Evaluating Soil Loss for Identification of Land Risk Area in the Kabe Watershed of Ethiopia

Soil erosion is the main cause of topsoil loss in farming land, which results in reduction in cropland productivity. Soil loss estimation is crucial for developing soil and water conservation strategies for Ethiopia. Te investigation aimed to estimate the soil loss in various intensifcations of land use patterns, including slope categories, using the soil estimation model associated with the ArcGIS process. It is analyzed in Ethiopian conditions based on erosivity, soil erodibility, vegetative cover ( C ) remote sensing data, slop-length factor (LS), and management practices ( P ). Te mean soil loss was relatively high (20.01t ha − 1 yr − 1 ) in the cultivated land, whereas it was lowest (0.17t ha − 1 yr − 1 ) under forest land. Soil loss in the watershed shows a signifcant variation under slope classifcation. Moreover, the land having a greater slope angle, or the upper slope of the watershed, contains maximum soil erosion, while the lower slope position has a minimum soil erosion rate. Te validation shows that there is an insignifcant variation between the predicted model and the experimental data. Terefore, this confrms that the model can be applied in the study watershed or elsewhere with similar agroecology to the study area. Tis research is also used to prepare an erosion management strategy for the conservation of soil and water in the watersheds.


Introduction
Soil erosion has various efects on the environment, society, and economy [1,2] since it removes fertile topsoil, which reduces the productivity of the crop feld, and fnally, it is the source of food production loss [3]. Te sediment transported in waterbodies could be the cause of the decline in water quality and freshwater bodies [4,5]. Heavy metals, contaminants, and chemicals that are generated from erosion in a landscape are transported with soil particles, causing higher sediment levels which eventually lead to water eutrophication and disturbance of delicate aquatic ecosystems [6]. Te excessive silt export caused by severe soil erosion that is deposited in water bodies results in disturbances of life in the water bodies and a decline in the quality of the water bodies [7]. Soil erosion is recognized as a serious threat to agricultural land's ability to operate sustainably since soil erosion can decrease the productivity and production of agricultural land by reducing soil nutrients and soil fertility [8][9][10][11]. Moreover, when the eroded soil reaches the water bodies, it can cause eutrophication, in which poisonous and injurious ingredients build up and decrease liquifed oxygen, which afects the hydrological ecosystems and biodiversity.
Te study conducted by the Global Soil Partnership (GSP) indicated that the rate of soil loss was greater than 75 billion t yr −1 [9]. Moreover, the economic cost of annual soil loss associated with crop felds is approximately US$400 billion around the globe [10]. Te annual assessment of soil loss varied around the world because of environmental and socioeconomic factors. For example, the annual erosion rate of soil in the US was 16 t ha −1 yr −1 , and in Africa, Asia, and South America, it ranged from 20 to 40 t ha −1 yr −1 [14,15]. In India and Syria, the erosion of soil in a year was 16.4 t ha −1 yr −1 and 5 t ha −1 yr −1 , respectively. According to Das et al. [16], the annual soil loss predicted by the RUSLE model in Arunachal Pradesh, India, was 1.38-59.05 t ha −1 yr −1 , whereas the soil loss in some watersheds of Ethiopia was 42 t ha −1 yr −1 [15] and 43 t ha −1 yr −1 in the upper Omo Gibe Basin of Ethiopia [17].
Many research fndings on soil erosion show that different approaches and methods were followed, such as feld experiments, the InVEST model, the WEPP model, and the RUSLE model, with the support of GIS technology. For example, Aneseyee et al. [17] used the InVEST sedimentary delivery ratio model, and Hussien [18] used the RUSLE model. Each of the models that apply in diferent watersheds has its limitations and drawbacks.
Te global cultivated land was afected by soil erosion siginfcantly [19], which impacts billions of people around the globe, particularly the population of Africa and less developing countiries [20]. In Ethiopia, the rate of soil loss could be greater than 300 t ha −1 yr −1 [8,21], which indicates that Ethiopia is the most afected country by soil erosion on the globe [15]. Te total soil loss is estimated at 1.5 billion t ha −1 yr −1 for the whole country, but agricultural land is the main source of soil erosion [22]. Te study in Ethiopia's highlands indicated that more than two million hectares of land were lost to rehabilitation [22]. Terefore, the management of soil erosion is the key issue for environmental conservation and improving food stability [20,21].
