Abstract
Storm Water Management Model (SWMM) is one of the most commonly used models in urban flood simulation. However, because the calibration and verification of the model's uncertainty parameters are extremely dependent on the measured flood data, it is difficult to apply the model in areas lacking data. This study proposes a parameter sample clustering method based on peer research results to determine the uncertainty parameters of SWMM, and compares the simulation results with the simulation results of the manual adjustment method based on measured data. The research shows that the Absolute error (AE), Relative error (RE), Nash efficiency coefficient (NSE), and Coefficient of determination (R2) of the water depth simulated by the parameter sample clustering method are 0.040m, 9.08%, 0.949, 0.967 compared with the measured value, respectively. The value of AE, RE, NSE, and R2 of the manual tuning method during the calibration simulation period are 0.066m, 15.95%, 0.881 and 0.924, respectively. Therefore, the parameter sample clustering method has a better simulation effect than manual tuning method, and it can be further promoted in areas without flood data.