Abstract: Storm Water Management Model (SWMM) is an effective tool for studying urban flood and waterlogging problems. The values of its sensitive parameters affect not only the calibration efficiency of model parameters but also its prediction accuracy. To accurately identify the sensitive parameters, this study uses both the modified Morris method and mutual information method to make sensitivity analysis from the local and global perspectives, respectively. The results show that for this model, the sensitive parameters for both peak flow and runoff coefficient are the Manning coefficient of the pervious area and the minimum infiltration rate, and either the modified Morris method or mutual information method can be used to identify these sensitive parameters, but the former is limited to relatively low recognition capability and the latter to a low recognition efficiency. Thus, we suggest that the modified Morris method be applied to the case of quick identification of the main sensitive parameters, and the two methods be combined to further identify other sensitive parameters.

Key words: sensitivity analysis, Storm Water Management Model, modified Morris method, mutual information method, Dahongmen drainage area