Understanding the Influence of Multiple Information Sources on Risk Perception Dynamics and Evacuation Decisions : An Agent-Based Modeling Approach

Evacuation is one of the most effective strategies to reduce risk during natural disasters. Although many studies have investigated factors affecting evacuation decisions (e.g., socio-economic, risk forecast, demographic, environmental cues, etc.), how all these factors interplay in forming the overall risk perception and evacuation decision is underexplored. Here, we present an agent-based model that leverages empirical evidence from post-disaster surveys to create evacuation tendencies of the agents (households) and integrates flood-related hazard risk, household socio-demographic factors, social network characteristics and decisions, and neighbors’ decisions to dynamically evolve complex decision-making pattern during hurricane evacuation. We simulate the effect of multiple information sources and their perceived credibility on evacuation participation by separately determining evacuation compliance (defined as evacuation from places where official mandatory evacuation order is issued) and shadow evacuation (defined as spontaneous evacuation from places where a mandatory order is not issued). We use socio-demographic data from United States Census Bureau, building footprint data from County Website, run off data from Stormwater Management Model (SWMM) of U.S. Environmental Protection Agency, and all other data were synthetically generated within the model. Our case study based on a community of Miami-Dade County indicates that higher perceived credibility of the hazard risk leads to higher evacuation compliance. We have also found that while social networks increase overall evacuation participation, they also increase shadow evacuation and higher trust on neighbor actions reduces evacuation rates. This study gives insights for emergency management to design appropriate strategies for providing hazard forecasts or communicating overall risk during disasters