Better forecasts, the old-fashioned way

For 17 years, Bill Gray has been unloading preseason hurricane forecasts with the precision of a Pedro Martinez fastball.

Many of Dr. Gray's colleagues, like hitters fanned by the famous Red Sox pitcher, shake their heads, and mutter: How does he do it?

The lanky, silver-maned Gray has become the first forecaster the media, the public, and several insurance companies look to for an early estimate of the number and strength of storms a hurricane season is likely to bring.

After years of grumbling, even the nation's "official" source for forecasts - the National Oceanic and Atmospheric Administration (NOAA) - has begun to recognize his work, by imitation. Last year, it began to issue seasonal hurricane forecasts of its own, which carry an explicit nod to Gray's efforts.

"When you come to within one or two storms a season in each category, you're doing a darn good job," says Tiruvalam Krishnamurti, a meteorology professor at Florida State University in Tallahassee. "What we don't understand is, why does it work?"

Ask Gray why it works, and his answer is absolute: "Our whole philosophy here is to study the past," he says.

Unlike a large number of weather researchers, Gray is an outspoken critic of what he sees as an overreliance on complex computer models. For years, he shimmied through storms in propeller planes to gather hard data and study weather patterns. Now, with governments sharing weather records, he takes his cues from history, not computers' number crunching of complex meteorological readings.

"You have all these years of data," he says. "You don't need to understand all the physical links between the processes if you know there's an association" between them.

Fruits of his labor

It's a process that has led him and his colleagues here at Colorado State University and at NOAA to predict 11 named storms in 2000, including seven hurricanes - three of them major. Hurricane Debby, which could affect the US later this week, is the fourth named storm and the second hurricane of the season, which runs from June 1 until Nov. 30.

It's also leading him to new frontiers of storm forecasting. Looking at past climate patterns, Gray says conditions are building that may lead to more Atlantic and Caribbean hurricanes reaching land over the next decade. As a result, he and his team are developing an approach for estimating the seasonal probability of hurricanes making landfall along the East and gulf coasts.

Peering at climate trends was not Gray's first choice of careers, though. His first love, he says, was baseball.

Two years in the college league in Kansas during the summers of 1948 and '49 landed him an invitation to try out for the Washington Senators farm system.

"I wasn't that good a hitter, so I probably couldn't have made it as an outfielder, but as a pitcher I always had a strong arm," he recalls. Asked about his ERA, he chuckles, "We didn't keep track."

In 1951, a knee injury ended his hopes for a major-league career, but led him down a new path.

It was October 1957 - a year after Gray finished a stint as a weather officer in the US Air Force - that the Russians launched Sputnik. Washington called for boosting federal support for science, and Gray headed to graduate school to study meteorology.

That decade saw some of the strongest hurricanes in years make landfall along US coasts. Connecting with a mentor who was keenly interested in tropical cyclones, Gray became captivated by them as well. In 1958, he strapped himself into the bombardier's seat in the nose of a modified B-29 Stratofortress and made his first research flight. The storm was hurricane Helene.

"That plane was really stressed," Gray recalls. "We were flying at 1,500 feet, bucking winds up to [138 m.p.h.]."

Gray's penchant for hard data on hurricanes placed him in the front ranks of hurricane research, colleagues say. Piecing together information from quick flights through pieces of storms, he is credited with early discoveries about the properties found in hurricane cores. In the late 1960s and into the 1970s, when international weather data was not widely shared, Gray traveled the globe gathering information that enabled him to put together the first worldwide study of where tropical cyclones tend to form and the conditions that spawn them.

"He came up with the picture we still use today," says William Frank, a meteorology professor at Penn State University in University Park.

No substitute for study

His disdain for the "glamour" of computers appears to extend to his own office PC, which displays more stains than a playroom rug. Computers are worthwhile, he says, but no substitute for study. "Numerical weather prediction has been a great success story for five to eight days out," he says. "But you can't take that and integrate it out for three or 10 or 100 years into the future."

Little wonder, then, that Gray also stands among the ranks of skeptics on human-induced global warming. "Is the climate changing? Sure it is," he says. "But it always changes."

His skepticism about the accuracy of forecasts from numerical models has rubbed off on some of his students, who have moved into climate and weather research careers of their own.

In a study in the current issue of the Bulletin of the American Meteorological Society, Gray protg Christopher Landsea and NOAA colleague John Knaff analyzed El Nio forecasts. They found that simpler statistical techniques - the approach Gray favors - outperformed their more-complex and much-heralded numerical siblings.

"With the results of this study," the two conclude, "one could have even less confidence in anthropogenic global-warming studies because of the lack of skill in predicting El Nio."

(c) Copyright 2000. The Christian Science Publishing Society