All Issue

2025 Vol.58, Issue 11 Preview Page

Special Issue: 지능형 도시홍수 예측

Urban inundation prediction with an HC-SURF-Trained Gated-CNN surrogate: The Sillim drainage district case study

HC‑SURF 물리모형 정보 연계 Gated‑CNN 기반 도시 침수 예측: 신림 배수분구 사례

Hyuna Wooa
,
Sang Bo Simb
,
Hyung-Jun Kimc
,
Hakseong Kimd
,
Seong Jin Nohe*
우 현아a
,
심 상보b
,
김 형준c
,
김 학성d
,
노 성진e*
aPh.D. Student, Department of Civil Engineering, Kumoh National Institute of Technology, Gumi, Korea
bPostdoctoral Researcher, Department of Hydro Science and Engineering Research, Korea Institute of Civil Engineering and Building Technology, Goyang, Korea
cSenior Researcher, Department of Hydro Science and Engineering Research, Korea Institute of Civil Engineering and Building Technology, Goyang, Korea
dUndergraduate Student, Department of Civil Engineering, Kumoh National Institute of Technology, Gumi, Korea
eAssociate Professor, Department of Civil Engineering, Kumoh National Institute of Technology, Gumi, Korea
a금오공과대학교 토목공학과 박사과정
b한국건설기술연구원 수자원하천연구본부 박사후연구원
c한국건설기술연구원 수자원하천연구본부 수석연구원
d금오공과대학교 토목공학과 학사과정
e금오공과대학교 토목공학과 부교수
*Corresponding Author
30 November 2025. pp. 1099-1111
PDF XML Citation
Abstract

Physics-based urban inundation models yield reliable results under limited observations but are constrained by computational cost for real-time use. This study develops a Gated-Convolutional Neural Network (Gated-CNN) deep learning model trained on HC-SURF (Hyper Connected Solution for Urban Flood) simulations and quantitatively evaluates how spatial resolution (5 m, 10 m) affects predictive performance. For the August 2022 event in the Sillim drainage district (Seoul), the 5 m model achieved CSI 0.78 and R² 0.89, whereas the 10 m model delivered higher overall accuracy and stability (CSI 0.82, R²0.90). Depth-specific analysis indicated comparable skill in shallow water (<0.15 m) and lower error in deep water (≥1.0 m) for the 10 m model (RMSE 0.34 m vs 0.53 m at 5 m). Inference for a 6 h event was about 65 times faster than the physics-based model (16 s vs 1,041 s), supporting suitability for real-time operation. These results can serve as reference data for selecting spatial resolution and for developing real-time urban flood forecasting systems.

Keywords
Urban inundation
Deep learning
Gated-CNN
Physical information
HC-SURF

도시 지역에서 물리 기반 침수 해석 모형은 관측 자료가 부족한 상황에서도 신뢰성 있는 결과를 제공하지만, 높은 계산 비용으로 인해 실시간 적용에는 한계가 있다. 본 연구에서는 HC-SURF 물리모형 정보를 활용해 학습한 Gated-Convolutional Neural Network (Gated-CNN) 기반 딥러닝 모형을 구축하고, 공간 해상도(5 m, 10 m)가 성능에 미치는 영향을 평가하였다. 서울 신림 배수분구의 2022년 8월 침수 사상에 대한 검증 결과, 딥러닝 모형은 5 m 해상도에서 CSI 0.78과 R² 0.89를 보였으며, 10 m 해상도에서는 CSI 0.82와 R² 0.90으로 성능이 향상되었다. 침수심 구간별 분석에서는 얕은 구간(<0.15 m)에서 두 해상도의 성능이 유사했으나, 깊은 구간(≥1.0 m)에서는 10 m 모형의 RMSE (0.34 m)가 5 m 모형(0.53 m)보다 낮았다. 또한 딥러닝 추론 시간은 물리모형 대비 약 65배 단축되어(평균 16초 vs 1,041초), 실시간 운영에 적합한 계산 효율성을 확인하였다. 본 연구는 도시 침수 예측 모형의 해상도 선택과 실시간 예보 체계 구축을 위한 기초 자료로 활용될 수 있다.

