P-Factor)
According to the land use/land cover and slope maps, the P-Factor represents up-anddown slope tillage through surface runoff altering flow patterns, gradients, or directions with a specific support land practices' effect. The P-Factor values of the investigated area ranged from 0.37 to 1.00 with mean of 0.52 (Table 11). Furthermore, with a CV of 22.80%, the P-Factor was characterized by moderate variability throughout the study area. The selection of appropriate conservation measures such as support practice, contour farming, strip cropping, cross slope, and up-and-down slope, can help to reduce the soil erosion caused by soil loss. The obtained results are consistent with those found by Dabral et al. [14] in a study that used USLE, GIS, and remote sensing to assess soil erosion in a hilly catchment in northeastern India. According to this study, the highest and lowest conservation factor (P) values were found to be 1 and 0.28, respectively. Figure 11 shows a map of the spatial distribution of P-Factor values.

1.
Based on the USLE model The annual average soil loss due to rill erosion was calculated by overlaying the values of the five factors: R, K, LS, C, and P. Based on the soil characteristics, land use, and topography map of the investigated area, these factors were employed in the USLE model and SSLM. According to USLE model, the soil loss rate varied widely spatially, as shown in Figure 12 and Table 11, with values ranging from 0.26 to 3.51 t ha −1 yr −1 , and an average of 1.30 t ha −1 yr −1 . Furthermore, the soil loss spatial variability was high with a CV of 64.30%. Based on classification standards, the average potential soil erosion rate (1.30 t ha −1 yr −1 ) indicated that the Dakhla Oasis as a whole is very slightly eroded. Therefore, the study area was combined into one soil erosion class, which was very low (<6.7 t ha −1 yr −1 ). Accordingly, the very low soil erosion class is the dominant class in this study area, as shown in Table 12. The prevalence of a very low class was related to land flatness and scarcity of rainfall. According to the spatial variability map (Figure 12), the soil loss rate was high in steeply sloping areas and places with little vegetation cover. Furthermore, the soil loss rates differed among agricultural lands, bare lands, crops and pasture lands, and rocky outcrops lands. As a result, the Dakhla Oasis is qualified as a promising area for sustainable agriculture if appropriate management is applied.

2.
Based on the SSLM The SSLM-based assessment of soil erosion produced results that are very similar to those obtained from the USLE model. According to SSLM, the soil loss rate widely varied spatially, as shown in Figure 13 and Table 11, with values ranging from 0.26 to 3.09 t ha −1 yr −1 , and an average of 1.30 t ha −1 yr −1 . Similar to the USLE model, the soil loss spatial variability was high with a CV of 53.38%. Furthermore, the study area was combined into a single soil erosion class as very low, as the USLE model presented. Figure 13 shows the spatial distribution of the soil loss rates according the SSLM. If the soil loss spatial distribution map based on the USLE model and the soil loss spatial distribution map according to the SSLM are compared, they are nearly congruent. This demonstrated that the proposed spatial-based model (SSLM) is fit and accurate for predicting, quantifying, and mapping soil loss under this study.
Agriculture 2021, 11, x FOR PEER REVIEW 20 of 29 Figure 11. Spatial distribution map of conservation practice factor (P-Factor).

Based on the USLE model
The annual average soil loss due to rill erosion was calculated by overlaying the values of the five factors: R, K, LS, C, and P. Based on the soil characteristics, land use, and topography map of the investigated area, these factors were employed in the USLE model and SSLM. According to USLE model, the soil loss rate varied widely spatially, as shown in Figure 12 and Table 11, with values ranging from 0.26 to 3.51 t ha −1 yr −1 , and an average of 1.30 t ha −1 yr −1 . Furthermore, the soil loss spatial variability was high with a CV of 64.30%. Based on classification standards, the average potential soil erosion rate (1.30 t ha −1 yr −1 ) indicated that the Dakhla Oasis as a whole is very slightly eroded. Therefore, the study area was combined into one soil erosion class, which was very low (<6.7 t ha −1 yr −1 ). Accordingly, the very low soil erosion class is the dominant class in this study area, as shown in Table 12. The prevalence of a very low class was related to land flatness and

