Sensitivity identification of SWMM parameters and response patterns of runoff pollution on hydrological and water quality parameters

Abstract

In the calibration process of urban flood and non-point source pollution models, sensitive parameters and their changing trends cannot be quickly obtained. In this study, we accurately identify the effects of hydrological and water quality parameters in the SWMM model on runoff water quantity and water quality indicators under different return periods based on the improved Morris screening method, obtain the parameter sensitivities under different objective functions. Meanwhile, the intrinsic influence mechanism of the sensitive parameters on runoff water quantity and pollution load indicators in each return period is explored, and the integrated response law of flow-concentration-load rate process to the sensitive parameters is comparatively analyzed. The results showed that sensitivity of Destore-Imperv to the peak flow at the outfall, total outfall volume, flow peak occurrence time, and pollutant load is the greatest. Under the low return period, the change of Destore-Imperv resulted in sharp fluctuations in the load rate at T = 57 min in the mid-rainfall period, with a rate of variation of 37.74 %. While Decay Constant has the most significant effect on the load rate at T = 77 min in the mid-late rainfall period, with a rate of variation of 35.14 %. As the return period increases, the sensitivity of all hydrological and hydraulic parameters decreases except for Conduit Roughness. Wash-off exponent(C2) of the road was the strongest sensitivity to pollutant loads and peak concentrations, followed by C2 of the roof and Max.Buildup of the road, respectively. When C2 = 1.03, the peak concentration trend produces a demarcation, the smaller C2 is, the more forward the pollutant concentration wave peak is, and the more pronounced the peak pollutant concentration is in the initial rain, and vice versa, the more backward the pollutant concentration wave peak was, the initial concentration was extremely low while the later concentration surged. As the return period increased, the pollutant concentration wave peaks of the same C2 were more forward, and the loading rate and concentration process trends were similar. The results of the study not only provide a reference for modeling calibration to determine the optimal parameter combinations, but also assist in finding solutions that can satisfy the regulation of urban runoff and non-point source pollution.