基于降雨和地形特征的输出系数模型改进及精度分析

降雨是非点源污染产生的主要驱动条件，地形是非点源污染输移的重要影响因素，两者均与非点源污染紧密相关。输出系数模型是大尺度流域非点源污染模拟的常用模型，从降雨和地形对非点源污染产生、迁移的影响出发，对现有输出系数模型进行了改进，并给出了改进模型中降雨影响因子和地形影响因子的确定方法，以提高模型在大尺度流域的模拟精度。最后，以长江上游1990和2000年TN负荷估算为例，分析了模型改进前后的精度变化。精度对比表明，通过模型改进，流域综合相对误差从4502%和4249%减小到2393%和1838%，直门达站、沱江大桥站和罗渡溪站控制流域的相对误差有明显减小。改进后的输出系数模型结构更合理，模拟更符合实际。

IMPROVED EXPORT COEFFICIENT MODEL CONSIDERING PRECIPITATION AS WELL AS TERRAIN AND ITS ACCURACY ANALYSIS

The dominant method in the field of non-point source(NPS)pollution research is NPS simulation model,in which export coefficient model is widely used.But for basins with obvious spatial differences of precipitation and terrain,such as upper reach of the Yangtze River,the model accuracy is limited for the sake of its neglecting of precipitation and terrain impact on NPS pollution.Considering that,an improved export coefficient model is proposed in this article,and the method for determining precipitation and terrain factors was also offered.The model was applied to simulate the total nitrogen load of upper reach of the Yangtze River in 1990 and 2003.It is indicated that average values of relative error for various watersheds reduced from 45.02% to 23.93% and from 42.49% to 18.38% respectively by using the improved model.Relative error values for watersheds control by Zhimenda,Luoduxi and Tuojiangdaqiao stations decreased obviously.The relative error values of Zhimenda Station reduced from 207.29% to 24.84%(1990 year)and from 163.25% to 18.16%(2003 year)respectively,those of Luoduxi Station decreased from-33.72% to-19.24%(1990 year)and from-41.20% to-24.35%(2003 year)respectively,and the relative error value of Tuojiangdaqiao Station reduced from 32.47% to-2.15%(2003 year).It can be concluded that the improved model is feasible and the predicted result is approximately in accordance with the monitoring data.For large scale basin,especially those with obvious spatial differences of precipitation and terrain,the improved model has its theoretical and practical significance.