So big the forecasters missed it

Last-minute warning dumped `humble pie' on weather service


January 26, 2000|By Frank D. Roylance | Frank D. Roylance,SUN STAFF

It was supposed to snow on the fish.

The coastal storm that erupted with unexpected heavy snows from the Carolinas to New England yesterday had been forecast to slide out to sea. Mariners would have a rough day, forecasters had said on Monday, but 35 million East Coast residents would be spared.

Instead, the low-pressure system pivoted unexpectedly up the coast and intensified, exploding with energy and moisture sucked from the Gulf Stream. The combination produced the biggest Northeast snowstorm since January 1996.

"Eating a lot of humble pie this morning, but not a lot of time to digest it," said Dewey M. Walston, a forecaster at the Baltimore-Washington forecast office in Sterling, Va.

Most Maryland residents had no inkling of the coming storm until after 10 p.m. Monday, when the National Weather Service issued winter storm warnings for 4 to 8 inches for the Baltimore-Washington region.

And even that was being described yesterday as a victory for the National Weather Service's much-touted new supercomputer, and a complex new "high-resolution global model" forecasting program that went into action for the first time Monday morning.

The computer's speed and the detailed modeling "were very important in alerting our forecasters and keeping them on the ball, so that by last evening we were definitely ready to go out with a [winter storm] warning," said Louis Uccellini, director of the National Centers for Environmental Prediction.

Without them, he said, TV weather forecasters and highway officials would not have had the snow forecast until after the 11 p.m. broadcasts.

But the real message from embarrassed weather officials yesterday was that the storm simply defied all their fancy computer models, and did what the infinitely more complex machinery of atmospheric physics told it to do.

"We were definitely hoodwinked in this case," said Paul G. Knight, a meteorologist with the independent Penn State Weather Communications Group, in University Park, Pa.

Yesterday's storm began as a low-pressure center that brought gusty rains and thunderstorms to Alabama, Georgia and northern Florida over the weekend.

Three models agreed

Forecasters knew this kind of system has the potential to intensify and race up the East Coast with the possibility of big snow. They even discussed the chance of that over the weekend. But no warnings were issued.

Three different computer models the weather service uses to build its forecasts agreed: There was little likelihood the storm would charge up the coast; it was most likely to slip out to sea.

"These computer simulations are pretty good, and when they showed the same trend over two or three different computers, it was a pretty good bet," Knight said.

Pretty good, but wrong.

Uccellini said it turned out that the storm in the Southeast was being influenced by another low over the Great Lakes. That low was acting as a "pivot" that helped to turn the Southeast storm to the left, swinging its track closer to the coast.

"We know it's difficult to simulate that interaction," he said. And the computers simply missed it.

First inklings

By Monday morning, the storm's turn was beginning to show up in the models. Snow advisories were issued for coastal regions, but still not for the big East Coast cities.

By late Monday afternoon, however, things clearly began to slide downhill.

The coastal low was intensifying. Like a worsening hurricane, its central pressure dropped sharply, its reach spread from 500 to 1,000 miles.

Uccellini doesn't like to call it a "bomb." He prefers "extreme cyclogenesis." Whatever, the storm clearly had blown up in the forecasters' faces.

Worst of all, its track shifted 100 miles closer to the coast.

Storm track forecasts are often wrong by 100 miles. But usually it happens over the ocean, or cornfields or the Rockies. When it happens to millions of people crowded along the East Coast, Knight said, "it can have a devastating effect."

The weather service began releasing its model "pre-forecasts" after 8 p.m. on Monday. The official winter storm warning, calling for 4 to 8 inches for the Baltimore-Washington region, was issued at 10 p.m.

"We made every effort to make sure to get it to the media before the 11 o'clock news," Uccellini said.

Staying behind

But events continued to run ahead of the forecasters all day yesterday. The snowfall intensified and got deeper. The weather service kept upping its accumulation estimates, and pushing the western edge of the heavy snow farther westward.

"Models still trying to catch up with the rapid, deep development and may still be a bit slow and underdoing precipitation," a Sterling forecast office meteorologist warned at 3: 20 a.m.

After daybreak, Atlantic moisture continued to pour in, bringing persistent, hurricane-like bands of precipitation into the Baltimore-Washington region.

Uccellini said the intricate mysteries of "cloud physics" and snowflake formation -- another component of snow forecasting that is still beyond the capacity of the new weather models -- pushed the depth of snow produced by each inch of melted precipitation from a normal 10 inches to a striking 20 inches.

`Extraordinarily intense'

"It is an extraordinarily intense storm," Uccellini said. He ranked it among just 28 such storms since 1950 -- though not as powerful as the 1996 blizzard.

As the weather service continues to upgrade both hardware and software in its new house-sized IBM System Parallel supercomputer, Uccellini insists that its forecasts will continue to improve.

"We'll do better than we did today," he said.

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.