Hurricane Evacuation Analysis with Large-scale Mobile Device Location Data during Hurricane Ian

TL;DR

This paper analyzes hurricane evacuation behaviors during Hurricane Ian using large-scale mobile device location data, revealing evacuation patterns, timing, and behavioral groups to improve disaster response strategies.

Contribution

Introduces a comprehensive algorithm for analyzing hurricane evacuation behavior using terabyte-scale GPS data, including categorization and spatiotemporal analysis during Hurricane Ian.

Findings

Lower out-of-zone evacuation in landfall area

Three distinct evacuation waves identified

Delayed evacuation orders influence behavior patterns

Abstract

Hurricane Ian is the deadliest and costliest hurricane in Florida's history, with 2.5 million people ordered to evacuate. As we witness increasingly severe hurricanes in the context of climate change, mobile device location data offers an unprecedented opportunity to study hurricane evacuation behaviors. With a terabyte-level GPS dataset, we introduce a holistic hurricane evacuation behavior algorithm with a case study of Ian: we infer evacuees' departure time and categorize them into different behavioral groups, including self, voluntary, mandatory, shadow and in-zone evacuees. Results show the landfall area (Fort Myers, Lee County) had lower out-of-zone but higher overall evacuation rate, while the predicted landfall area (Tampa, Hillsborough County) had the opposite, suggesting the effects of delayed evacuation order. Out-of-zone evacuation rates would increase from shore to inland. Spatiotemporal analysis identified three evacuation waves: during formation, before landfall, and after landfall. These insights are valuable for enhancing future disaster planning and management.