Author(s): Qian Zhu; Xujie Zhang; Xichao Gao; Chong Ma; Yueping Xu
Linked Author(s):
Keywords: Calibration strategy; SWAT; Natural runoff; SWAT-CUP
Abstract: Parameters are very important in hydrologic models, but they are often not exactly known, thus calibration is essential to determine the parameters. However, different calibration strategies may lead to different results, which probably introduce more uncertainty in flow simulation. In this paper, the SWAT (Soil and Water Assessment Tool) model is used to investigate the impact of two different calibration strategies on flow simulation in Xinanjiang catchment, one sub-basin of Qiantang River Basin, East China. The catchment is subdivided into 26 hydrological response units and six sub-basins. The discharges from 1981 to1987 at six control hydrological stations are used in the calibration. After a preliminary sensitivity analysis, twelve parameters are selected as sensitive ones and used in the calibration. Two calibration strategies are as follows: Strategy One is to calibrate the six sub-basins separately; Strategy Two is to calibrate the catchment as an integrated unit. Because of the unavailability of measured discharges from 1981 to 1987 at Luotongbu Station, which is the outlet of the whole catchment, an empirical formula is used to obtain the natural runoff. The optimization program SUFI-2 (Sequential Uncertainty Fitting Ver. 2) is employed to do the calibration. The Nash –Sutcliffe coefficient is used as the objective function, and others including R2 and MSE (mean square error) are also calculated as auxiliary indices to evaluate the efficiency of calibration. It turns out that the Nash –Sutcliffe coefficient and R2 based on Strategy One are larger than those based on Strategy Two, except for Tunxi sub-basin. The final conclusion is that Strategy One is more appropriate although more parameter uncertainties were also introduced.
Year: 2013