Greater than 85% of the Ethiopian population depends on agriculture, which indicates that agriculture is the backbone of the Ethiopian economy [25]. Agricultural farming provides a massive opportunity to create jobs for the majority of the population; it covers half of the country's GDP and also is the major source of foreign exchange income but farming activities have recorded low yields due to a decline in soil fertility and reduced agricultural feld productivity, which leads to incapable of achieving food selfsufciency [24,22].
To assess soil erosion risk and apply suitable soil and water conservation (SWC) technology on degraded land, several soil loss models have been advanced in recent years. To evaluate the soil loss, GIS and remote sensing data were acquired and signifcantly associated with the biophysical data [23,24]. Te RUSLE model is the well-identifed empirical soil erosion model used throughout the globe [25]. It is estimated soil loss with the input of diferent raster and vector data, even if it has its drawbacks such as the lack of hydrological connectivity and the inability to estimate the sediment export capacity of a given watershed.
Te origin of land degradation in Ethiopia is caused by farming on sloping land, poor practices of SWC measures, erratic patterns of rainfall, the absence of fallow land, a low supply of nutrients to the plant, vegetation, and forest degradation [17,30,31]. Terefore, the mismanagement of land by human activities such as poor cultivating practices and understanding the fuctuation of rainfall are signifcant infuences for defning the concentration and impact of soil loss [32]. Terefore, resource degradation, declining agricultural productivity, aggravating poverty, and food security are major challenges for the country. As a result of these, the struggle could be aimed at preserving the soil resources for maximizing the productivity and production of land, which would lead to improved livelihoods and sustainable use of the ecosystems.
Diferent soil and water conservation (SWC) measures have been introduced and implemented over the last decades by governmental and nongovernmental institutions to increase food production in the country [33]. Te emphasis has been largely on the construction of structural SWC measures in cultivated felds and the aforestation of hillsides to restore degraded land [34]. Conservation measures were opted in watersheds, leading to a decrease in runof and a considerable increase in groundwater recharge [35]. Moreover, the implementation of SWC has been triggered to improve crop production, increase vegetation cover, reduce soil erosion, and improve the food security and livelihoods of rural communities [36].
Regardless of the erosion severity and its efects in the Kabe watershed, there is a lack of studies conducted to compute erosion rates for better management of the land. Te land has a varied sensitivity to erosion based on its slope and land-use types features. Moreover, soil erosion predictions have been undertaken by many researchers at diferent times but their results show signifcantly varied. Terefore, estimating the soil loss rates and expressing the spatial mapping of soil erosion at the Kabe watershed is helpful for the planning of watershed development and for decision-makers. Tis research aims to (1) evaluate the soil loss rate in various patterns of land use systems, (2) explore the soil loss in diferent slope classes, and (3) validate the model to show the applicability and error of the model in the watershed.

Description of the Study Area.
Te research was undertaken in the Kabe watershed, which is part of the Blue Nile Basin of Ethiopia. Te study area is located 470 km from Addis Ababa, the main city of Ethiopia. Kabe watershed has diferent kebeles/villages/and its longitude is located at 39°41′10.713″E to 10°89′14.098″N and the latitude is located at 39°47′8.6279″E to 10°82′35.788″N ( Figure 1). Te elevation ranges are also based at 1428-2752 m above sea level, with a mean annual rainfall of 1130 mm, while the mean minimum and maximum temperatures of the district are 9 and 21°C, respectively. Te main types of crops grown are wheat, fenugreek, barley, and tef [37]. Te main economic activity in the study area was agriculture, which depends on rainfall farming. Moreover, traditional methods used to improve soil fertility, such as the application of farm residue manure and crop rotation, have been abandoned in the area. Te organic sources, such as crop residues, are completely removed from farmlands for animal feed, traditional fueling, and house construction purposes. Cow dung, which is supposed to be used as farm residue manure, is a major source of household energy sources. Crop yields under rainfed conditions are low due to the combined efects of limited input use and poor agronomic practices. Te degradation of the environment, such as soil erosion and nutrient depletion, causes a decline in agricultural production in the study area. Moreover, continuous drought, poverty, and crop failures were the common challenges, all of which in turn triggered a chronic shortage of food. Te study area has diferent topographic features with a wide range of altitude variations (see Figure 1). Consequently, diferent biodiversity exists in the watershed.

