Machine Learning–Based Prioritization of Sub-Watersheds for Soil Erosion Management: A Case Study of the Bardha Watershed

Soil erosion is a major environmental concern that affects land productivity and water quality. Although soil erosion is a serious global environmental challenge, understanding its potential influence in the Bardha Watershed is important due to its topographical characteristics and the dependence of local communities on land resources for agriculture. Morphometric analysis helps assess a watershed’s physical characteristics to understand its erosion potential. In this study, sub-watersheds were delineated using the Shuttle Radar Topography Mission (SRTM) Digital Elevation Model (DEM) to accurately derive drainage and terrain characteristics. To enhance the precision of sub-watershed prioritization, morphometric analysis is combined with multi-criteria decision-making (MCDM) techniques. This research ranks sub-watersheds in the Bardha watershed in Chhattisgarh using morphometric parameters in combination with four MCDM approaches: additive ratio assessment (ARAS), multi-objective optimisation by ratio analysis (MOORA), visekriterijumsko kompromisno rangiranje (VIKOR) and simple additive weighting (SAW). The criteria weights for these MCDM methods are determined using the criteria importance through intercriteria correlation (CRITIC) method. Furthermore, the novelty of this study lies in the integration of machine learning (ML) techniques, specifically support vector machine (SVM) and random forest (RF). By combining the outputs of all six methods, the study developed a unified priority map, which was subsequently classified into high, medium, and low priority zones. The study found that sub-watershed 3 (SW3) and SW4 fall into the common high-priority category; SW2, SW6, and SW7 into the medium category; and SW1 and SW5 into the low-priority group. This integrated method makes decision-making stronger by letting planners focus on high-priority sub-watersheds for strategic development, conservation, and optimal land management. This study aligns with SDG 15 by addressing land degradation through the identification and management of soil erosion-prone areas. • Developed a novel hybrid framework integrating morphometric analysis, VIKOR, MOORA, SAW (MCDM), and CRITIC-based objective weighting for prioritizing erosion-prone sub-watersheds. • Incorporated machine learning models (SVM, RF) to enhance prioritization accuracy high-risk erosion zones. • Provides a scientific decision-support framework for watershed managers to optimize soil conservation efforts in the Bardha Basin.