Weatherman Jay Trobec has been giving the forecast to 90,000 viewers of his Sioux Falls, S.D., TV station for 14 years, and he is usually right.

But "if I blow a forecast, I hear about it," he says.

When he predicted six inches of snow in Sioux Falls that never arrived, "people were coming up to me in the coffee shop and berating me," says Dr. Trobec, chief meteorologist at KELO-TV. "People who lay concrete for a living, people who put roofs on houses don't like it when the forecast isn't correct."

This year has already seen 10 weather disasters each costing more than $1 billion in damage, making it the most costly since the government started keeping records in 1980. And it has been one of the toughest years in memory for meteorologists. The technology used for forecasting has improved, and forecasts are more accurate compared with the past. But the job of the meteorologist is still both an art and a science.

‘STICKING THEIR NECKS OUT'

A lot of decisions hinge on forecasters' words. When Scott Nogueira, a meteorologist who specializes in aviation, told airline managers from Washington, D.C., to Philadelphia that a massive blizzard was expected to dump up to
27 inches of snow on airports in the region, "they gasped," he says.

In turn, the airlines began planning to cancel hundreds of flights, inconveniencing thousands of people. And when those forecasts don't pan out, tempers flare.

"Every day, meteorologists are sticking their necks out, sometimes into the guillotine," says Peter Neilley, vice president of global forecasting services for Weather Channel in Atlanta. "With a volatile weather year like we've had in 2011, it leads to a higher level of stress overall."

Working alone or in teams, some meteorologists specialize in certain industries, others in specific locales. All of them base their forecasts on computer models; radar and satellite images; and wind, temperature and precipitation data
collected from hundreds of sources on land and sea and in the atmosphere.

Improved data gathering, computer modeling and scientific analysis have made seven-day forecasts of broad weather patterns as accurate as five-day forecasts were 20 to 25 years ago, and five-day forecasts as accurate as three-day forecasts were then, says David B. Parsons, director of the school of meteorology at the University of Oklahoma in Norman.

Forecasts of precipitation and temperature also have become slightly more accurate in recent years, says Eric Floehr, founder and owner of ForecastWatch, which tracks the accuracy of weather forecasts. A one-day forecast comes within three degrees of hitting the mark, on average; a three-day forecast is usually accurate within four degrees.

There is still plenty of room for error. Storm fronts often move more slowly or quickly than expected, causing forecasters to "do a very poor job of predicting specifically where snow is going to land, and how much," Mr. Floehr says. Also, meteorologists often fail to predict the exact intensity of hurricanes because of a lack of data; the storms tend to destroy ocean buoys that measure wind speed and prevent piloted aircraft from getting close enough to analyze them.

WHAT DO WE DO?

Adding to the pressure: People increasingly look to meteorologists not only to predict the weather, but to tell them what to do about it. Nearly nine out of 10 Americans look at or listen to daily weather forecasts, usually more than three times a day, according to a 2009 survey.

When Hurricane Irene roared up the Atlantic seaboard in August, dozens of anxious people asked meteorologist Elliot Abrams where to park their cars or how much food to buy. Mr. Abrams, a senior vice president with AccuWeather, broad-cast hurricane updates on 15 radio stations and responded to callers who wanted to know whether to board up their windows.

"You don't want to make it sound so bad that people take risks and hurt themselves," Mr. Abrams says. "On the other hand, they need to take the right action."

Likewise, James Franklin was staying up most of the night helping figure out what forecasts the government should deliver. As a branch chief at the government's National Hurricane Center in Miami, Mr. Franklin barely slept as Irene churned toward the East Coast. He and the 10 meteorologists on his team agonized over the wording of the center's four daily advisories. Forecasters on each of three shifts held conference calls with other meteorologists and officials to discuss exactly where to issue hurricane watches vs. warnings. When Irene's winds seemed to weaken, Mr. Franklin urged caution over making sudden changes in their forecasts; an abrupt shift might force a reversal later.

"So many people are making decisions based on what we say, that I'm always worried," he says.