Authors : Njoku Peter Chinonyerem; Jonathan Nyekachi Amadi; Nbaakee Lebari Goodday

Volume/Issue : Volume 10 - 2025, Issue 9 - September

Google Scholar : https://tinyurl.com/mr6j5nak

Scribd : https://tinyurl.com/2s45592k

DOI : https://doi.org/10.38124/ijisrt/25sep1293

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Note : Google Scholar may take 30 to 40 days to display the article.

Abstract : Urban flooding remains one of the most pressing environmental and socioeconomic challenges in Nigeria, particularly in rapidly growing cities such as Lagos, Port Harcourt, and Ibadan. Conventional flood prediction approaches, often limited to hydrological and meteorological data, fail to capture the complexity introduced by urbanization, socioeconomic inequalities, and inadequate infrastructure. To address these gaps, this study develops a hybrid Artificial Intelligence (AI) framework that integrates spatial imagery with socioeconomic and climatic variables to improve urban flood risk prediction. The methodology combines Convolutional Neural Networks (CNNs) for analyzing geospatial and satellite imagery with Gradient Boosting Machines (GBMs) for modeling non-visual features, including poverty index, housing density, and rainfall intensity. A meta-learner ensemble strategy, using logistic regression, was employed to optimally fuse the predictions from both models. Comparative experiments were conducted to evaluate CNN-only, GBM-only, and hybrid ensemble models across multiple Nigerian cities, followed by visualization through flood risk maps and feature importance rankings. The findings demonstrate that the hybrid ensemble significantly outperformed individual models, achieving higher prediction accuracy and generalization. The integration of socioeconomic factors not only improved the model’s sensitivity to high-risk zones but also revealed critical drivers of vulnerability, such as unplanned housing and poor drainage systems. Case studies on Lagos Island and Port Harcourt showed that the hybrid model provided more realistic and actionable predictions compared to hydrology-only approaches. Flood risk maps effectively identified high, medium, and low-risk areas, offering valuable insights for targeted disaster response. This research highlights the potential of AI- driven hybrid modeling as a transformative tool for urban flood management in Nigeria. By integrating geospatial and socioeconomic intelligence, the framework enables data-informed policymaking, urban planning, and disaster preparedness. Future work should prioritize real-time flood alert systems and mobile-based decision support tools, ensuring that predictive insights translate into timely, community-level action.

Keywords : Floods, Disaster Management, Machine Learning, Convolutional Neural Networks, Gradient Boosting Machines, Spatial Analysis.

References :

1. “Social Economic and Health Implications of Recent Spate of Flooding in Nigeria,” International Journal of Research and Innovation in Social Science, 2022.
2. “2022 Nigeria floods,” Wikipedia, Oct. 2022.
3. “2023 Nigeria floods,” Wikipedia, Dec. 2023.
4. “2024 Lekki flood,” Wikipedia, Jul. 2024.
5. Lucas, Urban flood risks, impacts, and management in Nigeria, PreventionWeb.
6. Climate change and urban flooding: implications for Nigeria’s built environment, MedCrave Online.
7. Benin City environmental issues and flood management, Wikipedia, 2025.
8. “Impacts of flood disasters in Nigeria: A critical evaluation of health implications and management,”
9. “At least 115 killed in floods in central Nigeria,” The Guardian, May 2025.
10. “Teacher in Nigeria loses dozens of relatives and pupils in devastating floods,” AP News, Jun. 2025.
11. J. Rentschler, M. Salhab, and B. Sinha, “Flood exposure and poverty in 188 countries,” Nature Communications, vol. 13, no. 3527, 2022.
12. S. Hallegatte, J. Rentschler, et al., People in Harm’s Way: Flood Exposure and Poverty in 189 Countries. Washington, DC: World Bank, 2022.
13. IPCC, “Chapter 11: Weather and Climate Extreme Events in a Changing Climate,” in AR6 WG1, Cambridge Univ. Press, 2021.
14. IPCC, AR6 Synthesis Report: Summary for Policymakers. Geneva: IPCC, 2023.
15. IPCC, “Chapter 12: Climate Change Information for Regional Impact Assessment,” in AR6 WG1, Cambridge Univ. Press, 2021.
16. World Bank, “Flood risk already affects 1.81 billion people—climate change and unplanned urbanization are intensifying the danger,” 2022.
17. D. Bonafilia, B. Tellman, T. Anderson, and E. Issenberg, “Sen1Floods11: A georeferenced dataset to train and test deep learning flood algorithms for Sentinel-1,” in Proc. CVPR Workshops, 2020.
18. D. Bonafilia et al., “Sen1Floods11… (Open Access),” CVPRW 2020, paper and dataset.
19. A. Riche, A. Richards, and G. S. Lin, “A novel hybrid deep-learning approach for flood detection: RF, CNN, U-Net, and Res-U-Net comparison,” Remote Sensing, vol. 16, no. 19, 3673, 2024.
20. S. Kabir et al., “A deep CNN model for rapid prediction of fluvial flood inundation,” arXiv:2006.11555, 2020.
21. C. Meng et al., “A comparison of ML models for predicting flood susceptibility using NHAND,” Sustainability, vol. 15, no. 20, 14928, 2023.
22. T. Islam et al., “A systematic review of urban flood susceptibility mapping,” Remote Sensing, vol. 17, no. 3, 524, 2025.
23. M. S. Al-Sheriadeh et al., “Robustness of ML algorithms for flood susceptibility mapping; RF outperforms,” Georisk, 2024.
24. S. T. Seydi et al., “Comparison of ML algorithms for flood susceptibility mapping (Cascade Forest vs. others),” Remote Sensing, vol. 15, no. 1, 192, 2022.
25. S. Mangkhaseum et al., “Flood susceptibility mapping leveraging open-source remote-sensing datasets and ML,” Georisk, 2024.