Understanding Hurricane Evacuation Behavior from Facebook Population Data during Hurricane Ian

The frequency and intensity of hurricanes are increasing in coastal regions of USA posing a great threat to human lives. During these crisis events, emergency management officials need to identify vulnerable areas to move people to safe places within a short timeframe. To effectively plan and manage evacuations, it is crucial to understand evacuation trends in real time for proper allocation of limited resources. This study leverages an emerging data source available at a high spatial resolution to understand geospatial patterns of evacuation behavior in a rapidly intensifying hurricane. In this study, Facebook population data have been analyzed to understand spatiotemporal trends of evacuation behavior for Pinellas County, FL during evacuation periods of Hurricane Ian. After retrieving the Facebook population data at a census tract level, we have calculated evacuation rates of each census tract that contains an evacuation zone. The study has also explored various factors such as socio-demographic characteristics, type of evacuation zones, proximity to shelter locations etc. to examine the change of evacuation rates via a logistic regression model. The results reveal higher evacuation rates for census tracts located far from shelter locations, with people of high household income, and with high percentage of old, black, and Hispanic people. Evacuation rates were higher in high-risk evacuation zones compared to low-risk zones. This study illustrates the application of an emerging and widely available data source such as Facebook population data to understand evacuation behavior patterns in an unfolding hurricane at a high spatial resolution without compromising user privacy.