Diffusion of hurricane evacuation behavior through a home-workplace social network: A spatially explicit agent-based simulation model

With coastal populations experiencing the growing threat of hurricanes as a consequence of global climate change, models for predicting how household evacuation behavior may diffuse over time and space are needed for emergency management. This study models the effects of social influence on household evacuation behavior in anticipation of a hurricane event. An agent-based model was developed in this study to simulate: 1) a home-workplace social network between households residing in the Florida Keys, 2) the communication of a hurricane evacuation order among socially linked households, and 3) the resulting spatio-temporal diffusion of household evacuation behavior. Data sources informing model implementation include U.S. Census block group data, business databases, and statistics from hurricane evacuation surveys. Simulated model results from the model were validated with empirical traffic records observed at a Florida Keys monitoring station during evacuation from Hurricane Georges in 1998. This model builds upon previous research using agent-based models to simulate hurricane evacuation by incorporating multiple data sources and validating results with empirical traffic patterns. Such an empirically-grounded model facilitates locally relevant exploration of evacuation behavior to support the development of more effective evacuation plans and preparedness for future hurricane events.