Improved Hillslope Erosion Module for the Digital Yellow-River Model

Dawei Guo,Bofu Yu,Xudong Fu,Tiejian Li
DOI: https://doi.org/10.1061/(asce)he.1943-5584.0001117
IF: 2.4
2015-01-01
Journal of Hydrologic Engineering
Abstract:Severe soil erosion in the Loess Plateau region and associated large sediment discharge downstream in China have been of great concern for years. A physically based distributed model, the Digital Yellow River model (DYRIM), was developed in 2007 for land management at the basin scale in this region. In DYRIM, an explicit sediment yield equation was derived based on simple assumptions where sediment discharge would increase nonlinearly with the slope length, leading to unrealistic results as the spatial scale increases. The main aims of this study were to reformulate the module of DYRIM in comparison with other process-based erosion models; to develop a generic framework for erosion prediction and simulation of erosion and sediment delivery processes within storm events; to validate the new hillslope erosion module for improved performance across a range of spatial scales; and to evaluate parameter consistency among different runoff events. Observed flow and sediment discharge data for 16 runoff events and three runoff plots with different slope lengths (20, 40, and 60m) for each runoff event from the Tuanshangou experimental site in the Loess Plateau region were used to calibrate the new module for a 20m slope and the calibrated parameter values were then used to predict the sediment discharge for slope lengths of 40 and 60m at the same site. For calibration, the Nash-Sutcliffe coefficient of efficiency (NSE) was 0.92 on average for the 20m plot. For prediction, the average NSE values were 0.87 for 40m plot and 0.90 for the 60m plot. There is a marked improvement over the existing module, for prediction with the existing module, the average NSE values were only -0.86 for the 40m plot and -6.3 for the 60m plot. Whereas the calibrated parameter values vary from event to event, results show that using a common set of parameter values for all runoff events to predict sediment discharge, the overall NSE values using the combined data set are as high as 0.89 for the 40m plot and 0.90 for the 60m plot. This research shows that it is absolutely imperative to include a control term involving the sediment transport capacity, even for the highly erodible region of the Loess Plateau in China.
What problem does this paper attempt to address?