Spatiotemporal heterogeneity reveals urban-rural differences in post-disaster recovery

Abstract

Abstract A post-disaster recovery process necessitates significant financial and time investment. Previous studies have found the importance of post-disaster spatial recovery heterogeneity, but the recovery heterogeneity has not been extended to the directed recovery relationships despite the significance of sequential recovery plans. Identifying a causal structure between county-level time series data can reveal spatial relationships in the post-disaster recovery process. This study uses a causal discovery method to reveal the spatiotemporal relationships between counties before, during, and after Hurricane Irma in 2017. This study proposes node aggregation methods at different time scales to obtain internally validated causal links. This paper utilizes points of interest data with daily location information from mobile phones and county-level daily nighttime light data. We find intra-regional homogeneity, inter-regional heterogeneity, and a hierarchical structure among urban, suburban, and rural counties based on a network motif analysis. Subsequently, this article suggests county-level post-disaster sequential recovery plans using the causal graph methods. These results help policymakers develop recovery scenarios and estimate the corresponding spatial recovery impacts.