Dynamic Gravity Model for Hurricane Evacuation Planning

The paper presents a dynamic gravity model to estimate time-dependent origin–destination (O-D) trip tables for hurricane evacuation with survey data from Hurricane Floyd in South Carolina in 1999. The objective is to test whether a dynamic gravity model with a revised impedance function can be used successfully to model hurricane evacuation destination choice. A static gravity model is used to estimate dynamic O-D matrices on the basis of which the dynamic gravity model is subsequently calibrated, because the dynamic O-D movement observed from the survey data is too sparse to allow estimation of the dynamic gravity model. The static gravity model was developed with a combined impedance function and describes evacuees' travel behavior. The model was estimated through a chi-square minimization process. The model was found to produce a trip length distribution that was statistically significantly similar to observed values. A time-dependent travel demand was estimated on the basis of a sequential logit model. With the trip distribution from the static gravity model and an expansion factor, time-dependent O-D trip tables were computed. Dynamic traffic assignment was then performed to provide time-dependent O-D travel costs. The static gravity model was extended into a time-dependent version in which the time-dependent travel cost, distance from the projected path of the hurricane to the destination, and remaining accommodation at the destination feature in the formulation. The model is solved by ordinary least-squares regression after transformation. The time-dependent gravity model was found to perform well by comparison of predicted and observed trip length distributions.