Weathering vagaries of storm tracking

January 31, 2000|By Frank D. Roylance

THE TRUTH ABOUT the "surprise" snowstorm on Jan. 25 was that forecasters were not really surprised at all.

They knew a big East Coast snowstorm was a possibility, and not a remote one. But, stung by prior bad calls, they held their tongues rather than send a shiver up the East Coast that might prove to be another false alarm.

The truth is "we get gun shy," said meteorologist Paul Knight, of the Penn State Weather Communications Group. When forecasters get big storms wrong, they take heat from the public, and it can color how they handle their next forecasting dilemma.

Tuesday's meteorological debacle may have been rooted more in forecaster psychology, and in the public's unrealistic expectations of weather science, than in any failure of the National Weather Service's 2-month-old IBM supercomputer or its state-of-the-art weather models.

Meteorologists knew by the weekend that a coastal snowstorm was a possibility for Tuesday or Wednesday. The low was in place in the Deep South, and "we were watching this thing like a hawk," said Louis Uccellini, director of the National Centers for Environmental Prediction.

Sure, the computer models all said the storm was likely to move harmlessly out to sea. But the mathematical "confidence" level of that seaward storm track was low.

In other words, the thing was almost as likely to turn and charge up the coast. But we were never told.

The forecasters sided with the computers. They predicted cloudy skies for Baltimore and Washington, with maybe a chance of snow along Atlantic coast, closer to the storm's predicted offshore track.

The decision not to alert 35 million East Coast residents to the off-chance of a paralyzing storm was likely colored by two earlier forecasting failures, Mr. Knight said.

During the week that preceded the big storm, forecasters had predicted a significant Northeast snowstorm on Sunday, Jan. 23. But the strong low centered over the Great Lakes failed to move south as predicted. The snow never arrived.

The storm center moving over Alabama and Georgia, meanwhile, looked even weaker, and seemed more likely to go to sea.

"Since the Sunday thing fizzled, they said this thing was gonna fizzle also," Mr. Knight said.

But looming even larger in forecasters' minds, he said, was the still-fresh memory of the forecasts last September that led more than a million people to evacuate the coasts of Florida, Georgia and South Carolina to escape the 140-mph fury of an approaching Hurricane Floyd.

As it happened, Floyd stayed off the coast, skidding northward before crashing ashore in North Carolina. With hindsight, the evacuation, with its traffic gridlock, jammed motels and ruined holidays, seemed a costly error.

"Meteorologists were called on the carpet about that," Mr. Knight said. "`How dare you force a million and a half people to evacuate'" when the storm was headed somewhere else. "We got burned, and there is a certain psychology that makes one hesitant to cry wolf a second time."

In the weekend before the Jan. 25 snowstorm, forecasters could have alerted broadcasters and the emergency management people all up the East Coast to what they were discussing among themselves -- that this relatively weak Southern storm could turn on them.

All the elements were there. Arctic cold over the Northeast, and the Gulf Stream ready to supercharge the storm and send it spinning up the coast.

"There were signs that perhaps the system would hug the coast," Mr. Uccellini said. "But the dominant signal we got from the models was that it would be moving out to sea."

Mr. Knight said forecasters asked themselves a question: With the computers leaning toward flurries, did the potential impact of five or 10 or 20 inches of snow on East Coast cities outweigh the risks of touching off a snow panic among 35 million people, only to be proven wrong.

With the sting of the Floyd evacuations in mind, they clung to their computers. And then the computers missed the storm's actual track by 100 miles. That's a common margin of error, Knight said, and it usually has little consequence. But with snow, along the populous East Coast, 100 miles affects millions of lives.

In the end, the computers seem to have done well enough. The real problem appears to lie in our own unrealistic expectations, and perhaps with forecasters who allow our bluster to color their predictions.

They should share their thinking as well as their model runs, without risking their credibility. And if the storm doesn't come, they should not have to suffer our scorn. For if they keep us in the dark about the next uncertain threat, they may wager our lives and property.

Frank D. Roylance is a science writer for The Sun.

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.