키워드
도시침수
딥러닝
Gated-CNN
물리정보
HC-SURF
References
1

Burrichter, B., Hofmann, J., Koltermann da Silva, J., Niemann, A., and Quirmbach, M. (2023). “A spatiotemporal deep learning approach for urban pluvial flood forecasting with multi-source data.” Water, Vol. 15, No. 9, 1760.

10.3390/w15091760
2

Cao, X., Wang, B., Yao, Y., Zhang, L., Xing, Y., Mao, J., Zhang, R., Fu, G., Borthwick, A.G.L., and Qin, H. (2025). “U-RNN high-resolution spatiotemporal nowcasting of urban flooding.” Journal of Hydrology, Vol. 659, 133117.

10.1016/j.jhydrol.2025.133117
3

Choi, H., Lee, S., Woo, H., and Noh, S. (2023). “High-resolution urban flood modeling using cellular automata-based WCA2D in the Oncheon-cheon catchment in Busan, South Korea.” Journal of Civil and Environmental Engineering Research, Vol. 43, No. 5, pp. 587-899.

10.12652/KSCE.2023.43.5.0587
4

Guo, Z., Leitão, J.P., Simões, N.E., and Moosavi, V. (2021). “Data-driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networks.” Journal of Flood Risk Management, Vol. 14, No. 1, e12684.

10.1111/jfr3.12684
5

Han, H., Kang, N., Yoon, J., and Hwang, S. (2024). “Utilizing deep learning algorithm and high-resolution precipitation product to predict water level variability.” Journal of Korea Water Resources Association, Vol. 57, No. 7, pp. 471-479.

10.3741/JKWRA.2024.57.7.471
6

Hofmann, J., and Schüttrumpf, H. (2021). “floodGAN: Using deep adversarial learning to predict pluvial flooding in real time.” Water, Vol. 13, No. 16, 2255.

10.3390/w13162255
7

Hosseiny, H. (2021). “A deep learning model for predicting river flood depth and extent.” Environmental Modelling & Software, Vol. 145, 105186.

10.1016/j.envsoft.2021.105186
8

Jeong, M., Kim, K., Lee, C.K., and Kim, H.W. (2025). “Flood risk prediction using the grid-based machine learning models in Incheon.” Land and Housing Review, Vol. 16, No. 2, pp. 35-48.

9

Kabir, S., Patidar, S., Xia, X., Liang, Q., Neal, J., and Pender, G. (2020). “A deep convolutional neural network model for rapid prediction of fluvial flood inundation.” Journal of Hydrology, Vol. 590, 125481.

10.1016/j.jhydrol.2020.125481
10

Kim, S., Kim, H.-J., and Yoon, K.-S. (2021). “Development of artificial intelligence-based river flood level prediction model capable of independent self-warning.” Journal of Korea Water Resources Association, Vol. 54, No. 12, pp. 1285-1294.

10.3741/JKWRA.2021.54.12.1285
11

Lee, S., Nakagawa, H., Kawaike, K., and Zhang, H. (2016). “Urban inundation simulation considering road network and building configurations.” Journal of Flood Risk Management, Vol. 9, No. 3, pp. 224-233.

10.1111/jfr3.12165
12

Li, J., Pan, G., Chen, Y., Wang, X., Huang, P., Zhang, L., and Zhou, H. (2025). “Rapid-mapping maximum water depth map of urban flood using a highly adaptable machine learning based model.” Journal of Flood Risk Management, Vol. 18, No. 3, e70095.

10.1111/jfr3.70095
13

Lim, Y., Kim, D., Ji, Y., Young, S.J., and Kang, B. (2025). “Optimization of physical and hyperparameters for enhancing the response performance of an LSTM-based virtual flood level sensor.” Journal of Korea Water Resources Association, Vol. 58, No. 3, pp. 225-239.