Based on the SSLM
The SSLM-based assessment of soil erosion produced results that are very similar to those obtained from the USLE model. According to SSLM, the soil loss rate widely varied spatially, as shown in Figure 13 and Table 11, with values ranging from 0.26 to 3.09 t ha −1 yr −1 , and an average of 1.30 t ha −1 yr −1 . Similar to the USLE model, the soil loss spatial variability was high with a CV of 53.38%. Furthermore, the study area was combined into a single soil erosion class as very low, as the USLE model presented. Figure 13 shows the spatial distribution of the soil loss rates according the SSLM. If the soil loss spatial distribution map based on the USLE model and the soil loss spatial distribution map according to the SSLM are compared, they are nearly congruent. This demonstrated that the proposed spatial-based model (SSLM) is fit and accurate for predicting, quantifying, and mapping soil loss under this study. The spatial distribution of wind erosion hazards according to the ILSWE model is presented in Figure 14. Based on the ILSWE model values, the study area was classified into four classes. The four classes of wind erosion severity were very slight, slight, mod-   Figure 14. Based on the ILSWE model values, the study area was classified into four classes. The four classes of wind erosion severity were very slight, slight, moderate, and high, representing 1.0%, 25.2%, 41.5%, and 32.3% of the total study area, respectively. According to the spatial distribution map (Figure 14), the east and west parts of the oasis, which are characterized by bare land and a scarcity of vegetation, are at high risk of wind erosion (red color). The moderate soil erosion class (orange color) was the dominant class in this study area, as shown in Figure 14. Furthermore, the very slight and slight classes were observed in areas with dense vegetation cover, especially in palm trees and orchard cultivation areas. Accordingly, the Dakhla Oasis requires appropriate management and conservation practice against wind erosion for sustainable agriculture. According to Fenta et al. [32], who developed the index of land susceptibility to wind erosion (ILSWE) and used it to estimate and assess soil erosion severity and risks in the East Africa region, they reported that around 25% of the study area is experiencing risks of soil erosion by wind. Moreover, they stated that wind erosion risks are mainly attributable to poor vegetation cover (VC-Factor) and, hence, high surface roughness (SR-Factor). Table 13 shows the values and classes of wind erosion hazards based on the ILSWE for each soil profile. The wind erosion indexes ranged from 0.04 (very low) to 28.12 (high) with a mean of 11.88 (Table 13).

2.
Based on the SSLM The SSLM-based assessment of soil erosion produced results that are very similar to those obtained from the ILSWE model. Based on the SSLM, wind erosion indexes widely varied spatially, as shown in Figure 15 and Table 13, with values ranging from 0.43 to 27.87, and an average of 12.31 (Table 13). Similar to the ILSWE model, the study area was classified into four classes. The four classes of wind erosion severity were very slight, slight, moderate, and high, representing 0.9%, 25.4%, 43.9%, and 29.9% of the total study area, respectively. If the soil loss spatial distribution map based on the ILSWE model and the soil loss spatial distribution map according to the SSLM are compared, they are nearly congruent. This demonstrated that the proposed spatial-based model (SSLM) is suitable and accurate for predicting, quantifying, and mapping wind erosion under this study. Agriculture 2021, 11, x FOR PEER REVIEW 23 of 29 Figure 14. Spatial distribution map of wind erosion indexes and classes according to the ILSWE model.

