COMPARATIVE EVALUATION OF SUPPORT VECTOR MACHINE AND RANDOM FOREST FOR GULLY EROSION MAPPING IN URBAN ENVIRONMENTS USING REMOTE SENSING

The rapid expansion of urban areas has led to increasing concerns about the sustainability of natural resources. One of the most affected features is soil, which undergoes constant changes in land cover and use, often resulting in severe erosion and the formation of gullies. As gully formation in urban environments becomes an increasing problem, remote sensing methods have emerged to monitor these areas. To ensure accurate detection, appropriate image classification techniques must be evaluated. This study compares the performance of two classifiers, Support Vector Machine (SVM) and Random Forest (RF), using CBERS-4A satellite imagery to classify gully areas. Results indicate that while both classifiers delineated gully boundaries with good accuracy, they faced difficulties in differentiating urban features from bare soil in some instances. Notably, RF demonstrated superior performance compared to SVM for gully assessment in urban areas.