Small Scale Soil Erosion Susceptibility Modelling in A Protected Mountainous Grassland Using Sentinel-2, Field, and Climate Data

Soil erosion affects about half of the global land area resulting in annual average soil losses of eight to about 50 t/ha. Soil loss leads to challenges in agricultural production, threatens soil health, and ecosystem services, especially in mountainous regions. Ecosystem managers, therefore, must carry out an appropriate evaluation of the amount of soil lost by water erosion for sustainable soil management. Therefore, this study generated a small-scale soil erosion model with site-specific data obtained from field observations, Sentinel-2 multispectral images, 30 meters SRTM, and rainfall data using RUSLE with advanced GIS and remote sensing techniques. The results showed an average yearly soil erosion of 13 t/ha year, with a low to extremely high erosion covering about 21 % of the park. Areas with low vegetation protective cover of less than 40% experienced a high level of erosion, and areas with high topographic factors experienced a very low amount of soil loss. Validation was done using an error matrix between observed field soil erosion and predicted soil erosion resulting in an accuracy of 80%.