RUSLE, the most commonly used method for studying soil erosion, mainly serves to analyze soil loss by water erosion and to identify the most erosion prone areas (Renard et al., 1997). To some degree, the pattern of soil erosion sensitivity is consistent with the intensity of soil
erosion.
Generally, serious soil erosion areas are distributed in regions with extreme to high sensitivity, but there are differences in areas where high vegetation coverage causes such regions to have less soil erosion. The RUSLE model does not produce true soil erosion rates due to limitations such as accessibility and connectivity problems, the existence of temporary sediment stores, and the occurrence of lateral erosion in fluvial channels.
However, the accuracy of the results can be improved using remote sensing and GIS technology combined with field validation.
It is difficult and expensive to perform direct field measurements at a large scale, so model validation has always been largely ignored or restricted (Jetten et al., 2003).
In our study, therefore, validation was performed based on the comparison between soil loss measured from 137CS and predicted from the RUSLE model at the ecological observation stations.