Coming storms: method predicts intensity of U.S. hurricane seasons.

A new computer model that analyzes summer-wind patterns can predict whether the United States will suffer a damaging hurricane season, according to the scientists who developed the tool.

Hurricanes are among nature's most destructive disasters. Eight of the 10 costliest U.S. calamities were hurricanes, says Mark A. Saunders, an atmospheric physicist at University College London in Dorking, England. The average annual bill for hurricane damage in the continental United States is about $5.6 billion. However, last year, when six hurricanes slammed the Gulf Coast and eastern seaboard, insured losses tallied $18.8 billion in Florida alone.

Means to foretell whether an upcoming storm season will be particularly dangerous could benefit people who plan responses to large-scale emergencies or who work in the insurance and electrical-utility industries.

The new computer model, developed by Saunders and his colleague Adam S. Lea, considers the speed and direction of winds at altitudes between 750 meters and 7,500 in above sea level during the month of July over six broad regions of the eastern Pacific Ocean, the North Atlantic Ocean, and the central United States. The patterns of those winds, which influence the paths of weather systems and steer Atlantic storms either toward land or out to sea, typically persist throughout the hurricane season, says Saunders.

Although the North Atlantic hurricane season officially runs from June 1 to Nov. 30 each year, 86 percent of the hurricane strikes--and 96 percent of those from severely damaging hurricanes, category 3 or larger--have occurred after Aug. 1.

With their model, Saunders and Lea analyzed wind data gathered between 1950 and 2003. In nearly three-quarters of those years, their simulation accurately indicated whether U.S. hurricane damage was above or below average. That's a significant improvement over current models, says Saunders. The researchers describe their analyses in the April 21 Nature.

Although the new technique may be useful for assessing nationwide risk from hurricanes, it doesn't help meteorologists estimate the relative risks to different sections of the U.S. coastline, says James B. Elsner of Florida State University in Tallahassee.

Predicting the landfall of a hurricane, and therefore the location of its damage, is a tricky business, says Philip J. Klotzbaeh, an atmospheric scientist at Colorado State University in Fort Collins. The methodology Saunders and Lea used to develop their model for determining overall damage is "solid," he says. "I like that [the model] is linked to atmospheric physics" more directly than are some other models used to predict hurricane activity, he adds.