不同尺度流域月径流模拟对比与敏感性分析——以西江流域为例

通过新安江月径流模型耦合加速遗传算法对西江流域3个不同尺度子流域——新兴江流域(1 790 km2)、罗定江流域(3 164 km2)和贺江流域(8 326 km2)进行了模拟,同时对模型参数进行了敏感性分析.结果表明:1)在模型参数较多的情况下,用加速遗传算法能很快找到最优化参数,通过参数的初始区间的合理设定可以使参数值达到全局最优,并且使其符合参数的实际取值情况;2)在1 000～9 000 km2流域面积内只有1个气象站的情况下,利用新安江月径流模型模拟的Nash-Sutcliffe效率系数为65%以上;3)降雨径流的相关性越强,则模拟精度越高;4)20世纪80年代后,罗定江流域径流的变化主要受降雨的影响,人类活动影响相对其他2个子流域弱;5)新安江月径流模型的敏感参数为蒸发系数K、地下水出流系数KG、地下水消退系数CG和自由水蓄水容量SM(0～10),参数的取值范围会影响到模型参数的敏感性;异参同效仍然是流域水文模型的不确定性的一个重要方面,GLUE方法能够反映这一点.

Comparison of Monthly Runoff for Different Scaling Basin and Sensitivity Analysis:A Case Study in the Xijiang Watershed

By coupling Accelerate Genetic Algorithm,the study simulates the monthly runoff of three different scaling basins（Xinxing River basin by 1 790 km2,Luoding River basin by 3 164 km2 and He River basin by 8 326 km2）in the Xijiang watershed with Xin′anjiang monthly model,and the sensitivities of parameters are then analyzed.The results show that：1）when there are too many parameters,by using the AGA optimal parameters can be found quickly,and the method can make the parameters get global optimal values and correspond with the reality by designing reasonable initial ranges of the parameters;2）For the basin with 1 000~9 000 km2 area,Nash-Sutcliffe efficiency coefficient can be more than 70% by using Xin′anjiang monthly model even though there is only one meteorological station;3）The stronger the relationship between precipitation and runoff,the higher the simulation accuracy will be.4）After 1980,the change of runoff on Luoding River basin is affected by precipitation and the influence of human activities is weaker in the other two basins;5）The sensitive parameters of Xin′anjiang monthly model are evaporation coefficient（K）,impermeable area ratio（IMP）,effluent coefficient of groundwater（KG）,degrading coefficient of groundwater（CG）and the capacity of free water（SM）,and the range of the parameters will affect the sensitivity;and meanwhile GLUE method can show equifinality of the parameters,which is an important aspect of the uncertainty of hydrological model.