Development of urban inundation measurement data based inundation prediction model using LSTM

Jaewoong Cho; Hoseon Kang; Hyejin Moon; Hanseung Lee

doi:10.3741/JKWRA.2026.59.1.77

All Issue

2026 Vol.59, Issue 1 Preview Page Next Page

Research Article

Development of urban inundation measurement data based inundation prediction model using LSTM 을 이용한 계측데이터기반 침수 예측모델 개발

31 January 2026. pp. 77-84

PDF XML

Abstract

Due to the recurring urban flood damage occurring annually, the National Disaster Management Research Institute (NDMI) has installed and operated such sensors in cities including Ulsan, Busan, Incheon, Cheonan, and Changwon Since 2018. These sensors are used for both real-time monitoring and validating threshold rainfall levels that cause urban flooding. This study aims to utilize the urban flood measurement data collected over multiple years to predict inundation levels using a time-series prediction model. Specifically, we adopted the Long Short-Term Memory (LSTM) model, which excels in learning long-term dependencies. The performance analysis of the model showed that it accurately predicted the onset and end time of flooding, as well as the maximum flood depth, using the both ground surface and rain water inlet sensor data. The model achieved a Mean Absolute Error (MAE) of 1.211 and Root Mean Square Error (RMSE) of 3.166 for ground surface flood depth predictions, and an MAE of 4.136 and RMSE of 7.157 for rain water inlet flood depth predictions. Notably, all correlation coefficients exceeded 0.94, indicating high prediction accuracy. The inundation prediction model developed in this study is expected to contribute to the establishment of real-time flood forecasting and response system in urban areas.

Keywords

Urban flood

Flood measurement sensor

Monitoring

Flood prediction

LSTM

최근 도시침수 피해가 매년 반복적으로 발생함에 따라 국립재난안전연구원은 2018년부터 울산, 부산, 인천, 천안, 창원 등에 침수센서를 설치하여 운영하고 있다. 본 연구에서는 다년간 수집된 도시 침수 계측데이터를 활용하여 침수 예측에 활용하기 위해 시계열 예측모델 중에서 장기 의존성 학습에 강점이 있는 LSTM을 이용하여 계측데이터 기반의 침수심 예측모델을 설계하였다. 모델의 성능분석 결과 지표 및 빗물받이 계측 데이터를 활용하여 침수 시점, 종료 시점, 최대 침수심을 정밀하게 예측하였으며, 지표 침수심의 예측 성능은 MAE 1.211, RMSE 3.166, 빗물받이 침수심의 예측 성능은 MAE 4.136, RMSE 7.157로 분석되었으며, 특히 상관계수는 0.94 이상의 높은 성능을 갖는 것으로 분석되었다. 본 연구에서 개발한 계측데이터 기반의 침수 예측모델은 향후 도시 내 실시간 침수 예측 및 대응 체계에 기여할 수 있을 것으로 기대된다.

키워드

도시침수

침수계측센서

실시간 모니터링

침수예측

LSTM

References

Chung, J., Gulcehre, C., Cho, K., and Bengio, Y. (2014). “Empirical evaluation of gated recurrent neural networks on sequence modeling.” arXiv preprint, arXiv:1412.3555.

Gujar, S., and Vishwakarma, D. (2023). Level sensors market - By type (Contact sensor, contactless sensor), by monitoring type (Continuous level, point level), by application (Liquid measuring & monitoring, fluidized solid measuring & monitoring), End-use industry & forecast, 2023-2032, accessed 31 October 2023, <https://www.gminsights.com/toc/details/level-sensors-market>.

Hochreiter, S., and Schmidhuber, J. (1997). “Long short-term memory.” Neural Computation, Vol. 9, No. 8, pp. 1735-1780. doi: 10.1162/neco.1997.9.8.1735.

10.1162/neco.1997.9.8.1735

Korea Meteology Administrate (KMA) (2024). Summary for policymakers.

Kratzert, F., Klotz, D., Herrnegger, M., Sampson, A.K., Hochreiter, S., and Nearing, G.S. (2018). “Rainfall-runoff modeling using LSTM networks.” Hydrology and Earth System Sciences, Vol. 22, No. 11, pp. 6005-6022.

10.5194/hess-22-6005-2018

Le, X.H., Ho, H.V., Lee, G., and Jung, S. (2019). “Application of Long Short-Term Memory (LSTM) neural network for flood forecasting.” Water, Vol. 11, No. 7, 1387. doi: 10.3390/w11071387.

10.3390/w11071387

National Disaster Management Research Institute (NDMI) (2022). Design of sensor-based inundation prediction model.

Seoul Metropolitan Government (2023). 2022 flood report.

United States Environmental Protection Agency (EPA) (2018). Smart data infrastructure for wet weather control and decision support. Washington, D.C., U.S.

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, Ł., and Polosukhin, I. (2017). “Attention is all you need.” Advances in Neural Information Processing Systems, Vol. 30, pp.5998-6008.

Wang, Z., Yan, W., and Oates, T. (2017). “Time series classification from scratch with deep neural networks: A strong baseline.” International Joint Conference on Neural Networks (IJCNN), Anchorage, AK, U.S., pp.1578-1585.

10.1109/IJCNN.2017.7966039

Information

Publisher :KOREA WATER RESOURECES ASSOCIATION
Publisher(Ko) :한국수자원학회
Journal Title :Journal of Korea Water Resources Association
Journal Title(Ko) :한국수자원학회 논문집
Volume : 59
No :1
Pages :77-84
Received Date : 2025-04-29
Revised Date : 2025-11-25
Accepted Date : 2025-12-02
DOI :https://doi.org/10.3741/JKWRA.2026.59.1.77

Journal of Korea Water Resources Association ISSN:2799-8746(Print) 2799-8754(Online) 한국수자원학회 논문집

All Issue