Model for Severity of Hurricanes in Gulf of Mexico

Abstract

Observational and theoretical evidence suggests hurricane severity can be affected by local oceanographic features. The most important of these features in the Gulf of Mexico are the Loop current and its associated eddies. These features are large bodies of warm subsurface water that, theory suggests, can act as an energy source for passing hurricanes. Present design standards for the offshore oil industry assume no spatial variability, so these findings have substantial financial implications for design of new structures and requalification of thousands of existing ones. We propose a statistical model and an estimation procedure that accounts for the spatial variation of the probability distribution function of hurricane severity. The model is based on the representation of the hurricane process as a marked line process where the hurricane tracks are idealized as linear segments with respect to a given location, and where the marks measure the severity of hurricanes. The efficiency of the estimation procedure is evaluated through simulations for different types of spatially dependent probability distribution functions. The procedure is finally applied to a data set comprising the 100 most severe hurricanes in the Gulf of Mexico from 1900 to 1989.