LSTM Model-Based Rapid Prediction Method of Urban Inundation with Rainfall Time Series

Xinxin Pan, Jingming Hou (), Xujun Gao, Guangzhao Chen, Donglai Li, Muhammad Imran, Xinyi Li, Nan Yang, Menghua Ma and Xiaoping Zhou
Additional contact information
Xinxin Pan: Xi’an University of Technology
Jingming Hou: Xi’an University of Technology
Xujun Gao: Xi’an University of Technology
Guangzhao Chen: Xi’an University of Technology
Donglai Li: Xi’an University of Technology
Muhammad Imran: Xi’an University of Technology
Xinyi Li: Xi’an University of Technology
Nan Yang: Gansu Research Institute for Water Conservancy
Menghua Ma: China Power Construction Group, Co. LTD, Northwest Engineering Corporation Limited
Xiaoping Zhou: China Power Construction Group, Co. LTD, Northwest Engineering Corporation Limited

Water Resources Management: An International Journal, Published for the European Water Resources Association (EWRA), 2025, vol. 39, issue 2, No 6, 688 pages

Abstract: Abstract In recent years, with the increasing frequency of extreme rainfall events, the resulting urban inundation disasters have become increasingly severe. Rapid and accurate urban flood simulation and prediction are of great significance for disaster prevention and mitigation. However, physically-based numerical models require substantial computation time for simulating urban flood processes. In this study, we introduce the LSTM algorithm to replace physically-based numerical models for the rapid prediction of flood processes at urban inundation points. First, a hydrological-hydrodynamic numerical model for the study area is constructed to simulate flood processes under different rainfall scenarios, forming a result database. Next, the LSTM algorithm is used to train and learn from the simulated flood data, and the reliability of this learning method is verified. Finally, a rapid prediction model for flood processes at inundation points in the study area is developed. The results indicate that the prediction model achieves high accuracy, with R² values above 0.90 for predicting flood processes and peak flood characteristics at single inundation points. The MAE is no greater than 0.069, and the RMSE is no greater than 0.077. The error in the inundation process ranges between − 0.5% and 0.5%. In terms of efficiency, the average time taken to predict a single rainfall event is only 0.193 s, compared to 4625.92 s for the hydrodynamic model, representing a speedup of approximately 23,968 times relative to the physically-based numerical model. These findings demonstrate that this method meets the needs of daily urban early warning and forecasting work, enhances the city’s disaster prevention and mitigation capabilities, and effectively reduces the loss of life and property.

Keywords: Urban Inundation; LSTM Algorithm; Rainfall Time Series; Rapid Forecasting; Physically-based Numerical Model (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s11269-024-03972-z Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:waterr:v:39:y:2025:i:2:d:10.1007_s11269-024-03972-z

Ordering information: This journal article can be ordered from
http://www.springer.com/economics/journal/11269

DOI: 10.1007/s11269-024-03972-z

Access Statistics for this article

Water Resources Management: An International Journal, Published for the European Water Resources Association (EWRA) is currently edited by G. Tsakiris

More articles in Water Resources Management: An International Journal, Published for the European Water Resources Association (EWRA) from Springer, European Water Resources Association (EWRA)
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().