Author(s): Feng Wenwen; Chao Wang
Linked Author(s):
Keywords: Water system City Simulation model Uncertainty SWMM
Abstract: River and lake systems are not only vital ecological landscapes and water resource carriers in urban areas but also serve as essential foundations for achieving urban water storage, detention, and drainage functionalities. In the simulation of urban river and lake systems, the complex hydrological and hydraulic processes often result in low accuracy of simulation models. This paper constructs a high-precision simulation model for urban river and lake systems based on hydrological and hydraulic processes. Nonlinear equations are employed to accurately describe the rainfall-runoff processes, thereby reducing the complexity of the water system. To address the issue of numerous parameters and the difficulty in calibration in the SWMM (Storm Water Management Model), a parameter optimization and calibration method based on the BP (Back Propagation) neural network is proposed. This method reduces the uncertainty of the water system and improves the accuracy of the simulation model through piecewise calibration. The results of long-term rainfall tests show that the average Nash-Sutcliffe Efficiency (NSE) coefficient at the observation sections is 0.556, and the average Root Mean Square Error (RMSE) is 0.077. These results confirm the effectiveness and precision of the proposed simulation model.
Year: 2025