Based on the SSLM
The SSLM-based assessment of soil erosion produced results that are very similar to those obtained from the ILSWE model. Based on the SSLM, wind erosion indexes widely varied spatially, as shown in Figure 15 and Table 13, with values ranging from 0.43 to 27.87, and an average of 12.31 (Table 13). Similar to the ILSWE model, the study area was classified into four classes. The four classes of wind erosion severity were very slight, slight, moderate, and high, representing 0.9%, 25.4%, 43.9%, and 29.9% of the total study area, respectively. If the soil loss spatial distribution map based on the ILSWE model and the soil loss spatial distribution map according to the SSLM are compared, they are nearly congruent. This demonstrated that the proposed spatial-based model (SSLM) is suitable and accurate for predicting, quantifying, and mapping wind erosion under this study. Overall, many studies have concluded that different land use/land cover have varied impacts on soil erosion [68][69][70]. There are four land use/land cover types in the study area, and in the following order of total area: barren land (82.7%) > agricultural land (13.4%) > sand dunes (3.1%) > built-up land (0.7%), considering the Dakhla Oasis in 2018 [56]. The soil types in the study area with high clay concentrations in the top layers may increase the water runoff and, consequently, the erodibility [70,71]. Due to the very low annual  Overall, many studies have concluded that different land use/land cover have varied impacts on soil erosion [68][69][70]. There are four land use/land cover types in the study area, and in the following order of total area: barren land (82.7%) > agricultural land (13.4%) > sand dunes (3.1%) > built-up land (0.7%), considering the Dakhla Oasis in 2018 [56]. The soil types in the study area with high clay concentrations in the top layers may increase the water runoff and, consequently, the erodibility [70,71]. Due to the very low annual precipitation in the study area, the effective soil erosion factors are the K-Factor, C-Factor, and P-Factor, which are associated with soil characteristics and management practices [70]. Furthermore, very low soil erosion risk intensity was most prominent, indicating that there is currently no erosion hazard due to water erosion, while wind erosion ranged from very slight to high across the study area. Although anthropogenic activities may be important locally, the main causes of soil erosion spatial variability in the study area might be associated with natural factors such as soil types and land cover. According to the outputs of the different models, the study area is slightly eroded due to water erosion and moderately eroded by wind.

Model Validation
The geometric mean algorithm of the five factors was used to calculate soil erosion, and statistical calibration was performed using the average linear regression (ALR) method. The ALR value, which represented the R 2 , was 0.94 and 0.97 for the USLE and ILSWE models with the SSLM, respectively (Figures 16 and 17). These values indicate a very high correlation between the estimated erosion rates based on both the USLE and ILSWE models and the proposed model (SSLM). The outputs of the soil erosion according to the USLE model and the SSLM, as well as the ILSWE model and the SSLM, were close to each other (Tables 12 and 13) and had a strong positive correlation (r = 0.97 and 0.98, respectively). These results revealed a strong relationship between the USLE model and the SSLM and between the ILSWE model and the SSLM, which indicated that the proposed spatial-based model (SSLM) is valid to quantify the soil loss rate. Furthermore, the high congruence between the outputs and maps of the spatial distribution of the USLE and the ILSWE models with the SSLM indicated that the proposed spatial-based model (SSLM) has a high accuracy for predicting and calculating soil erosion rates. Policy planners can utilize the predicted soil erosion maps, because they provide a useful guide for future sustainable landscape development and natural resource management.
high congruence between the outputs and maps of the spatial distribution of the USLE and the ILSWE models with the SSLM indicated that the proposed spatial-based model (SSLM) has a high accuracy for predicting and calculating soil erosion rates. Policy planners can utilize the predicted soil erosion maps, because they provide a useful guide for future sustainable landscape development and natural resource management.

Conclusions
The present study is an attempt to adopt a spatial-based model to compute and map potential soil erosion. It proposed a spatial-based model to quantify the soil loss rate and to identify and map the spatial distribution of soil erosion in the Dakhla Oasis, Egypt. The proposed spatial-based model (SSLM) was designed by integrating GIS analyst environment with the universal soil loss equation (USLE) and Index of Land Susceptibility to Wind Erosion (ILSWE) models. The annual average of soil loss due to rill erosion was high congruence between the outputs and maps of the spatial distribution of the USLE and the ILSWE models with the SSLM indicated that the proposed spatial-based model (SSLM) has a high accuracy for predicting and calculating soil erosion rates. Policy planners can utilize the predicted soil erosion maps, because they provide a useful guide for future sustainable landscape development and natural resource management.

