Development and Application of a Flood Vulnerability Index Using Canonical Correlation Analysis

Global warming increases heavy rainfall, escalating flood damage and highlighting the importance of flood risk prediction research. The objective of this paper is to develop a new model for the prediction of flood damage risk areas, utilizing multivariate statistical methods. We used data on land characteristics by street number to build the model and flood damage data to validate the model. The variables were categorized into direct and indirect risk factors, and a direct vulnerability index and an indirect vulnerability index reflecting the weights of the variables were calculated through factor analysis and canonical correlation analysis. We combined two indices to produce the final flood vulnerability index(FVI), which was then reclassified into Stanine scores according to the risk level to improve the utility. The method was applied to Gyeonggi-do data, and its scalability was verified by predicting the FVI for Jeonbuk-do. The methodology proposed in this study confirms the ability to predict and prevent flood damage risk using the FVI without actual flooding data, and it presents a method that can effectively predict and avoid flood damage risk even in areas where flood damage data does not exist. We also implemented a dashboard to visually understand the risk of flood damage and utilize it for countermeasure planning.