Tropical cyclone intensity estimation in the North Atlantic basin using an improved deviation angle variance technique

This paper describes results from an improvement to the objective deviation angle variance technique to estimate the intensity of tropical cyclones from satellite infrared imagery in the North Atlantic basin. The technique quantifies the level of organization of the infrared cloud signature of a tropical cyclone as an indirect measurement of its maximum wind speed. The major change described here is to use the National Hurricane Center's best-track database to constrain the technique. Results are shown for the 2004-10 North Atlantic hurricane seasons and include an overall root-mean-square intensity error of 12.9 kt (6.6 m s ^-1, where 1 kt50.514ms ^-1) and annual root-mean-square intensity errors ranging from 10.3 to 14.1 kt. Adirect comparison between the previous version and the one reported here shows root-mean-square intensity error improvements in all years with a best improvement in 2009 from 17.9 to 10.6 kt and an overall improvement from 14.8 to 12.9 kt. In addition, samples from the 7-yr period are binned based on level of intensity and on the strength of environmental vertical wind shear as extracted from Statistical Hurricane Intensity Prediction Scheme (SHIPS) data. Preliminary results suggest that the deviation angle variance technique performs best at the weakest intensity categories of tropical storm through hurricane category 3, representing 90%of the samples, and then degrades in performance for hurricane categories 4 and 5. For environmental vertical wind shear, there is far less spread in the results with the technique performing better with increasing vertical wind shear.