RAPID/Collaborative Research Examining Household Movements and Evacuation Decision-Making in a Compounding Risk Event

Description of Project and Data: This study examines evacuation and sheltering-in-place decisions of households before and after two simultaneous disasters: Hurricane Ida and COVID-19 pandemic. Building upon the previous work examining protective action decision-making during disasters, and through the extension of the PADM framework to account for the presence of simultaneous risks, this research advances understanding and offers new insights pertaining to household decision-making in dual risk environments. The research team developed a questionnaire that was administered by Qualtrics eight months following Hurricane Ida making landfall in Southeastern Louisiana. Surveys were administered via internet and phone to six parishes in southeast Louisiana: Orleans, Jefferson, St. Charles, St. John the Baptist, Terrebonne, and St. Bernard Parish. Data was collected beginning April 29th, 2022, and concluding June 1, 2022. 

The survey questionnaire consisted of four sections: (1) questions inquiring about hurricane and COVID-19 risk perceptions, (2) questions pertaining to what protective actions were undertaken (e.g. evacuate shelter-in-place), along with additional information about the timing of decisions and evacuation destinations (3) questions asking about the extent of damage sustained to their properties as well as their experiences with secondary hazards stemming from the hurricane and, later, Tropical Storm Nicholas, and (4) questions about their household characteristics, vaccination status, and previous evacuation and sheltering-in-place experiences.

This project also examines evacuation and recovery patterns during compound risk events. The data here includes home locations of mobile phone users estimated from their trajectories from location-based services (LBS) geolocation data, computed for each half-day (12 hours). Using these time-varying home locations, their status of evacuation from and return to home is determined. This project is important in big passively-collected data-based studies as it introduces concepts of distance-based clustering, home detection, and large-scale high-resolution GPS data analysis for decision-making, primarily for academics but also to some extent to disaster modelers and policymakers. The data files are stored in the Apache Parquet format which can be accessed via tools like Python, R, MySQL, SQLite, etc. The description of the fields in these files is provided in ReadMe.txt.