JWSS - Journal of Water and Soil Science

The annual soil erosion in different regions of the world has been estimated using various empirical and numerical methods whose accuracy is very dependent on their utilized parameters. One of the most common methods in the evaluation of the mean annual soil erosion especially in sheet and furrow regions is the USLE method. In this relationship, almost all factors that normally affect the soil loss process such as land cover, slope, precipitation, soil type, and support practice parameter of soil have been employed but, in this research, it was shown that the accuracy of this method in mountainous areas covered by rock and snow is somewhat low. To do this, a part of the Tibet plateau in China, where observation soil loss data were available, was selected for investigation. To implement the numerical and analytical analysis, many maps including DEM, NDVI, orientation, soil type, mean monthly and annual precipitation for 30 years were collected. For increasing the accuracy of the model, the cover management parameter was extracted from high accuracy NDVI maps and all USLE parameters were calculated in ArcGIS. The final results were shown that the amount of annual soil loss which was estimated by the USLE method is more than the observed data which were collected by Chinese researchers. This is because the large areas of the study area are covered by lichen and snow where soil loss due to the erosion process is very low but these regions cannot be recognized from NDVI maps. Also, the analysis of the NDVI maps was shown that the relationships of Fu, Patil, and Sharma were not suitable for soil loss estimation in elevated mountainous areas. If the other relationships such as Lin, Zhu, and Durigon are used for the regions with a height of more than 5500 m, a new correction coefficient needs to be used for the C factor which was calculated as 0.2 for the study area.