Conclusions
The present study is an attempt to adopt a spatial-based model to compute and map potential soil erosion. It proposed a spatial-based model to quantify the soil loss rate and to identify and map the spatial distribution of soil erosion in the Dakhla Oasis, Egypt. The proposed spatial-based model (SSLM) was designed by integrating GIS analyst environment with the universal soil loss equation (USLE) and Index of Land Susceptibility to Wind Erosion (ILSWE) models. The annual average of soil loss due to rill erosion was

Conclusions
The present study is an attempt to adopt a spatial-based model to compute and map potential soil erosion. It proposed a spatial-based model to quantify the soil loss rate and to identify and map the spatial distribution of soil erosion in the Dakhla Oasis, Egypt. The proposed spatial-based model (SSLM) was designed by integrating GIS analyst environment with the universal soil loss equation (USLE) and Index of Land Susceptibility to Wind Erosion (ILSWE) models. The annual average of soil loss due to rill erosion was calculated using the USLE model and SSLM by overlaying the five factors' values (R, K, LS, C, and P). The outputs of the soil loss according to the USLE model and the SSLM were close to each other. Soil loss values based on the USLE model and the SSLM varied from 0.26 to 3.51 t ha −1 yr −1 with an average of 1.30 t ha −1 yr −1 and from 0.26 to 3.09 t ha −1 yr −1 with a mean of 1.33 t ha −1 yr −1 , respectively. Therefore, and according to the assessment of both the USLE and the SSLM, one soil erosion class, the very low class (<6.7 t ha −1 yr −1 ), was reported to be the prevalent erosion class in the study area. These findings indicate that the Dakhla Oasis is slightly eroded and more tolerable against soil erosion factors under current management conditions. Therefore, based on a vulnerability assessment of water erosion in the study area, the very low soil erosion class was the most prevalent in the study area due to the land flatness and rainfall scarcity.
Furthermore, the study area was classified into four classes of wind erosion severityvery slight, slight, moderate, and high-representing 1.0%, 25.2%, 41.5%, and 32.3% of the total study area, respectively, based on the ILSWE model, and 0.9%, 25.4%, 43.9% and 29.9%, respectively, according to SSLM. The moderate wind erosion class was the dominant class in this study area. Moreover, very slightly and slightly classes were observed in areas with dense vegetation cover, especially in palm trees and orchard cultivation areas. Accordingly, the Dakhla Oasis requires appropriate management and conservation practices against wind erosion for sustainable agriculture. According to the outputs of the different models, the study area is slightly eroded due to water erosion and moderately eroded by wind. Consequently, the Dakhla Oasis is qualified as a promising area for sustainable agriculture when appropriate management is applied. Crop types, soil types, drainage patterns, and land barrenness are the most important factors of soil erosion in the study area, though anthropogenic activities may play a role locally.
Comparing the findings obtained from both the USLE and ILSWE models to those of the SSLM, there were strong positive correlations (r = 0.97 and 0.98, respectively), as well as strong relationships based on the average linear regression (R 2 = 0.94 and 0.97, respectively). Furthermore, the high congruence between the outputs and maps of the spatial distribution of the USLE and ILSWE models and those of the SSLM indicated that the proposed spatial-based model (SSLM) has a high accuracy for predicting and calculating soil erosion rates. This study also pointed out that designing soil erosion spatial models using available data sources and the integration of the USLE and ILSWE models with GIS techniques is a viable option for calculating soil loss rates and classes. Finally, the proposed soil erosion spatial model is suitable for calculating and assessing soil loss rates under this study and is valid for use in other studies under similar arid regions.