Chicagoans may quibble for years about how the Blizzard of 1999 stacked up against great snowfalls of years past, but one distinction seems indisputable: This year’s onslaught was among the best-predicted winter storms in the city’s history.
As early as a week ago Sunday, when the ruinous storm was nothing but a gleam in the eye of some wispy clouds far out over the Pacific Ocean, local forecasters saw the signs gathering for a big snow to hit the Chicago area by Friday. Those predictions never wavered.
Meteorologists chalk up their success in part to what many are calling “a textbook snowstorm” that would have been hard to miss even in the worst of circumstances.
But the ability to spot such dangers nearly a week in advance–a luxury inconceivable for victims of blizzards that hit in 1967 and 1979–is the result of sophisticated data-gathering techniques and superfast forecasting computers that have become available only in the last 10 years.
The power of such tools is getting more formidable. Weather-computing capacity will grow thirtyfold over the next few years as the government brings on-line two IBM RS/6000 supercomputers. That’s the same family of computers that produced Deep Blue, the computer that in 1997 beat Garry Kasparov, the world’s reigning chess champion.
The most sophisticated of the new models should be available in late 2000, with mind-boggling processing speeds around 2.5 trillion calculations per second. By comparison, the fastest computers used today by the National Weather Service perform nearly 16 billion calculations per second.
“Thirty years ago, human forecasters mostly used their intuition and experience,” said Buzz Burek, a meteorologist with the National Centers for Environmental Prediction in Camp Springs, Md. “Now it’s about deciding which computer model is right.”
More than eight computer models from different countries and national agencies tracked the elements that made the colossal storm, crunching the raw data using supercomputers. The primary model in the U.S., called the Medium Range Forecast, is currently fueled by three Cray supercomputers in Suitland, Md..
Starting last weekend, all eight models forecast a major Midwest storm for Friday.
The computers make their calculations based on measurements from around the world on such factors as wind speed, temperature and water content of the air.
Forecasts are becoming more accurate in part because meteorologists have more ways of collecting the data used to make weather predictions, said Jim Allsopp, a meteorologist at the National Weather Service office in Romeoville. The weather service still uses tried-and-true methods, such as weather satellites and weather balloons that are launched twice a day at 80 sites across the United States.
But new methods have emerged since the late 1980s. Among them are “profilers,” small machines that send microwave pulses into the atmosphere to measure wind speeds. According to Allsopp, they’re similar to radar guns used by police.
Another fairly new forecasting tool is a network of automated weather stations that collect a wide range of data. Started in the early 1990s, those stations are gradually replacing the human observers that the weather service had relied upon for decades.
In the Chicago area, automated stations are located at all the major and smaller airports, and they relay their information 24 hours a day to the weather service’s computer network.
The drama of the Blizzard of 1999 began last weekend, when two weather fronts started forming thousands of miles apart–one comprising subtropical air far west of California and a cold northern front in the Gulf of Alaska.
As the northern front moved southeast, the Pacific storm front forged due east, passing over California on Thursday. It then took a southerly turn over the southern Rocky Mountains and down into Texas, where it arrived Friday.
As it moved toward the Gulf of Mexico, the southern front picked up huge amounts of warm moisture–a prerequisite for a large snowfall. The water-laden clouds followed a low-pressure corridor north through Missouri and southern Illinois on Friday, eventually meeting up with the cold system coming from the northwest.
Because warm air carries more water vapor than cold air, the southern front began to dump its moisture as it ran into the northern branch, creating snow in northern Illinois and rain farther south.
More Top Picks Tmy Mini
“The two fronts got together and combined their energy,” Burek said. “It’s either very good or very bad timing on the part of Mother Nature, depending on whether you like snow.”
As the two systems clashed, cloud banks started swirling counter-clockwise, leaving snow in its wake–right in the middle of the Chicago area. Instead of the usual winds blowing west to east, downtown Chicago was slapped with winds coming straight from the northeast, off Lake Michigan.
The lack of buildings and trees slowing the wind off the lake made for shoreline gusts of up to 60 miles per hour, easily surpassing the standard of 35 m.p.h. sustained winds that makes up part of the definition of a blizzard.
Meteorologists foresaw all of this days in advance. The current national forecasting supercomputers deliver regional predictions out to five days and general forecasts for 10 days in advance.
“The improvement in forecasting was flat between the 1930s and 1960s, then these models made it creep forward,” said Tom Skilling, meteorologist for WGN-Ch. 9. “They say the forecast to Day 5 today is as accurate as the Day 2 forecast was in the 1960s.”
Skilling conceded that the newfangled systems aren’t perfect. The computers still give false alarms of major storms at least once a month, according to most experts.
That leaves the decision of which models to believe up to experienced meteorologists–who readily acknowledge their fallibility.
“We want to give as much lead time as possible,” Allsopp said, “but we don’t want to get too cocky.”