基于偏相关法的暴雨管理模型参数敏感性分析

摘要

基于拉丁超立方抽样法对暴雨管理模型(SWMM)输入参数进行采样,采用偏相关法对参数进行全局敏感性分析.取某小区为研究区域构建SWMM模型,分析水文水力参数变化对模型输出变量的影响大小.研究发现,峰值流量最敏感的参数是透水区糙率系数,但该系数并不起决定性作用,它和其他较敏感参数共同决定着峰值流量的大小;峰现时间最敏感的参数是最小渗透率,但是所有参数的敏感性都较低,对峰现时间的影响都很小;总产流量最敏感的参数依次是最小渗透率、面积修正因子、渗透衰减系数和最大渗透率.通过敏感性分析识别出敏感参数,可提高参数率定效率和模拟结果的可靠性.%Partial correlation analysis was used to assess the global sensitivity of SWMM input parameters which were sampled by the Latin hypercube sampling method. The influence of the change of input parameters to its output were analyzed by choosing a plot used to build the SWMM model. The results show that Manning's N for pervious area is the most sensitive parameters for peak flood,but it does not play a de-cisive role,Manning's N and other sensitive parameters affect the size of the peak flow together.The parameter of the greatest impact on peak time is the minimum infiltration rate,but all the influence of parameters on the peak time is very small. The descending order of parameter sensitivity to the total runoff volume is the minimum infiltration rate,area correction factor,decay coefficient of infiltration and the maximum infiltration rate. The analysis of the sensitivity of input parameters can improve the calibration efficiency of model parameters and upgrade the accuracy of simulation result.