ABSTRACT Gully erosion is a severe form of land degradation and soil erosion that is more distinct in the arid and semi‐arid areas of the globe. Due to the autochthonous characteristics, gully erosion leads to several types of land degradation activities and becomes a threat to food security and ecological stability. Therefore, identification of gully erosion zones is highly required for maintaining ecological sustainability. This study aims to identify probable gully erosion hotspots using machine learning applications and investigate the spatial pattern and annual soil loss using the Revised Universal Soil Loss Equation (RUSLE) method in the North Koel River Basin, Jharkhand. The gully inventory map has been prepared based on field investigation and satellite imagery. Both of the applied models, that is, support vector machine (SVM) and random forest (RF), identified that the southern and northern parts of the North Koel River Basin are prone to high and very high gully susceptibility. The RF model has better predicting capability (AUC = 0.778) than SVM, and it predicts that approximately 30% area of the basin is under the high and very high gully, susceptible classes. The estimated mean annual potential soil loss within the North Koel River Basin is 377789.6 tons/year. The potential soil erosion map shows that total soil loss is mainly generated from a small area, particularly from hilly tracts, exposed surfaces, and cultivated areas. This study also quantifies the sediment loss, identified gully‐ prone areas, highlighting the gully triggering factors that prioritise the conservation strategies. Gully erosion management includes a combination of diverting surface runoff, stabilising the gully head, reducing water velocity within the channel, and implementing vegetative controls. The primary techniques comprise construction check dams, shaping slopes, planting deep‐rooted grasses, and ensuring suitable agricultural practices, such as contour farming.
Saha et al. (Sun,) studied this question.