Estimation of Soil Erosion at the Watershed.
A RUSLE equation has the capability of estimating soil loss by using erosivity, erodibility, topography, vegetation cover, and conservation practices [38]. Te fve parameters were used to estimate soil erosion on the model, such as erosivity (R), erodibility (K), slope and steepness (LS), crop cover (C), and conservation (P). Te RUSLE has computed the mean erosion rate in diferent land use systems and slope classifcations, as given in the following equation: where A is the eroded soil expressed in tons per hectare per year (t ha −1 yr −1 ), R is rainfall erosivity (MJ , LS is the length of slope and steepness, C is the vegetation cover (dimensionless), and P is conservation practice (dimensionless) ( Figure 2).

Applied and Environmental Soil Science
Calculator (Spatial Analyst) tool of ArcGIS 10.4 [39]. A maximum likelihood classifcation method of supervised classifcation was used to identify the land use patterns of the investigated watersheds. During the feld visit, ground control points (GCPs), which represent the diferent land cover classes, were taken using handheld GPS. Te taking of GCPs was used to sample representative regions of interest (RoI) (signatures) for the diferent land cover types to regulate the accuracy of the image classifcations.

Erosivity (R).
Rainfall erosivity denotes the energy that began with the sheet, then rill erosion, and fnally creates gully erosion. Te erosivity estimates in the rainfall data are a straight raindrop infuence and are partly due to the runof that rainfall produces.
Estimating erosivity is based on Hurni's [40] equations, derived from a spatial analysis regression Helldén [41] adapted for Ethiopia using annual precipitation, but there are many diferent computational techniques to compute erosivity factors in the world.
where R � the annual rainfall erosivity and P � the mean annual precipitation of nearby stations acquired over the last 30 years.
To compute the R factor, a formula based on the average yearly rainfall was used. Interpolation has been performed to show the spatial surface distribution of soil erosion based on average 30 -year (1986-2015) mean annual precipitation data. Based on the Hurni [40], as provided in equation (2), the average rainfall and erosivity of the three stations were 1145 mm and 634 MJ Mm −1 ha −1 yr −1 , respectively ( Figure 3) and Table 1.

Erodibility of Soil (K).
Te erodibility of soil (K) is determined based on the soil type, which is afected by the structure and texture of the soil, organic matter (OM) contents, and soil permeability (see equations (3)- (7)). For this study, the FAO soil map was used to derive the data on soil properties. Te study area has three major soil types. In each soil type, soil properties were studied using standardized laboratory methods by taking 48 soil samples using systematic sampling techniques. In other words, sixteen (16) soil samples were taken from the three soil types based on systematic sampling techniques to analyze the organic carbon matter content and soil textures (silt, loam, and clay). Based on the Norman et al. [42] equation, a fraction of sand, silt, clay, and organic carbon content for the watershed has been taken as 0.37. After establishing the value of the K factor, it was put into the geo-database based on Kouli et al. [43] to create a raster map with a spatial resolution of 30 m cell size (Figure 4). where where SAN, SIL, and CLA are % sand, silt, and clay, respectively; C � the organic carbon content; SN1 � sand content subtracted from 1 and divided by 100; Fcsand � soil erodibility factor for low; Fsicl � soil erodibility factor for high clay to silt ratio; Forgc � factor that reduces soil erodibility for soil with high organic content; Fhisand � factor that reduces soil erodibility for soil with high sand content.

Topographic Factors (LS).
Te slope of the land infuences the velocity and level of runof. In other words, a higher slope triggers a higher velocity of runof, which aggravates soil erosion. Tere are diverse topographic features in the land use system, such as high and low slopes, fat land, and steep slopes. Te slope was classifed into six in the Kabe watershed ( Figure )5. With this data, slope length and steepness factors can be investigated for their efect on soil erosion [44]. DEM from the USGS was important to compute the slope length factors with the help of the ArcGIS environment. According to Moore and Burch [45], the LS factor was analyzed using the following equation: where LS is the collective slope length and steepness factor. DEM was used to develop fow accumulation with a resolution of 30 m and sin of slope (degree). Te LS factor for the Kabe watershed was computed, and it was 4.94 at the maximum and 0 at the minimum value (see Figure 6).

