Soil erosion rates assessed by RUSLE and PESERA for a Chinese Loess Plateau catchment under land‐cover changes

ABSTRACT On the Chinese Loess Plateau, soil erosion models are often employed to predict erosion rates and responses to land‐use/‐cover changes (LUCCs). Previous Loess Plateau studies employed individual models with specific emphases but model comparisons have not been undertaken so the relative performance of different models is not known. In this study we employed two extensively applied models (RUSLE and PESERA) to investigate the impact of LUCCs during 1990–2000 and 2000–2011 on soil erosion rates for a typical Loess Plateau catchment (i.e. Huangfuchuan), and compared their modelling results. Land‐cover patterns for 1990, 2000 and 2011 were derived from Landsat images. The catchment was dominated by grassland (over 70%) and experienced considerable LUCCs: vegetation coverage increased from 38.3% in 1990 to 48.7% in 2011. Modelling results suggested that mean soil erosion rates of the catchment increased under the 1990–2000 LUCC and decreased under the 2000–2011 LUCC. Sandy land and scrubland were found to suffer from most severe soil erosion and thus should be the focus of future conservation work. Mean soil erosion rates on steep slopes (i.e. > 25°) were predicted to increase under the 2000–2011 LUCC, implying that further work is still needed to study soil erosion processes and their conservation on steep slopes. Model comparisons showed that RUSLE predictions were higher than PESERA predictions for most areas (particularly for steep slopes), and the former were generally closer than the latter to check‐dam sediment yield measurements. RUSLE and PESERA results were not linearly correlated, possibly due to differences in their underlying principles and their sensitivity to crucial parameters – RUSLE is more sensitive to slope gradients while PESERA is more sensitive to vegetation coverage. PESERA needs improvement to better account for steep slope erosion processes on the Loess Plateau, while RUSLE needs improvement in the description of vegetation effects. © 2019 John Wiley & Sons, Ltd.