融合RF与USLE的太平河小流域土壤侵蚀控制效果评估研究OA

To address soil erosion issues in the Taiping River sub-watershed and scientifically evaluate the effectiveness of ecological restoration projects,a coupled RF-USLE model was developed by integrating the traditional Universal Soil Loss Equation(USLE)with the Random Forest(RF)algorithm.Using remote sensing and geographic information system technologies,multiple factors such as slope length and slope(LS),land use type,and normalized difference vegetation index(NDVI)were extracted to quantify the contribution of each influencing factor to soil erosion intensity.Pixel-level difference analysis was conducted by combining remote sensing imagery with model output layers,revealing the spatial pattern evolution of soil erosion intensity before and after intervention through spatial distribution maps,difference quantification,and uncertainty interval assessment.The results showed that after implementation of the restoration project,the average soil erosion intensity in the watershed decreased from 26.84 t/(km2·a)to 11.52 t/(km2·a),with the overall erosion level shifting toward mild conditions.Compared with the RUSLE model,the RF-USLE model reduced prediction errors by approximately 23%,demonstrating higher stability and interpretability in complex terrain areas.These findings provide a quantifiable and scalable technical approach for assessing the effectiveness of ecological restoration in small watersheds.