Attributed graph neural network-based flood vulnerability assessment of river basin systems

Abstract

Flood vulnerability assessment is a critical component of flood risk management, particularly in regions with complex river networks and limited hydrological data. This study proposes a graph neural network-based framework to evaluate river system vulnerability, representing each river segment as a node with hydrological and geomorphological attributes. Two GNN models were employed to generate vulnerability scores by jointly considering node attributes and network structure. High-risk river segments were first identified based on these scores, and the results were then aggregated to delineate flood-sensitive sub-basins. A case study in Guangxi, China, using the Xijiang River system, shows that the two models converge on similar high-risk areas, with an overlap of 60% in identified high-risk segments, aligning well with observed flood patterns. This highlights the robustness and practical reliability of the proposed approach. The framework offers a practical, data-efficient tool for identifying vulnerable river segments and flood-prone sub-basins, supporting flood risk management and decision-making in complex river systems.

Keywords

Vulnerability assessment / River system / Graph neural network (GNN) / Flood risk management / Basin-scale analysis

References

[1] Abedi R, Costache R, Shafizadeh-Moghadam H, Pham QB. Flash-flood susceptibility mapping based on XGBoost, random forest and boosted regression trees. Geocarto Int. 2022, 37: 5479-5496.

[2] Ajin RS, Senan CPPC, Devi BRA, Costache R, Nagar JK, Rajaneesh Aet al. . Flood risk mapping in an urbanized tropical river basin in India using MCDA-AHP: a post-storm event evaluation. Smart Construction and Sustainable Cities. 2025, 3. 10

[3] Allstadt KE, McBride SK, Godt JW et al (2025) Preliminary field report of landslide hazards following Hurricane Helene. US Geological Survey

[4] Aparcedo A, Lopez C, Kotta A, Li M (2024) Multimodal Power Outage Prediction for Rapid Disaster Response and Resource Allocation. 2024 IEEE 52nd Photovoltaic Specialist Conference (PVSC). IEEE 1093–1099

[5] Beijing Daily O. (The People's Government of Beijing Municipality) (2025) Beijing’s Flood Control Press Conference Held; Over 300,000 People Affected by the Flood Disaster. https://www.beijing.gov.cn/ywdt/gzdt/202508/t20250801_4163942.html. Accessed. (Chinese)

[6] Bentivoglio R, Isufi E, Jonkman SN, Taormina R. Rapid spatio-temporal flood modelling via hydraulics-based graph neural networks. Hydrol Earth Syst Sci. 2023, 27: 4227-4246.

[7] Cai B, Yu Y. Flood forecasting in urban reservoir using hybrid recurrent neural network. Urban Clim. 2022, 42: 101086.

[8] Cai H, Zhao W, Njock PGA. Multi-tier scheduling algorithm of dispatching systems for urban water logging. Smart Construction and Sustainable Cities. 2024, 23.

[9] Chang F-J, Chiang Y-M, Chang L-C. Multi-step-ahead neural networks for flood forecasting. Hydrol Sci J. 2007, 52114-130.

[10] Cho M, Kim C, Jung K, Jung H. Water level prediction model applying a long short-term memory (lstm)–gated recurrent unit (gru) method for flood prediction. Water. 2022, 14: 2221.

[11] Ding K, Li J, Bhanushali R, Liu H (2019) Deep anomaly detection on attributed networks. Proceedings of the 2019 SIAM international conference on data mining. SIAM 594–602

[12] Fan H, Zhang F, Li Z (2020) Anomalydae: Dual autoencoder for anomaly detection on attributed networks. ICASSP 2020–2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE 5685–5689

[13] Feizi H, Sattari MT. Streamflow forecasting based on PatchTST, LSTM, and ensemble learning approaches. Water Resour Manage. 2026, 4044.

[14] Gai Y, Wang M, Wu Y, Wang E, Deng X, Liu Yet al. . Simulation of spring discharge using graph neural networks at Niangziguan Springs, China. J Hydrol. 2023, 625130079.

[15] Hou Y, Wei Y, Wu S, Li J. Mapping the social, economic, and ecological impact of floods in Brisbane. Water. 2023, 153842.

[16] Joshi B, Singh VK, Vishwakarma DK, Ghorbani MA, Kim S, Gupta Set al. . A comparative survey between cascade correlation neural network (CCNN) and feedforward neural network (FFNN) machine learning models for forecasting suspended sediment concentration. Sci Rep. 2024, 1410638.

[17] Lehner B, Grill G. Global river hydrography and network routing: baseline data and new approaches to study the world's large river systems. Hydrol Process. 2013, 272171-2186.

[18] Li B. From maps to decisions: A GeoAI framework for multi-hazard infrastructure resilience and equitable emergency management. Am J Big Data. 2025, 6: 139-153.

[19] Li J, Zheng A, Guo W, Bandyopadhyay N, Zhang Y, Wang Q. Urban flood risk assessment based on DBSCAN and K-means clustering algorithm. Geomat Nat Hazards Risk. 2023, 14: 2250527.

[20] Li W, Kiaghadi A, Dawson C. Exploring the best sequence LSTM modeling architecture for flood prediction. Neural Comput Appl. 2021, 335571-5580.

