Modeling runoff and soil erosion in the Three-Gorge Reservoir drainage area of China using limited plot data

This paper presents a modeling approach to simulate runoff and soil erosion at the small watersheds of the Three-Gorge Reservoir drainage area in China by using limited plot data on runoff-soil erosion. The approach coupled the empirical relationships between soil loss and runoff. This relationship is derived from the experimental plots under different land use types with a spatially distributed hydrological model, WetSpa Extension, to calculate soil loss in grid cells. A topographic factor was also developed to account for the impacts of topography on soil loss. Finally, a constant Sediment Delivery Ratio (SDR) was applied to calculate sediment yields transported to the catchment outlet from grid cells. The coupled model was first calibrated for the study areas using the observed continuous discharge data and event-based sediment data over the period from May 1st through October 31st of 2001 in a small sub-catchment in Sichuan, China. The calibrated model was then applied to the same sub-catchment over the same period of every year from 1993 to 2000 as well as the neighboring catchments to assess the performance and stability of the model. A comparison of observed and simulated stream flows indicates that the model performance was acceptable in six (1993, 1994, 1996, 1998, 1999 and 2000) of the 8 years with Nash–Sutcliffe Efficiency (NSE) of 0.654 and 0.729 for weekly and monthly flow discharges, respectively. A comparison between the observed and simulated sediment yields showed that the simulation of sediment yields was very satisfactory (NSE = 0.88) and the spatial variability of soil erosion rates within the catchments was predominantly controlled by land use types. Finally, the model was applied to assess the efficacy of soil conservation through land use changes in Sub-catchment No. 2. The results clearly indicated that land use changes after 1990s have been very effective in reducing both runoff and sediment. This study suggests that experimental plot data is an effective supplement to modeling spatial variation of soil erosion, albeit their various limitations.