Sensitivity Analysis of Hurricane Evacuation Casualties and Costs in Florida

Hurricane evacuations are expensive endeavors. Some evacuation costs are necessary to protect lives, but forecast error at the time that evacuation orders are issued can result in the potential for casualties in areas not appropriately warned, as well as unnecessary evacuation costs in areas unaffected by the storm. Given the interplay between hurricane forecasting, issuance/cancellation of evacuation orders, and overall evacuation costs, it is logical to test the sensitivity of those costs to improvements in hurricane forecasting. To examine this interplay, evacuation costs are calculated for over four hundred thousand simulated hurricane evacuations of the Florida coastline, and an analysis is performed on the sensitivity of those evacuation costs to three major variables: accuracy of the synthetic hurricane forecasts, length of coastline evacuated beyond a simulated forecast cone of uncertainty, and lead time for issuance of evacuation orders. A reduction in forecast track error reduces the average total overall cost of evacuation but may result in higher casualties if evacuations do not cover an adequate stretch of coastline. Evacuating approximately 250 km on each side of the cone of uncertainty is optimal because average casualty costs are not appreciably decreased with a larger evacuation area. A longer evacuation lead time results in a decrease in average casualty costs as a greater percentage of affected households evacuate, while the average combined cost of evacuation and casualties increases by at most a few percent.