Living with and without water: modeling human-infrastructure interactions in disaster preparedness

Abstract When disasters strike, people rush to secure essential goods. This paper introduces a system dynamic model that elucidates the interplay between human and infrastructure systems during disaster preparation. Harnessing human mobility data related to household visits to grocery stores for bottled water, the model captures the dependence of human activities on transportation networks and grocery services. The model also accounts for flood disruptions and interactions among households, where households’ visits to the same facilities may engage in stockpiling behavior. Tested in Wilmington, DE, during simulated Nor’easter-like flooding, our findings reveal that physical vulnerabilities such as road and facility inundation and isolation hinder access to critical facilities, resulting in diminished resource availability for communities. Moreover, social vulnerability exacerbates resource disparities, with affluent households stockpiling and affecting socially vulnerable counterparts except those benefiting from geographical advantages in accessing facilities. Despite sufficient average resources at the census block level, many households receive insufficient resources for two consecutive days, underscoring the importance of household-level disaster preparedness assessments. Additionally, probabilistic stockpiling behavior further influences resource distribution across communities. Our model can be expanded with empirical studies using granular household data to capture unobserved heterogeneity better and enhance disaster supply pre-positioning, targeted support, and resilience planning.