The CSLE model based soil erosion prediction: Comparisons of sampling density and extrapolation method at the county level

The Universal Soil Loss Equation (USLE) is an empirical equation commonly applied to predict soil erosion throughout the world. In China, the Chinese Soil Loss Equation (CSLE) has been developed for estimating annual soil erosion by water on the basis of the USLE. It is proven that two factors (i.e., sampling density and extrapolation method) are important in the application of CSLE for soil erosion survey and evaluation. In this study, based on the SPOT 6 image (1.5 m resolution) and the 1:10,000 topographic map in Yishui county of the Yimeng Mountain Area (YMA), eastern China, two sampling densities (1% and 4%) were selected and three extrapolation methods (grid based calculation, direct extrapolation, and Kriging extrapolation based on sampling units) were adopted to compare their effect on the CSLE calculated soil erosion condition and status. Results showed that estimated soil erosion by direct extrapolation and Kriging extrapolation methods varies largely with sampling density. The discrepancy of the soil erosion area ratios calculated by the two methods are 6.5% and 7.7% under the 1% and 4% sampling density, and the relative discrepancies are 15.9% and 16.2%, respectively. But the grid calculation method is less affected by sampling density. The discrepancy of soil erosion area ratios is only 0.9% and 2.0%, respectively, under the 1% with 4% sampling density. Considering both field workload and prediction accuracy of soil erosion, the 1% sampling density with the grid calculation method is recommended when high resolution remotely sensed data are available. Otherwise, the 4% sampling density with either extrapolation method should be used. The research can provide useful reference information for ensuring the accuracy and reducing the survey workload on the application of CSLE in other areas.