Coastal Flood Risk Prediction Using GIS and Machine Learning in Balayan, Batangas Province

Abstract. Recent coastal flooding in Balayan, Batangas, driven by Typhoon Neneng (NESAT, 2022), Severe Tropical Storm Kristine (Trami, 2024), and Typhoon Carina alongside the enhanced southwest monsoon (2024) which highlights the increasing vulnerability of coastal communities to extreme weather events. With climate change accelerating, there is an urgent need for data-driven flood risk assessment methods that integrate both environmental and socio-economic factors. This study employs a Geospatial Artificial Intelligence (GeoAI) workflow leveraging the Random Forest algorithm to estimate weights for key flood risk indicators within the IPCC risk framework, which conceptualizes risk as a function of hazard, exposure, and vulnerability. The model incorporates diverse hydroclimatic and socio-environmental variables, such as elevation, slope, projected population density, proximity to mangroves and seagrasses, distance to the coastline, and sea level rise projections under RCP 8.5 that represents a high-emissions climate scenario. To ensure consistency and transferability, all input variables were normalized to a 0–1 scale via min-max scaling, and standardized risk thresholds were applied. The GeoAI workflow was trained on historical flood data spanning 2022–2024 and validated under future climate conditions to minimize subjectivity and data imbalance while enhancing robustness. The Random Forest (RF) model achieved high predictive performance, with a mean cross-validation accuracy of 98.62% and test accuracy of 98.73%, which effectively discriminates flooded from non-flooded zones. Furthermore, this workflow supports near-term planning by providing spatially explicit risk maps and quantifying potential impacts such as projected agricultural losses, thereby informing climate-resilient adaptation strategies and policymaking.