The Effect of Tropical Cyclone Characteristics on U.S. Landfall Probability

When a tropical cyclone threatens the coastline, decision makers can take preparatory actions designed to mitigate the damage caused by landfall. Those in this situation must decide whether and when to begin their preparations. Regnier and Harr have developed a dynamic decision model in which the decision maker has the option of delaying preparation and waiting for an updated, more accurate forecast. Regnier and Harr combine their decision model with a Markov model of tropical cyclone motion derived from fifty-three years of Atlantic hurricane data. We examine the state space Ω used in the Markov model with the goal of including additional state variables for storm characteristics such as direction of travel and wind speed. Logistic regression analysis is used to examine characteristics that had a statistically significant effect on landfall probability for Atlantic tropical cyclones between 1950 and 2007. The revised state space developed here will improve the Markov model, enabling Regnier and Harr’s dynamic decision model to provide decision makers with more valuable information.