Hurricane evacuation is a complex process driven by human decision-making under uncertainty and governed by social, technical, spatial, and temporal constraints. This project takes a holistic approach to modeling the process by integrating behavioral modeling of household decision making with methodologies based on emerging concepts in network science, stochastic modeling and agent based simulation. Data will be collected through scientific telephone surveys developed using qualitative methods and integrated into a temporal GIS framework including 2010 Census demographic data, spatial risk zones, and transportation route information. Communication and assimilation of evacuation warnings will be modeled using emerging complex network approaches including social contagion on random networks. The resulting impact of warning propagation on risk perception will be an input to household agent-based evacuation decision models that will incorporate social and temporal constraints. Evacuation trips thus generated (including timing, potential routes/destinations) will then be analyzed and evaluated using stochastic optimization models incorporating nontraditional objectives such as evacuation to safety and determine optimal routes, departure time, and safe destinations.
The importance of this project is that it will integrate social science and engineering to model a process in which social and technical aspects cannot each be studied in isolation. This interconnection is faced by forecasters and emergency managers as much as by researchers in multiple disciplines. As such it should open new paths in cross-disciplinary research and teaching applied to problems that are both complex and of consequence to society.