A CECREH study published in Urban Climate examines NFIP-insured residential flood losses associated with the 2016 Tax Day Flood, Hurricane Harvey, and Hurricane Irma. The research was authored by Temidayo Popoola, Jesse Andrews, Kaifa Lu, Katharine Hayhoe, and Ali Nejat.
The study integrated hazard, exposure, and vulnerability indicators at the census-tract level and compared regression and classification approaches. Classification models consistently performed better than regression models for distinguishing tracts with and without observed NFIP-insured losses.
The findings show that flood-loss screening depends on the combined influence of hazard, exposure, and vulnerability rather than precipitation alone.
Cross-event testing showed stronger transfer between the two Texas flood events than between Texas and Florida. Population density and precipitation were influential across the analysis, while geographic context remained important when applying models across regions.
The results support locally validated flood-risk screening and reinforce the importance of considering community exposure and vulnerability when prioritizing mitigation and recovery planning.