A RUSLE-based Comprehensive Strategy to Assess Soil Erosion in a Riverine Country, Bangladesh

Recognizing soil as a vital resource for food production and animal habitat, this study employed a comprehensive, nationwide erosion assessment in Bangladesh using the Revised Universal Soil Loss Equation (RUSLE) model. Rainfall, soil data, the Digital Elevation Model (DEM), Landsat 8 imagery, and land use and land cover (LULC) maps were employed as inputs for the prediction of potential soil erosion. Utilizing the zonal statistics tool within a geographic information system (GIS) platform, the study area's most erosion-vulnerable zones were subsequently identified and mapped. These zones were designated as 'erosion hotspots' (EHs) with demonstrably clear connections to river basins and networks. Analysis revealed an annual average erosion rate of 1.3 t ha-1 yr-1, resulting in an alarming loss of 17.68 million tons of topsoil annually. Notably, four distinct EHs were identified within the study area. The southeast region, due to its specific topography and climate, was found to be highly susceptible to erosion. Within this region, 2.22% (2995.93 km2) experience erosion exceeding 10.0 t ha-1 yr-1. RUSLE-based results were validated by comparison with outputs from the Global Soil Erosion Modeling Platform (GloSEM) and Google Earth imagery. In addition, the model's performance was evaluated using the receiver operating characteristic (ROC) curve and area under the ROC curve (AUC). The high AUC value of 0.94 achieved in this study reinforces the accuracy and confidence of the soil erosion estimations.