This study proposes a novel hybrid LSTM-SWMM model that integrates the advantages of SWMM model and the LSTM neural network for the first time to predict runoff in urban areas. The aim is to build an efficient and rapid model that takes into account the physical mechanism, so as to better respond to and simulate urban flood. The results show that, in the training period and testing period, the simulated discharge process of LSTM-SWMM model and the observed discharge process are in good agreement, which can reflect the actual rainfall runoff process. The R² of LSTM-SWMM model is 0.969, and the R² of LSTM model is 0.954. Also, when the forecasting period is 1, the NSE value of LSTM-SWMM is 0.967, which is the best forecasting accuracy; when the forecasting period is 6, the NSE value of LSTM-SWMM is 0.939, which is the worst. With the increase of the forecasting period, the NSE values show a downward trend, and the accuracy gradually decreases.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.