The combination of intense climate change and rapid urbanization means that urban storms and floods capable of causing substantial economic losses are becoming one of the principal disasters that threaten sustainable urban development. Therefore, atmospheric–hydrological modeling, which could effectively extend the forecast period and support related research in regions lacking precipitation data, has received increasing attention in relation to urban hydrology. In this study, considering more comprehensive optimization of the physical parameterization schemes of the Weather Research and Forecasting (WRF) model, an atmospheric–hydrological modeling system based on the WRF model and Storm Water Management Model (SWMM) was proposed and applied to Beijing's sub-center. Results showed that the cumulus parameterization scheme has greatest impact on simulation of heavy rainfall. Overall, the combination of the WRF Single-Moment 5-class scheme, Grell–Freitas scheme, Yonsei University scheme, the newer version of the Rapid Radiative Transfer Model, Revised MM5 Monin–Obukhov scheme, Noah land surface model, and the Urban canopy model produced the best performance. Application of the proposed system demonstrated the feasibility of simulating urban storms and floods via WRF–SWMM system, which can represent the entire process from the rainfall produced by atmospheric movements to the formation of the urban floods.