Where do evacuees go? Understanding evacuation destinations in Georgia during Hurricane Milton

This study proposes a methodological framework to analyze spatial and temporal patterns of Florida residents' evacuation to Georgia during Hurricane Milton, focusing on changes in Points-of-Interest (POIs) visitor share within Georgia census tracts before and during the hurricane. Using mobile GPS-based mobility data, we track weekly POI visit volumes linked to visitors' home census tracts. An origin-destination matrix is constructed to assess where evacuees went, and how destinations varied by the severity of evacuation orders. Quantile regression results reveal a U-shaped pattern in explanatory power, with stronger model fit at the lower and upper quantiles, indicating that large positive or negative changes in visitor share are better explained by the model.

Social ties are significant predictors of destination choice, especially in the upper quantiles of tracts with increased visitor share, with amplified effects among evacuees from tracts under mandatory evacuation. Distance from Atlanta's airport and major urban centers were negative predictors of visitor share change. The presence of clustered hotels serves as a consistent pull factor, while malls play a dual role, attracting evacuees in upper quantiles and deterring them in others. These findings highlight the need for disaster planning strategies that extend beyond administrative shelter assignments to incorporate actual evacuation behaviors and relational geographies. Policy recommendations include pre-identifying high-demand destinations, strengthening service capacity in rural receiving areas, and integrating social ties data into evacuation forecasting models to improve regional resilience.