PESERA-LP: A coarse-scale process-based fluvial erosion model for topographically complex regions

The Chinese Loess Plateau, characterized by complex and fragmented topography, has undergone severe soil loss over the past century. While over thirty soil erosion models have been used in the region, most contemporary models are either catchment-scale or event-based. There is a notable absence of regional-scale models that account for erosion-relevant processes specific to the Plateau. In this study, we developed a new scheme (PESERA-LP) for the simulation of soil erosion processes on the Loess Plateau. The model integrated advanced hydrological, vegetation, and erosion modules to enhance our understanding and prediction of soil erosion dynamics on the Plateau. In our scheme, the key parameter of the hydrological module was spatialized based on precipitation, while the terrain factor from the Revised Universal Soil Loss Equation (RUSLE) model and the erodibility factor of the Erosion Productivity Impact Calculator (EPIC) model were incorporated into the erosion module. Additionally, the parameters of the vegetation growth module were also optimized. PESERA-LP was implemented in both equilibrium and time-series modes, with a validation conducted based on field measurements. Validation of runoff depth in the equilibrium mode demonstrated a Root Mean Square Error (RMSE) of 0.47 mm a⁻¹ and a Nash-Sutcliffe Efficiency (NSE) of 0.63, while the time-series mode exhibited an RMSE of 0.25 mm m⁻¹ and an NSE of 0.58. As for erosion rate, RMSE and NSE were 6.04 t ha⁻¹ a⁻¹ and 0.89 in the equilibrium mode, compared to 0.99 t ha⁻¹ m⁻¹ and 0.52 in the time-series mode. Sensitivity analysis demonstrated that modelled runoff depth was sequentially impacted by precipitation, temperature, and vegetation cover, while modelled erosion rates were sequentially influenced by vegetation, precipitation, slope gradient, and temperature. The equilibrium mode is suitable for assessing spatial variability of average erosion rates across large areas, whereas the time-series mode is preferentially used for analyzing continuous monthly erosion rates in relatively small areas.