Vegetative Cove Factor (C).
Soil erosion could be diferent depending on rainfall erosivity and the morphology of the plant cover. Te falling rainfall protected by vegetation cover could reduce soil erosion on certain land. Te protecting plants could be crops, weeds, or trees. Diferent stages of crop growth afect the generation of crop management factors and the need for the growth period and year of the plants.
To determine crop management factors (C), data on land use was produced from Landsat images of 30 m resolution (see Table 2). To classify the land use system, GIS and remote sensing applications such as the maximum likelihood classifcation algorithm were carried out on the remote sensing data. Te crop management factor values associated with Ethiopian contexts based on the available land use maps were performed using Hurni [40] and set into a geo-database (see Figure 5).

Conservation Practice (P-Factors).
Conservation practices (P-value) are considered the application of soil conservation practices on the landscape, like terracing, mulching, and gulley control. If no erosion control practice is found in a landscape, then the P-value is equal to one, which indicates that the landscape has a high capability of reducing soil erosion. Te P-value indicates a range between 0 and 1. Ploughing the farmland on high, sloppy land could increase soil erosion instead of reducing it. Terefore, the farming system in the landscape needs to apply diferent SWCs with diferent P factors. According to Hurni [40], the conservation practices (P-factors) values for diferent conservation practices in a land use system were provided (Table 3).
Based on Hurni [40], the management practices collected during feld observation have classifed the watershed as indicated in Table 4, and it has been put into the geodatabase; hence, P-value was analyzed using Arc GIS (see Figure 7).

Model Validation.
Te RUSLE model was computed to compare simulated and observed data. Te observed data was obtained from the Ministry of Water and Energy of

Applied and Environmental Soil Science
Ethiopia. Te unit for model output (simulated) was expressed on an annual basis, i.e., t ha −1 yr −1 , and the observed data described by streamfow and suspended sediment concentration data were expressed on a daily basis by m s −3 and g ml −1 , respectively. Terefore, the unit for the observed and predicted data requires to make a similar unit for consistent analysis. Te observed data were computed using the streamfow and sediment concentration based on the following equation for the gauged stations, as introduced by Sadeghi et al. [46]: where SC (t day −1 ) refers to soil erosion, Q refers to the rate of streamfow (m s −1 ), and b and c are constants, obtain from the analysis of the streamfow and sediment concentration (g ml −1 ) data. Te coefcient of determination (R 2 ), mean Percentage Bias Error (PBIAS), and Residual Root Mean Square (RRMSE) were used to check the model's performance [47,48]. If the statistical value indicates a high value, then the performance of the model becomes very good and applicable to the watershed [48,49].

Result and Discussion
Te model of RUSLE used to estimate soil loss in the study area is provided in Figure 8. Te overall maximum soil erosion assessed was 0 to 125.24 t ha −1 yr −1 because of continuous cultivation on a steep slope, forest cover reduction, loss of organic matter, and the absence of appropriate conservation measures. Moreover, the northwestern and southern parts of the watershed have the highest risk of soil loss due to the lack of modern types of soil conservation structures.   Land cover P-factors Forest land supported by terraces and bunds 0.  Applied and Environmental Soil Science

