Flood Vulnerability Analysis for Illinois Using a Decision-Tree Based Model for Social Vulnerability

With the rise of urbanization and global climate change, there has been an increase in the occurrence of natural disasters over the past 50 years. While part of this rise is due to natural events, the resulting disasters are not the outcome of geophysical occurrences by themselves. Social vulnerability describes the human aspect of vulnerability and works to identify the social conditions that leave societies susceptible to disasters. In this study, the decision tree (DT) methodology is proposed as an alternative to assess flood vulnerability in Illinois, specifically in the Illinois River Basin, which is heterogenous in demographics and land use. We also derive a PCA index using the performance optimization methodology developed by the University of Illinois at Chicago - School of Public Health’s (UIC-SPH) team. The predictive performance of the two derived indices and existing flood vulnerability indices for Illinois is evaluated using National Flood Insurance Program claim loss data as realized loss. The Flood Vulnerability Index derived using DT methodology (FVI.DT) has the highest overall predictive performance (87.2%). Existing FVIs for Illinois significantly under- and overestimate county level vulnerability, which implies gaps in mitigation planning and resource allocation throughout the state.