Coupled Study Of Distributed Time Variant Gain Model And Storm Water Management Model

With the rapid development of urbanization in China,urban flooding has become increasingly serious,and the construction of sponge cities has gradually become one of the key tasks in urban construction and water environment management in China.The response of runoff to rainfall in sponge cities has significant time-varying nonlinear characteristics,however,the current research on the time-varying nonlinear mechanism of rainfall and flooding in complex substrates of sponge cities is relatively rare.Therefore,this paper takes the sponge city pilot area in Fengxi New City as the research object and couples the Distributed Time Variant Gain Model(DTVGM)flow production calculation module,which can better characterize the rainfall-runoff nonlinearity,with the pipe network confluence calculation module of the Storm Water Management Model(SWMM)to provide a new tool for urban rainfall flood simulation.The main research results obtained in this paper are as follows:(1)A coupled framework of DTVGM-SWMM model was developed to better characterize the rainfall-runoff nonlinearity in the study area.First,MATLAB software was used to rate the parameters of DTVGM;second,a call engine was employed in the Visual Studio 2022 platform to replace the results of the DTVGM yield flow calculation at the current time step with the results of the variable vOutflow in the SWMM calculation engine,which characterizes the flow values leaving the catchment area and associated with the catchment module;finally,based on the distributed time-varying The open source of SWMM software and the convenience of hybrid programming in MATLAB and Visual Studio platforms provide the technical space for the construction of the coupled model framework.(2)Based on the rainfall data of the one-year Chicago design storm,the parameters of the DTVGM flow-producing module are rate determined by a genetic algorithm with the Nash efficiency coefficient(NSE)normalization as the objective,and the convergence calculation of the coupled model is performed based on the parameters.The parameters of the coupled model are validated with 20170820 rainfall,and the NSE of the production and outflow flow process lines are calculated to be 0.83 and 0.95,respectively,and the relative errors of flood peak are-54.4%and-14.3%,respectively;and the relative errors of runoff depth are-19.8%and-13.4%,respectively.The evaluation results of the outflow flow process line of the coupled model are good,and the relative error of the flood peak of the produced flow process has a low impact on the outflow process,and the accuracy of the coupled model is high.(3)The 20170909 and 20170916 rainfall scenarios were used for rainfall simulation in the study area to explore the practical application of the coupled model.The NSE of the outflow flow process lines of the two rainfall scenarios calculated by the coupled model are 0.90 and 0.76,respectively;and the relative errors of flood peak are 30.3%and-36.3%,respectively;and the relative errors of runoff depth are-7.5%and-18.5%,respectively;and all the results satisfy the accuracy of flood forecasting except for the large relative errors of flood peak.The good evaluation index results of the coupled model make it can be a valuable supplementary or replacement model for the SWMM model.(4)A global sensitivity analysis method based on cumulative distribution function(PAWN)was used to derive a density-based sensitivity index using the parameters of the DTVGM flow production module as input and the coupled model flow production flood flow as output results,and to analyze the sensitivity of the model parameters within the range of output results greater than the actual values.The sensitivity ranking of its parameters according to this index(from largest to smallest):surface produced flow Ⅱ time-varying gain factor,hysteresis parameters related to basin soil properties and evapotranspiration,surface produced flow Ⅰ time-varying gain factor,and subsurface produced flow Ⅲ time-varying gain factor;however,when focusing on the interval when the model output produced peak flood flow is greater than 15 m3/s,the sensitivity indices of each parameter in the ranking are scaled down by 49%,38%,58%,and 45%,respectively;which is caused by the reduced range of the model output of interest.The obtained parameter sensitivity results can provide the basis for the subsequent improvement of the coupled model parameter rate and selection.