Efects of Land Use Systems on Soil Erosion.
Te study area has three land-use types ( Figure 9). Te mean soil erosion was relatively high (20.01 t ha −1 yr −1 ) in cultivated land, while it was lowest in forest land (0.17 t ha −1 yr −1 ). Te overall average soil erosion was 6.95 t ha −1 yr −1 for the entire watershed (Table 5). Te study shows a lower average soil erosion of 6.90 t ha −1 yr −1 (Table 5) compared to the tolerable rate of soil erosion (10 t ha −1 yr −1 ) [40], and it also showed a lower tolerable soil loss rate in tropical Africa of (11 t ha −1 yr −1 ) [50]. Te maximum soil loss rate in the watershed was 125.24 t ha −1 yr −1 . Tis is the highest soil loss due to a slope greater than 75% and a high slope length and steepness value. Te forest land soil loss was lower because of the protective ability of the vegetation and the OM added to the soil that makes the soil stick together. However, on cultivated land, soil loss is highest because continuous cultivation of land could be triggered by the loss of organic matter and top fertile soil, which are easily eroded by wind and water. Generally, the simulated erosion of soil and the description of spatial mapping is accurate, as related to other studies conducted in preceding times. For example, Mati et al's. [51] study shows the mean soil loss of Ethiopia's highland was 100 metric t ha −1 yr −1 in cropland. Of course, this is not a similar estimate to our studies. Te soil erosion was enormously high in Ethiopia's highlands, which is a computed mean soil loss of 20 t ha −1 yr −1 [40]. According to Hurni [52], the average soil erosion in the feld of cultivated land was 42 ha −1 yr −1 . Te soil erosion computed annually in the watershed of Medego in Ethiopia was 9.63 t ha −1 yr −1 [53], and the average soil loss in a year for the watershed of Chemoga in the Blue Nile Basin of Ethiopia was 93 t ha −1 yr −1 [54]. Terefore, this fnding indicated that there are inconsistencies in estimating the rate of soil erosion.

Te Efects of Slope on Soil Erosion in the Kabe Watershed.
Te Kabe watershed was classifed into six slope classes ( Figure 10). Te slight place of the study area was on a very high slope (>75%), and most areas were found under (0-15%) gentle slope positions. Nevertheless, the low slope conditions afect average soil erosion insignifcantly (Table 6). Te analysis showed that the average loss from erosion under diferent slope positions is signifcantly diferent. Te analysis showed that the highest (13.71 t ha −1 yr −1 ) soil loss was observed under the upper slope position (Table 6), whereas the smallest soil erosion (1.69 t ha −1 yr −1 ) was found under the lower slope position of the watershed. Similar studies were conducted in the Tigray Region of Ethiopia [55], which showed the maximum soil erosion was found on the upper slope and the minimum soil loss was observed under the lower slope position.

Model Validation in the Kabe Watershed.
Te observed and simulated data have shown an insignifcant variation in soil loss in the Kabe watershed (P < 0.05, Figure 11). Terefore, the model used in this watershed is suitable for estimating soil erosion in the watershed. Te observed mean soil erosion values of the three gauged stations were 7.72 t ha −1 yr −1 , 7.29 t ha −1 yr −1 , and 7.45 t ha −1 yr −1 , respectively, which is reliable with results derived from the existing model.
Te experimental and predicted erosion of soil were 7.49 and 6.95 t ha −1 yr −1 , respectively, with a variation of 0.54 t ha −1 yr −1 . Te very few inconsistencies (error � −3.4%) of soil erosion recommend that land use/cover and other climatic factors have been adequately recognized by the RULSE.     Tus, the performance of the RUSLE model indicates a very good performance based on the statistical analysis (PBIAS � −3.22%, R 2 � 0.86 and RRMSE � 0.84). Terefore, it indicates that the experimental data from the study watershed is a good ft with the RULSE models' predictions.

Conclusion
Te study analysis indicates that there was a signifcant rate of soil loss because of the signifcant dynamics of land use systems, which are contributed by climate variabilities such as increasing temperature and rainfall fuctuation. Te analysis also shows cultivated lands have generated a higher soil erosion rate because the protective capacity of the land becomes low and the absence of forest cover. Moreover, in the vegetation and grazing land, the soil loss declined due to the protective capacity of the vegetation and grassland. Te analysis shows that a higher sloppy area has shown a higher soil loss, whereas a lower soil loss has triggered in the lower slope area. Te RUSLE model in the Kabe watershed predicted a lower rate of average soil erosion compared to the tolerable soil erosion rate estimated for Ethiopia and tropical Africa. Terefore, a watershed with high soil erosion needs to provide urgent interventions to decline soil erosion using conservation strategies, appropriate planning, community participation, and integrated approaches.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.

Authors' Contributions
FA collected, analyzed, and interpreted the data and made the fnal write up. AB wrote and edited the paper and performed GIS analysis; TS and EA edited the fnal manuscript. All authors read and approved the fnal manuscript.