Dynamic route choice in hurricane evacuation

In this research a framework is developed for modeling route choice in hurricane evacuation. Two behavioral hypotheses are evaluated which together with the route choice model, constitute the contributions of the research. The first hypothesis states that beside congestion, other variables such as familiarity with the route, availability of fuel and shelter, facility class, and length of route have an effect on an evacuees\u27 route choice. The second hypothesis states that as time passes and storm conditions change, the impact each variable has on route choice changes. The logit structure was used for modeling the choice process and stated choice data previously collected from the New Orleans area on hypothetical storms was used to calibrate the model. The study found that accessibility of a route, familiarity with a route, facility class, length of a route, and availability of services (gas stations and hotels) had an effect on evacuation route choice. The magnitude of the coefficients of perceived service, accessibility, and distance differed among those evacuating in the first half of the evacuation period versus those that evacuated in the second half but coefficients of facility class were not significantly different between two time intervals. Observed traffic count data from hurricane Katrina evacuation was used to validate the model. Comparison of traffic volumes predicted by the model with actual traffic volumes from hurricane Katrina shows error percentages of 17.5, 0.01, and 28 percent of error for volumes on I-10, I-55, and US-61 respectively