城市降雨径流水文模拟的参数局部灵敏度及其稳定性分析

基于Morris筛选法,在厦门城市小流域采用8场实测降雨数据对城市降雨径流模型SWMM的水文水力模块进行局部灵敏度及其稳定性分析.结果表明,Area、%Imperv和Dstore-Imperv是影响降雨总径流量和流量峰值最灵敏的参数.对于总径流量,Area、%Imperv和Dstore-Imperv的灵敏度分别为0.46～1.0、0.61～1.0和-0.050～-5.9;而对于流量峰值,它们的灵敏度分别为0.48～0.89、0.59～0.83和0～-9.6.其中降雨强度最小的场次降雨的各个参数降雨总径流量和流量峰值的灵敏度都最大,而降雨强度较大的场次降雨的总径流量和流量峰值的灵敏度都较小.不同场次降雨模型参数的灵敏度分析具有很大的差异性,但%Zero-Imperv对总径流量和流量峰值模拟输出影响的稳定性最小,表现在变异度最大,高达221.24%和228.10%,而Conductivity参数稳定性最大,变异系数都为0.

Local Sensitivity and Its Stationarity Analysis for Urban Rainfall Runoff Modelling

Sensitivity analysis of urban-runoff simulation is a crucial procedure for parameter identification and uncertainty analysis. Local sensitivity analysis using Morris screening method was carried out for urban rainfall runoff modelling based on Storm Water Management Model (SWMM). The results showed that Area, % Imperv and Dstore-Imperv are the most sensitive parameters for both total runoff volume and peak flow. Concerning total runoff volume, the sensitive indices of Area, % Imperv and Dstore-Imperv were 0.46-1.0, 0.61-1.0, -0.050(-) - 5.9, respectively; while with respect to peak runoff, they were 0.48-0.89, 0.59-0.83, 0(-) -9.6, respectively. In comparison, the most sensitive indices (Morris) for all parameters with regard to total runoff volume and peak flow appeared in the rainfall event with least rainfall; and less sensitive indices happened in the rainfall events with heavier rainfall. Furthermore, there is considerable variability in sensitive indices for each rainfall event. % Zero-Imperv's coefficient variations have the largest values among all parameters for total runoff volume and peak flow, namely 221.24% and 228.10%. On the contrary, the coefficient variations of conductivity among all parameters for both total runoff volume and peak flow are the smallest, namely 0.