Gully Erosion Probability Mapping Using Random Forest Algorithm in Machine Learning Framework

Gully erosion threatens soil quality and agricultural productivity. This research assessed gully erosion risk in West Bengal’s Sita Nala basin using random forest algorithms. By analyzing 16 environmental factors across 200 sample points, the study identified key contributors including relief, drainage density, and vegetation cover. The resulting probability map classified areas into three risk levels, with 13.42% showing significant erosion risk. The model proved highly accurate with an 85.1% ROC value.