Soil erosion-induced watershed prioritization in the beas river sub-basin of the western himalayas: A multi-criteria decision-making and XAI-Based approach

Beas River Sub-Basin is experiencing severe soil erosion due to extreme hydro-climatic events and anthropogenic activities. Therefore, prioritizing erosion-prone watersheds is essential for effective soil resource management. This study proposes an integrated approach combining morphometric analysis using weighted compound factor (WCF) and soil erosion susceptibility modelling (SESM) through fuzzy analytical hierarchy process (FAHP). The results from both analyses were integrated using fuzzy overlay to derive watershed priority rankings, from highest (rank 1) to lowest (rank 20). Watersheds A01BEA09, A01BEA14 and A01BEA19, located in the middle sub-basin, showed the highest susceptibility and were ranked 1st, 2nd and 3rd, respectively. In contrast, watersheds A01BEA01 and A01BEA02 in the eastern region showed the lowest susceptibility and were ranked 20th and 19th. The results also show that 7141 km² (57.7 %) of the northern and northwestern sub-basin is highly susceptible to soil erosion, while 4238 km² (33.6 %) and 1219.55 km² (9.6 %) show moderate and slight susceptibility, respectively. The SHapley Additive exPlanations (SHAP) analysis identified land use land cover (LULC), slope, rainfall and sediment transport index (STI) as the most influential parameters. Sensitivity analysis confirmed LULC as the dominant factor, followed by rainfall, slope and STI. Excluding LULC reduced the model's accuracy from 0.843 to 0.628. The findings highlight the critical role of LULC management in mitigating soil erosion, which threatens agricultural productivity, accelerates reservoir sedimentation, and endangers food and water security. This study provides valuable insights for stakeholders and policymakers to develop targeted strategies and implement watershed management frameworks in high-risk areas.