An Intelligent SWMM Calibration Method and Identification of Urban Runoff Generation Patterns

Zixin Yang ^1* Jiahong Liu ² Youcan Feng ¹ Jia Wang ² Hao Wang ² Changhai Li ¹

• ¹ Jilin University, Changchun, China
• ² China Institute of Water Resources and Hydropower Research, Beijing, Beijing Municipality, China

The reliability of simulation and prediction results in urban runoff processes simulated by SWMM depends on parameter calibration. This paper proposes a universal and effective method to improve model simulation accuracy by optimizing the parameter value ranges using an unsupervised intelligent clustering algorithm. Based on this, scenarios with varying proportions of pervious and impervious areas are set, and the key model parameters' sensitivity ranking is used to determine the main runoff generation patterns. The results show that when the impervious area is less than 10%, the parameters %Zero.Imperv, N.Imperv, and Dstore-Imperv are the top three key parameters, indicating that runoff primarily originates from pervious areas. Conversely, when the impervious area increases, runoff primarily originates from impervious areas. Since the impervious area runoff pattern uses the Unit Hydrograph Model, which has fewer parameters and is easier to calibrate, better simulation accuracy is often achieved. This study can also provide a reference for SWMM accuracy requirements under different surface conditions.