We will discuss a project that aims to understand and improve hurricane risk communication and protective decision making by investigating how evolving, uncertain weather forecasts and warnings interact with social information flow and decisions as a hurricane approaches and arrives. To accomplish this goal, the research integrates study of the real-world hazard information system with computational physical and social modeling, using knowledge and methods from atmospheric and related sciences, computer and information science, and social and behavioral sciences. To understand the real-world system, we are examining how people communicate and respond to information about hurricane threats by analyzing data collected from social media, complemented by focus groups with more vulnerable populations. This is combined with computational modeling research that includes hurricane/storm-surge modeling and agent-based modeling of social actors who pursue, process, and transmit information.
We will present the project goals and design along with initial results from the research, with an emphasis on agent-based modeling experiments. The project has developed version 1 of a spatially-explicit agent-based model that is informed by knowledge about hurricane prediction, communication, and decision making. During the days leading up to landfall, the model inputs historical storm information and forecasts, which are passed among different types of agents (e.g., broadcasters, public officials, members of the public) through information networks. That information is processed by public agents who, as the storm approaches, decide whether to gather information more or less frequently and whether to take protective action. Using the agent-based model, we examine how different hurricane forecast and warning information, with different levels of uncertainty and skill, propagates through the informational system and influences patterns of evacuation decision making.