[21] Li Z, Liu H, Zhang C, Fu G. Real-time water quality prediction in water distribution networks using graph neural networks with sparse monitoring data. Water Res. 2024, 250: 121018.

[22] Liu X, Lyu H-M, Shen S-L (2025) Assessment of geo-disaster risk levels induced by extreme rainfall using integrated FCM-VIKOR approach. Georisk: Assess Manag Risk Eng Syst GeohazardS 1–20

[23] Liu Y, Ao X, Feng F, Ma Y, Li K, Chua T-S, et al (2023) FLOOD: A flexible invariant learning framework for out-of-distribution generalization on graphs. Proceedings of the 29th ACM SIGKDD conference on knowledge discovery and data mining. 1548–1558

[24] Lyu H-M, Yin Z-Y, Zhou A, Shen S-L. MCDM-based flood risk assessment of metro systems in smart city development: A review. Environ Impact Assess Rev. 2023, 101107154.

[25] Lyu H-M, Yin Z-Y, Zhou A, Shen S-L. Sensitivity analysis of typhoon-induced floods in coastal cities using improved ANP-GIS. Int J Disaster Risk Reduct. 2024, 104104344.

[26] Lyu H, Yin Z, Shen S, Chen X, Su D. A novel framework incorporating machine learning into GIS for flood susceptibility prediction of urban metro systems. SCIENCE CHINA Technol Sci. 2025, 681620701.

[27] Malla S, Ohgushi K. Flood vulnerability map of the Bagmati River basin, Nepal: a comparative approach of the analytical hierarchy process and frequency ratio model. Smart Construction Sustain Cities. 2024, 216.

[28] Meenal R, Michael PA, Pamela D, Rajasekaran E. Weather prediction using random forest machine learning model. Indones J Electr Eng Comput Sci. 2021, 221208-1215

[29] Potapov P, Hansen MC, Pickens A, et al (2022) The Global 2000-2020 Land Cover and Land Use Change Dataset Derived From the Landsat Archive: First Results. Front Remote Sens https://doi.org/10.3389/frsen.2022.856903.

[30] Roman-Stork HL, Subrahmanyam B. The impact of the Madden–Julian Oscillation on Cyclone Amphan (2020) and southwest monsoon onset. Remote Sens. 2020, 12: 3011.

[31] Saber M, Boulmaiz T, Guermoui M, Abdrabo KI, Kantoush SA, Sumi Tet al. . Examining LightGBM and CatBoost models for wadi flash flood susceptibility prediction. Geocarto Int. 2022, 377462-7487.

[32] Santos LBL, Escobar-Silva EV, Satolo LF, Oyarzabal RS, Diniz MM, Negri RGet al. . Machine learning-based hydrological models for flash floods: a systematic literature review. Smart Construction Sustain Cities. 2025, 3: 21.

[33] Schoppa L, Disse M, Bachmair S. Evaluating the performance of random forest for large-scale flood discharge simulation. J Hydrol. 2020, 590125531.

[34] Shao X, Zhang X, Gao Y, Li B (2025) A method for fast prediction of flood evolution process based on graph neural network. Comput Graph 104471

[35] Shen S-L, Du Y, Moqsud A, Jia R. Lowlands: smart disaster prevention and sustainability. Smart Construction Sustain Cities. 2024, 220.

[36] Sun AY, Jiang P, Yang Z-L, Xie Y, Chen X. A graph neural network approach to basin-scale river network learning: the role of physics-based connectivity and data fusion. Hydrol Earth Syst Sci Discuss. 2022, 20221-35

[37] Supriyadi B, Windarto AP, Soemartono T. Classification of natural disaster prone areas in Indonesia using K-means. Int J Grid Distrib Comput. 2018, 11: 87-98.

[38] Taghizadeh M, Zandsalimi Z, Shafiee-Jood M, Alemazkoor N (2025) Multi-fidelity graph neural networks for efficient and accurate flood hazard mapping. Environ Model Softw 106654

[39] Tehrany MS, Pradhan B, Mansor S, Ahmad N. Flood susceptibility assessment using GIS-based support vector machine model with different kernel types. CATENA. 2015, 125: 91-101.

[40] Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, et al (2017) Attention is all you need. Adv Neural Inform Process Syst 30

[41] Xu H, Ma C, Lian J, Xu K, Chaima E. Urban flooding risk assessment based on an integrated k-means cluster algorithm and improved entropy weight method in the region of Haikou, China. J Hydrol. 2018, 563: 975-986.

[42] Xu Y, Zhao J, Wan B, Cai J, Wan J. Flood forecasting method and application based on informer model. Water. 2024, 16: 765.

[43] Zhao G, Pang B, Xu Z, Peng D, Xu L. Assessment of urban flood susceptibility using semi-supervised machine learning model. Sci Total Environ. 2019, 659940-949.

[44] Zheng Q, Zhou A, Shen S-L. Mapping bushfire risk based on scale division and factor analysis: a case study from Victoria, Australia. Int J Disaster Risk Reduct. 2025, 118105222.

[45] Zhou X-H, Zhou A, Shen S-L. How to mitigate the impact of climate change on modern cities: lessons from extreme rainfall. Smart Construction and Sustainable Cities. 2023, 17.

RIGHTS & PERMISSIONS

The Author(s)