Integration of geospatial techniques and soil loss modelling for soil conservation in a drought prone river basin of eastern India
摘要
Soil erosion is a major environmental challenge in India, where millions of tonnes of fertile soil are lost annually, adversely affecting agricultural productivity and ecosystem sustainability. Reliable spatial assessment of erosion risk is therefore essential for prioritizing conservation interventions. In parts of eastern India, characterized by intense monsoonal rainfall, undulating topography, and dynamic land-use patterns, soil erosion remains a critical concern. Within this region, parts of the Mahanadi River basin have been identified as highly susceptible to soil erosion. Jonk River Basin, being a part of Mahanadi basin, has been recognized by the National Bureau of Soil Survey and Land Use Planning as a highly vulnerable area for soil conservation; however, detailed spatial assessment of soil erosion at basin and sub-basin scales remains largely unexplored. This study presents the novel contribution for spatial assessment of soil erosion in the Jonk River Basin using the Revised Universal Soil Loss Equation (RUSLE) integrated with Geographic Information Systems (GIS), remote sensing, and statistical computing in the R programming language. Rainfall, soil properties, topography, land-use/land-cover, and conservation practice datasets were integrated to derive RUSLE factors and generate basin-wide soil loss estimates. In addition, a factor influence analysis was performed in R to quantify the relative contribution of individual erosion-controlling parameters. The results reveal that 4.81% of the basin area falls under moderate to very high erosion risk classes, with soil loss ranging from 55 to 137.5 t ha−1 yr−1. Sub-basin level analysis identifies SB-8 as the most erosion-prone region, primarily due to its moderately to highly dissected plateau terrain with hills and valleys that enhance runoff and sediment transport, whereas SB-5 experiences relatively low erosion risk. Model validation using field-observed erosion sites and Receiver Operating Characteristic (ROC) analysis demonstrates strong predictive performance, with an Area Under the Curve (AUC) value of 0.97, indicating very high model accuracy. The Random Forest-based importance analysis revealed the dominant role of slope length and steepness (LS) and conservation practice (P) factors in controlling erosion dynamics within the basin. The findings also highlight significant spatial heterogeneity in erosion processes and critical management gaps, including exposed slopes lacking protective measures and the prevalence of water-intensive paddy monocropping in drought-prone conditions. The integration of RUSLE with GIS, remote sensing, and the R environment provides a reproducible and scalable framework for soil erosion assessment. The study therefore, provides a scientific basis for targeted soil and water conservation planning and sustainable watershed management in erosion-prone regions.