The Sobol method based on variance decomposition was used to calculate the sensitivity of parameters in the storm water management model (SWMM) to runoff, peak flow, and peak time under two different rainfall conditions, and the convergence of the results was briefly analyzed, in order to quantitatively explore the global sensitivity of hydro-hydraulic parameters in the typical urban SWMM. Results showed that the sensitive parameters and relative ranking of hydro-hydraulic modules in SWMM were different under different rainfall intensities. The impervious areas’ Manning roughness was the most sensitive parameter for all three kinds of output variables under the heavy rainfall intensity condition, and the total sensitivity values were all greater than 0.6. The Manning roughness coefficient of the concrete pipe section was much more sensitive to the peak time than to the other two types of output variables. Percent of impervious area with no depression storage was sensitive parameter, and the impervious ground surface store was the most sensitive parameter with the first-order sensitivity index of 0.396 in the small rainfall intensity condition. The calculated Sobol indexes fluctuated in the first 63 000 runs of the model and tended to converge after 84 000 runs.

Keywords： sensitivity ; Sobol method ; storm water management model (SWMM) ; model parameter ; quantitative analysis