Good, long-range forecasting is goal


November 02, 1992|By Douglas Birch | Douglas Birch,Staff Writer

CAMP SPRINGS -- It may seem a cheeky goal for a profession that can, the day before, confidently predict sunshine for a ballgame that gets rained out.

But a small cadre of meteorologists across the United States is hunting for what may be the profession's Holy Grail, a reasonably reliable way of forecasting weather trends a month, a season or even a year in advance.

After decades of research, it's a goal that remains as elusive as the legendary Grail itself.

"Our accuracy in long-range forecasting is very poor," admits Dr. Anthony S. Barnston of the National Weather Service's Climate Analysis Center.

The 44-year-old Dr. Barnston, who is thin as an isobar, works nights at the center trying to improve the statistical computer program or "model" that the agency uses to produce its 30-day and 90-day weather forecasts.

The one-time psychology student, who found human behavior even less predictable than the weather, is not overly optimistic about the prospects for progress. The obstacles, he notes, are daunting.

But the potential payoff is enormous. Reliable long-range forecasts could save lives, improve agricultural output and help industry plan for the production of everything from parkas to electric power.

In 1955, computers using dynamic models -- programs that mimic the flow of temperatures, pressures and winds through the atmosphere -- could accurately calculate the local weather patterns for only about three days in advance, says A. James Wagner, senior forecaster with the Climate Analysis Center.

Since then, computers have become vastly more powerful. Scientists have developed improved computer models. And researchers have a deeper grasp of the forces driving the Earth's climate.

The result? Detailed local predictions can be made, at most, about six days into the future, Mr. Wagner says.

Long-range regional forecasts, meanwhile, can't rely on the same dynamic models used for short-range local forecasts -- the dynamic models can't keep up with the complicated changes that occur in the atmosphere over long periods.

So in making its 30-day and 90-day forecasts, the National Weather Service mostly relies on what it calls statistical models, which use elaborate and ingenious methods for comparing current weather maps with historical data.

The results are not exactly impressive.

The weather service's 24-hour forecasts are more than 95

percent reliable, Mr. Wagner says. But monthly forecasts are a minuscule 14 percent better than random guesswork for temperatures and 5.7 percent better for rain or snow. Ninety-day forecasts score a dismal 5 percent for temperatures and 3 percent for precipitation.

The problem is that the weather is what is known to physicists as a non-linear, or chaotic, system. A change of a few degrees in the air temperature over a small area can eventually lead to big changes in regional weather patterns.

That doesn't necessarily mean that forecasters will never be able to predict trends in climate over large regions, researchers say. But it seems to mean that forecasters will never be able to say in November that temperatures will reach 98 degrees on July 4, with thunderstorms in the afternoon.

"The fact that the climate is a chaotic system doesn't mean that you can't predict it," says Eugene M. Rasmusson, a meteorologist with the University of Maryland. "It means your ability to predict it is limited."

One ray of hope for long-range forecasters is the recognition in the past decade of the way El Nino drives global climate.

Every three to seven years, a pool of warm surface water in the western Pacific near the equator spreads eastward to the coast of South America. Meteorologists think this warm pool may be the planet's largest "engine" for pumping heat from the oceans into the atmosphere.

This warming, called El Nino or "The Christ Child" because it comes near Christmas, has been linked to droughts in Australia and sub-Saharan Africa, warmer-than-normal temperatures in central Siberia, heavy rains in northwest Europe and cold snaps in Greenland.

In the United States, it generally signals cooler, wetter winters in the southeast, and warmer, drier-than-normal weather across the northern states.

Most long-range forecasters track changes in the sea surface temperatures of the equatorial Pacific, figuring that the trend there will reverberate throughout the globe -- particularly in an El Nino year.

The most ambitious effort to achieve climate clairvoyance is probably one led by Ants Leetmaa at the National Weather Service.

Dr. Leetmaa's team of nine scientists has fashioned a model similar to those used for three- and five-day forecasts. But instead of just looking at North America a few days ahead, the BTC model is designed to predict the weather over a wide swath of the globe on both sides of the Equator, including the southern United States, nine months in advance.

Dr. Leetmaa publicly says his team has had "a modicum of success." Privately, team members have told colleagues their tests show exciting results.

Exciting or not, all the forecasters appear to have a long way to go.

Three major groups studying the El Nino cycle have come up with three different predictions of Pacific sea surface temperatures -- one shows it cooling, one warming and one predicts it will stay near normal.

Timothy P. Barnett and Nicholas Graham, research meteorologists at the Scripps Institute of Oceanography in California, have a statistical model that predicts a cooling in the next few months.

Stephen E. Zebiak and Mark Cane at Columbia University, meanwhile, predict a flat year -- nothing out of the ordinary -- in the equatorial Pacific.

Dr. Leetmaa's group predicts what no one else expects: a warming trend, though not a full-fledged El Nino, beginning in December.

"I'm not very upset with my forecast . . . We're out on a limb," Dr. Leetmaa says.

Baltimore Sun Articles
Please note the green-lined linked article text has been applied commercially without any involvement from our newsroom editors, reporters or any other editorial staff.