Application of Multinomial Logistic Regression to Model the Impact of Rainfall Genesis on the Performance of Storm Overflows: Case Study

Abstract In this study, a mathematical model was proposed to analyze the performance of storm overflows. The model included the influence of rainfall genesis on the duration of storm overflow, its volume, and the maximum instantaneous flow. The multinomial logistic regression model, which has not been used to model objects located in a stormwater system, was used to simulate the duration of storm overflow. The Iman–Conover method, involving theoretical cumulative distributions determined based on 30-year rainfall sequences, was adopted to simulate the rainfall characteristics describing the overflow performance (total and maximum 30-min rainfall depth and rainfall duration). Simulations showed a significant impact of rainfall genesis on the parameters of the storm overflow. The model presented in this work and the results obtained can be used at the dimensioning stage of storm overflows and to create an early warning system against undesirable phenomena in the stormwater system within urban catchments.