Hurricane Harvey (2017) revealed local social vulnerabilityand risk perception of hurricanes

Abstract:

On August 25 of 2017, as Hurricane Harvey churned toward the Texas coast, the low evacuation rate and underestimated hurricane strength at Houston city together led to more than 120,000 rescues reported and 80 fatalities. To investigate local social vulnerability conditions, we constructed multiple social vulnerability factors, including maximum inundation rate and local traffic connectivity during hurricane Harvey. We mapped these factors with local rescue requesting rates (self-reported data from the red cross web site), a revealed preference we selected to represent local social vulnerability, to zip-code level (in total 137 zip codes covering Houston region). We performed logistic regression to zip code level rescue requesting rates on social vulnerability factors. The local inundation rate (p<0.001), traffic condition (p<0.001), English speaking ability (p<0.001), poverty rate (p = 0.003) and young population (p = 0.001) are significantly impacting the local rescue requesting rates (shown in Table. 1). We further analyzed the potential impact of local flood risk communication policy on residents' risk perception of hurricanes. We selected whether a person is assigned in a flood zone (FZ) as a proxy for risk communication level. Though people in the FZs are naturally under higher potential flood risks, we found that the rescue requesting rates are not simply higher in FZs than outside FZs in each zip code. For the zip codes with a larger portion inside FZs, the residents in these FZs show lower rescue requesting rates than those outside FZs in the same zip code (shown in Fig. 1). When a zip code is 50% covered by FZs, the residents in these FZs show 60% less probability to ask for rescue than outside FZ. However, when a zip code is with only ~10% FZ coverage, the residents in these FZs show 100% more probability to ask for rescue than outside FZ. The results provided quantitative evidence that when more residents are risk communicated (even on a low level), they will do more to avoid hurricane risks.