Predicting Blizzards: Model Misbehavior

Amazing weather photos for every month!
The Old Farmer's Almanac
2013 Weather Watcher's Calendar

$9.99

Related Articles

PrintPrintEmailEmail
Your rating: None Average: 4.5 of 5 (15 votes)

 

Think blizzard and heavy snowfall comes to mind. Officially, however, the deadly winter storm is defined by these factors:

  • wind gusts over 35 mph
  • visibility of less than a quarter-mile (though if you've ever been caught in a blizzard, you'll probably swear it's closer to a few inches)
  • duration of at least 3 hours
  • temperature below 20°F (-7°C)

Although blizzards are tracked by satellites, forecasters use computer models to predict their paths. The models reside in mammoth supercomputers and are constantly fed information about the current state of the atmosphere.

According to the National Weather Service, the computers have an overall accuracy level of 85 percent—a respectable rate, as some have learned the hard way.

On February 5, 1978, weathermen ignored a computer's prediction of a massive East Coast snowstorm and put their faith in more tried-and-true methods, such as looking through reams of hand-drawn charts. The computers won; areas of the Northeast wound up buried under more than 2 feet of snow.

Nonetheless, while computers can often "see" blizzards coming, they're far from foolproof.

On January 24, 2000, models persuaded forecasters in Washington, D.C., to predict "less than an inch of snow" for the following morning. The models misfired by about 250 miles (400 kilometers), and a major blizzard dumped over a foot of snow. The foul-up was blamed in part on bad timing; weather balloons delivered new information just after the evening weather shows were over.

Another problem: Models are only as good as the data they're given-and that's never an accurate reflection of real conditions. A tiny change in the model can deliver a different answer.

In the case of the D.C. blizzard, the model didn't adequately gauge the effect of an eruption of thunderstorms over Georgia. This storm was a missing variable that pushed the blizzard on a more western track.

The computer model's degree of accuracy also depends on the blizzard itself. Frontal systems have less predictability than high pressure systems. Small miscalculations in location can mean big differences in what you're going to shovel the next day.

For the Record

Forecast: Fair, slightly colder

Fact: "Colder brisk westerly winds and fair weather" was the official forecast issued by the War Department Signal Service (a precursor to today's National Weather Service) on March 11, 1888. The next day, the Blizzard of 1888 dumped four feet of snow from New York City, north through New England, killing 400.

The Vermont Bellows Falls Times described conditions this way: "No paths, no streets, no sidewalks, no light, no roads, no guests, no calls, no teams, no hacks, no trains, no moon, no meat, no milk, no paper, no mails, no news, no thing—but snow."

More Articles:

Comments

One of THE best predictors of

One of THE best predictors of TWO major snowstorms in my area (DC) was -- The Birds.... As the snow began late in the afternoon, a light dusting of tiiiny tiny flakes started to fall.. and at the same time, the birds started to FRANTICALLY search for food, even while the ground was still dry and uncovered. "Better go home NOW" I said to my guests, who had been waffling about "blizzrd? or not?" -- and they left. By daybreak we had almost 2 feet of snow on the ground.

im so sick and tired of the

im so sick and tired of the last several years of big predictions-little results i love snow yet years have gone by with "warmer than usual" winters why? when i was a girl we could expect a white Christmas and a good blizzard nearly every year-im sad my children & grandchildren dont have the deep snow to play in!

Post new comment

The content of this field is kept private and will not be shown publicly.
By submitting this form, you accept the Mollom privacy policy.