10.3741/JKWRA.2025.58.3.225
14

Löwe, R., Böhm, J., Jensen, D.G., Leandro, J., and Rasmussen, S.H. (2021). “U-FLOOD - Topographic deep learning for predicting urban pluvial flood water depth.” Journal of Hydrology, Vol. 603, 126898.

10.1016/j.jhydrol.2021.126898
15

Ministry of Environment (ME) (2022). Flood control plan report for specific river basins (Dorim Stream, Siheung Stream).

16

Noh, S.J., Lee, J.-H., Lee, S., Kawaike, K., and Seo, D.-J. (2018). “Hyper-resolution 1D-2D urban flood modelling using LiDAR data and hybrid parallelization.” Environmental Modelling & Software, Vol. 103, pp. 131-145.

10.1016/j.envsoft.2018.02.008
17

Pianforini, M., Dazzi, S., Pilzer, A., and Vacondio, R. (2024). “Real-time flood maps forecasting for dam-break scenarios with a transformer-based deep learning model.” Journal of Hydrology, Vol. 635, 131169.

10.1016/j.jhydrol.2024.131169
18

Seleem, O., Ayzel, G., Bronstert, A., and Heistermann, M. (2023). “Transferability of data-driven models to predict urban pluvial flood water depth in Berlin, Germany.” Natural Hazards and Earth System Sciences, Vol. 23, No. 2, pp. 809-822.

10.5194/nhess-23-809-2023
19

Sim, S.B., and Kim, H.-J. (2024a). “An efficiency analysis of urban inundation simulation by urban building reflection techniques.” Journal of the Korean Society of Safety, Vol. 39, No. 6, pp. 53-60.

20

Sim, S.B., and Kim, H.-J. (2024b). “An urban flood model development coupling the 1D and 2D model with fixed-time synchronization.” Water, Vol. 16, No. 19, 2726.

10.3390/w16192726
21

Tak, Y.H., Kim, Y.D., Kang, B., and Park, M.H. (2016). “Sewer overflow simulation evaluation of urban runoff model according to detailed terrain scale.” Journal of Korea Water Resources Association, Vol. 49, No. 6, pp. 519-528.

10.3741/JKWRA.2016.49.6.519
22

Woo, H. (2025). Estimating urban inundation using physics-informed deep learning. Master Thesis, Kumoh National Institute of Technology.

23

Woo, H., Choi, H., Kim, M., and Noh, S.J. (2024). “Estimating urban inundation using physics-informed deep learning: A case study of the Oncheon-cheon catchment.” Journal of Korea Water Resources Association, Vol. 57, No. 12, pp. 989-1001.

10.3741/JKWRA.2024.57.12.989
24

Xu, L., and Gao, L. (2024). “A hybrid surrogate model for real-time coastal urban flood prediction: An application to Macao.” Journal of Hydrology, Vol. 642, 131863.

10.1016/j.jhydrol.2024.131863
25

Zhou, Y., Wu, W., Nathan, R., and Wang, Q.J. (2021). “A rapid flood inundation modelling framework using deep learning with spatial reduction and reconstruction.” Environmental Modelling & Software, Vol. 143, 105112.

10.1016/j.envsoft.2021.105112
Information
  • Publisher :KOREA WATER RESOURECES ASSOCIATION
  • Publisher(Ko) :한국수자원학회
  • Journal Title :Journal of Korea Water Resources Association
  • Journal Title(Ko) :한국수자원학회 논문집
  • Volume : 58
  • No :11
  • Pages :1099-1111
  • Received Date : 2025-09-08
  • Revised Date : 2025-10-02
  • Accepted Date : 2025-10-10
  • DOI :https://doi.org/10.3741/JKWRA.2025.58.11.1099
Journal Informaiton Journal of Korea Water Resources Association Journal of Korea Water Resources Association
Online Submission
Archives
  • scopus
  • KCI-수자원
  • NRF
  • KOFST
  • KISTI Current Status
  • KISTI Cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close