Flood forecasting using traditional physical hydrology models requires consideration of multiple complex physical processes including the spatio-temporal distribution of rainfall, the spatial heterogeneity of watershed sub-surface characteristics, and runoff generation and routing behaviours. Data-driven models offer novel solutions to these challenges, though they are hindered by difficulties in hyperparameter selection and a decline in prediction stability as the lead time extends. This study introduces a hybrid model, the RS-LSTM-Transformer, which combines Random Search (RS), Long Short-Term Memory networks (LSTM), and the Transformer architecture. Applied to the typical Jingle watershed in the middle reaches of the Yellow River, this model utilises rainfall and runoff data from basin sites to simulate flood processes, and its outcomes are compared against those from RS-LSTM, RS-Transformer, RS-BP, and RS-MLP models. It was evaluated against RS-LSTM, RS-Transformer, RS-BP, and RS-MLP models using the Nash–Sutcliffe Efficiency Coefficient (NSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Bias percentage as metrics. At a 1-h lead time during calibration and validation, the RS-LSTM-Transformer model achieved NSE, RMSE, MAE, and Bias values of 0.970, 14.001m³/s, 5.304m³/s, 0.501% and 0.953, 14.124m³/s, 6.365m³/s, 0.523%, respectively. These results demonstrate the model's superior simulation capabilities and robustness, providing more accurate peak flow forecasts as the lead time increases. The study highlights the RS-LSTM-Transformer model's potential in flood forecasting and the advantages of integrating various data-driven approaches for innovative modelling.

Over the past several decades, urban flooding and other water-related disasters have become increasingly prominent and serious. Although the urban rain flood model’s benefits for urban flood simulation have been extensively documented, the impact of rainfall input to model simulation accuracy remains unclear. This systematic review aims to provide structured research on how rain inputs impact urban rain flood model’s simulation accuracy. The selected 48 peer-reviewed journal articles published between 2015 and 2019 on the Web of Science™ database were analyzed by key factors, including rainfall input type, calibration times and verification times. The results from meta-analysis reveal that when a traditional rain measurement was used as the rainfall input, model simulation accuracy was higher, i.e., the Nash–Sutcliffe efficiency coefficient (NSE) of traditional technology for rain measurement was higher than the 0.18 for the new technology rain measurement with respect to flow simulation. In addition, the single-field sub-flood calibration model was better than the multi-field sub-flood calibration model. NSE was higher than 0.14. The precision was better for the verification period; NSE of the calibration value showed a 0.07 higher verification value on average in flow simulation. These findings have certain significance for the development of future urban rain flood models and propose the development direction of the future urban rain flood model. Finally, in view of the rainfall input problem of the urban storm flood model, we propose the future development direction of the urban storm flood model.