Why Hurricane Models Can't Be Fully Trusted

By Grace Muller, AccuWeather.com Staff Writer
September 07, 2012; 5:23 AM
Share |

Meteorologists forecast a hurricane's track and intensity by looking at, among other things, previous hurricanes' tracks, radar, satellite and forecast models.

A forecast model uses data collected globally by satellites and plugs it into an equation that predicts how the hurricane will move. The equations use atmospheric pressures, the temperature of the air and ocean, the amount of moisture in the air and other factors to determine what direction a storm will go.

Since universities developed the first forecast models in the 1950s, the equations have evolved. When meteorologists first started researching hurricanes, simplistic computer models took rotating cylinders of air and tried to move them. The equations did not account for the sophisticated physics that affect an actual hurricane. The forecast models have continued to evolve since then, with better computers and more data as the main factors in creating more accurate forecasts. Faster computers can process more data that comes from research flights into storms.

"Ten, 15 years ago, computer models were much more difficult to deal with because they waffled all over the place," AccuWeather Expert Meteorologist Dan Kottlowski said. "You'd get a run and it would take the storm in one direction. Then you'd get another run and it would take it in another direction, more wildly than what we see nowadays."

Forcasters have a hard time forecasting storms when they are more than three or four days out from landfall. The room for error increases the farther the storm is away. According to Kottolowski, five days out from landfall, a storm can deviate 300 to 400 miles from the projected point of landfall, which is why information that far ahead should be used with caution. However, the error in forecasting the storm, say within 24 hours, is less than 90 miles.

"They're just models," AccuWeather Expert Meteorologist Henry Margusity said. "They're wrong. You have to look at other things to forecast the weather. You have to look at the weather pattern and understand what the weather pattern is there. Models are just tools to use, they're not always right. You also have to look at water temperatures and the state of the atmosphere. There are a whole bunch of things meteorologists look at."

Words to Know When You're Talking Models:

Consensus model: Rather than expect one forecast model to estimate a storm's track accurately, meteorologists combine a few trusted models into a "consensus model," which does a better job of forecasting a track.

"It's been proven time in, time out, that a consensus of models beats the individual model," Kottolowski said.

Verified forecast: An accurate estimate by a meteorologist that pinpointed where that storm is and the location and the intensity.

A run: When data is entered into a model's equation. Computer models are "run" at specific intervals, between every six and every 24 hours. It takes between three and six hours to get the solution to the atmospheric equation so by the time you get the model output the data is already around three to six hours old.


Comments left here should adhere to the AccuWeather.com Community Guidelines. Profanity, personal attacks, and spam will not be tolerated.

More Weather News

  • Weekend Rain to Soak Billings, Bismarck, Regina

    August 23, 2014; 3:13 AM ET

    A swath of soaking rain will slowly shift from the northern Plains to the Canadian Prairies this weekend, making people reach for their umbrellas and heightening concerns for flooding.

  • Pittsburgh: Storms to Retreat By Sunday

    August 23, 2014; 3:12 AM ET

    While residents will face more disruptions to outdoor activities on Saturday, dry air will push southward across Pittsburgh to end the weekend.

Daily U.S. Extremes

past 24 hours

  Extreme Location
High N/A
Low N/A
Precip N/A


This Day In Weather History

Dallas/Ft. Worth, TX (1980)
105 degrees -- the 60th consecutive day with a high temperature of at least 100 degrees.

Lake Ontario (1987)
Strong northwesterly winds caused upwelling. On the south shore, the water temperature was 70 degrees, but along the north shore, it was only 41 degrees.

Virginia/WV (1990)
Heavy rains caused floods. Kilmarnock, VA, had a two-day total of 13.50 inches, and Nassawaddox, VA, had 12 inches.