1. Key Laboratory of Forest Ecology and Environment of State Forestry Administration, Research Institute of Forest Ecology, Environment and Protection, Chinese Academy of Forestry, 100091, Beijing, China;
2. Co-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, 210037, Nanjing, China
The simulation accuracy of watershed hydrological model depends largely on the description of the watershed characteristics of the input data. High resolution input data can accurately describe the watershed characteristics, and at the same time increase the difficulty of data collection and processing. So it is meaningful to study the response of the model simulation results to spatial data resolution, which would help to improve model running efficiency without reducing model simulation accuracy. Based on the previous studies, taking the SWAT for example, we review the impact of data quality on the accuracy of watershed hydrological simulation, mainly including the resolution of DEM (digital elevation model),the delineation of watershed, the quality of land use/ land cover (LU/ LC) and soil maps, the number and distribution of weather and precipitation gauge stations. The main conclusions are as follows: 1) Threshold of resolution of spatial data exists in the simulation of different subjects (runoff, sediment, nutrient, etc. ), and too high or too low resolution of spatial data could result in the decrease of simulation accuracy. 2) The impact of DEM resolutions on the relative error of flow simulation results is far less than the relative error of sediment simulation, and the simulation results of inorganic nitrogen and organic phosphorus are consistent with the results of flow and sediment, respectively. 3) Further improvement of DEM resolution does not always result in the improvement of simulation accuracy after the resolution reaching certain accuracy, while the decrease of slope degree caused by the decrease of DEM resolution would decrease the flow production and delay the summit of peak flow. 4) The number of subwatersheds has little impact on the simulation of flow, but significantly influences the simulation of sediment. The number of subwatersheds and HRUs has significant impact on the upstream sediment yield, while it has little impact on the sediment load at the outlet of the watershed. 5) Effects of resolution of spatial landuse data and soil data on the simulation results of the model mainly result from their impact on the generation of HRUs. 6) The spatially distributed precipitation data could increase the accuracy of surface runoff simulation, because the distributed precipitation data could embody the locally intensified rainfall events that could significantly influence the surface runoff. The results could provide references for the development, application and improvement of watershed hydrological model and increase the accuracy of hydrological model simulation.