Runoff–Sediment Simulation of Typical Small Watershed in Loess Plateau of China

The implementation of measures such as check dams and terraces in the Loess Plateau of China has had a groundbreaking impact on water and sediment conditions. The question of how to accurately simulate the runoff–sediment process under complex underlying surface conditions has become key to clarifying the water cycle law. This study focused on the Chenggou River basin, a small watershed located in the Loess Plateau, to examine the effect of the underlying surface characteristics on the runoff production process, and the spatial distribution of the dominant runoff process in the runoff generation mechanism was determined according to the land application, slope and vegetation coverage of the watershed. A runoff–sediment model was constructed that was combined with the traditional hydrological physical mechanism and a deep learning algorithm. Different parameters were calibrated depending on the spatial distribution of the dominant runoff process and we then ran the runoff–sediment simulation model to very its serviceability in the typical watershed of the Loess Plateau. Different parameters were calibrated for each type of hydrological response unit (HRU), according to the division of each HRU and the actual flood process, to calculate the runoff yield of each HRU. An LSTM algorithm was used for flow routing and a CSLE algorithm was used to simulate soil erosion. The results show that there were 29 flood events in the Chenggou River basin from 2013 to 2017. The average runoff depth had an 8.86% margin of error, while the peak flow had a slightly higher 9.44% deviation. The Nash efficiency coefficient was 0.84, and the relative error of soil erosion was 14.45%. The model simulation effect is good and can be applied to the typical watershed of the Loess Plateau. The model can provide a scientific basis for the highly efficient and sustainable utilization of water resources, ecological environment construction and the sustainable development of agriculture.