Parameter sensitivity analysis of SWMM model based on local and global methods

Abstract

[Objective] The sensitivity analysis of model parameters is conducive to enhancing the optimization efficiency of model parameters, reducing model uncertainty, and improving the model accuracy. [Methods] A residential community in Qian'an City was selected as an example, and the SWMM model was constructed. The analysis of parameter sensitivity was conducted by employing the modified Morris method and the Sobol method, with peak flow and runoff coefficient as optimization objectives under different rainfall scenarios. The advantages and disadvantages of these two methods were also evaluated. [Results] The result indicate that both the modified Morris method and Sobol method can identify high sensitivity parameters under different objective functions. For peak flow, both methods identify the roughness coefficient of the pipeline as the most sensitive parameter, with sensitivities of-0.563~-0.157 and 0.116~0.915, respectively. For runoff coefficient, the modified Morris method identifies the maximum infiltration rate as the most sensitive parameter, with sensitivities of-0.358~0. The Sobol method identifies the storage capacity of impermeable areas as the most sensitive parameter, with sensitivities of 0.238~0.961. The second-order sensitivity index calculated by the Sobol method shows that the interaction between parameters is mainly the combined effect between the two parameters. The modified Morris method can qualitatively sort the sensitivity of parameters, and the calculation is efficient and convenient. The Sobol method can comprehensively analyze the influence mechanism of parameters and their interactions. [Conclusion] The comprehensive application of qualitative and quantitative analysis method has important guiding significance for in-depth analysis of model parameter sensitivity, accurate identification of parameter calibration direction, and improvement of model reliability.

Keywords

SWMM model / sensitivity analysis / modified Morris method / Sobol method / peak flow / runoff coefficient / rainfall / numerical simulation