Intelligent Flood Risk Management in Jeju Island: Grid-Based AI Prediction and Assessment

Abstract

Jeju Island, a volcanic island in Korea, exhibits distinct hydrogeomorphological features that pose challenges for disaster management. Traditional hydrological models and qualitative risk assessment methods developed for inland areas are insufficient for addressing Jeju Island’s unique conditions. A localized approach is necessary to integrate hydrogeomorphological and urban drainage factors with hazard-related data. This study develops a grid-based AI flood risk assessment method tailored to Jeju Island’s characteristics, considering urban drainage systems to improve prediction accuracy. The objectives are to: (1) establish a grid-type flood damage and flood risk influencing factor database that reflects Jeju Island's unique hydrogeomorphological and hazard characteristics, (2) create a binary classification deep learning model integrating hydrological, hazard, and urban drainage factors to assess flood risk, and (3) propose warning standards based on flood damage data and risk assessments, accounting for the specific disaster patterns of Jeju Island. This study seeks to overcome the limitations of existing disaster management approaches by leveraging AI and hydrological analysis technologies, addressing Jeju Island's increasing vulnerability to disasters. Through the integration of grid-based data and advanced ML models, this research aims to provide a comprehensive framework for predicting and managing flood risks, ultimately enhancing resilience and safety on the area.