Evaluating a revised 24-hour Atlantic statistical hurricane intensification model with operational numerical models

A new statistical hurricane intensification model was developed to forecast the changes in the 24-hour maximum sustained winds. This model differs from the previous version in that it can be used throughout the entire track of a tropical cyclone rather than the time period near and during the rapid intensification period (RIP). A sample of all Atlantic tropical systems from 1981-2006 was developed using key intensification variables for discriminant function analyses (DFA) and multiple regression analyses.

A two-step discriminant function analysis (DFA) was used to help predict the strongest the tropical system will eventually become and the relative magnitude of the 24-hour maximum sustained wind speed change. The first step of the DFA was used to select among three initial categories: tropical storm, minor hurricane, and major hurricane. The new version incorporated the additional category of tropical storm. Rather than using the proximity to the RIP, as in the previous version, the new version used the second step of the DFA to select among five categories relating to the magnitude of intensification: rapid intensification, slight intensification, little to no change, slight weakening, and rapid weakening.

Multiple regression models were created for the 15 combinations to forecast the changes in the 24-hour maximum sustained winds. The model output was tested independently on the 2007 season and the results were then compared with other operational dynamic and statistical intensity models.