VIC模型与新安江模型径流模拟对比研究

COMPARATIVE STUDY ON THE PERFORMANCE OF VIC MODEL AND XIN'ANJIANG MODEL IN SIMULATING RUNOFF

分别选取VIC模型与新安江模型作为分布式和集总式水文模型的代表,以太湖上游山区西苕溪流域为例,以Nash-Suttcliffe效率系数为目标函数,用SCE-UA(shuffled complex evolution algorithm)自动优化算法进行参数自动率定,探究不同结构的模型在径流模拟中的差异性.模拟结果表明:VIC模型和新安江模型在该地区日降雨径流模拟中都表现出了较好的适用性,率定期模型效率系数分别为0.76和0.77,验证期分别为0.84和0.82;但是在月时间尺度上,VIC模型和新安江模型则表现出了较大的差异,率定期分别为0.78和0.87,验证期则分别为0.85和0.91;从月均方根误差和平均绝对误差来看,新安江模型也表现了更好的模拟性能.从丰水(25%)、平水(50%)和枯水(25%)三个流量等级探究VIC模型和新安江模型的模拟及预测效果,结果显示:VIC模型在丰水时模拟较好,新安江模型在平水和枯水时模拟较好,尤其是在低流量时,模拟能力远优于VIC模型.

Hydrological models have different structures, and usually represent different aspects of hydrological processes. No one can convincingly claim that a particular model is superior to others for all applications. In this study, two different models were used to simulate rainfall-runoff processes in the Xitiaoxi catchment, Taihu Basin, and the SCE-UA （Shuffled Complex Evolution algorithm） was used to calibrate the two models. Data illustrated that both VIC model and the Xin＇anjiang model could represent daily rainfall runoff processes in Xitiaoxi catchment very well; the Nash-Suttcliffe coefficient of efficiency was greater than 0.75 in calibration period and 0. 84 in validation period, both performed better on monthly scale than on daily scale. The Xin＇anjiang model performed better than VIC model on monthly scale, with improvement of approximately 24.8% and 24. 6% in calibration and validation periods, respectively. The objective function of Root Mean Square Error （RMSE） and Mean Absolute Error （MAE） showed similar result. Data were analyzed according to different discharges. It was found that accuracy of VIC model was higher in simulation of high discharge, but Xin＇anjiang model simulated middle and low flow better.