An anemometer measures winds near a tornado in Goshen County, Wyo., in 2009. (Ryan McGinnis)

While violent, rotating thunderstorms spawn almost all of the most powerful tornadoes, most of them produce only weak twisters — or none at all. Unfortunately, meteorologists can’t predict which of these swirling storms, known as supercells, will spin out damaging and deadly tornadoes, such as those that devastated Birmingham, Ala., and Joplin, Mo., last year.

To understand what causes one storm to be devastating and another to fizzle — and to enable forecasters to give earlier warnings when a deadly twister is on the way — a federally funded program known as VORTEX2 was launched several years ago. About 1,300 tornadoes hit the United States in an average year; about a quarter of them are classified as particularly deadly, with winds of at least 118 mph.

In 2009 and 2010, 120 scientists and students crisscrossed the Great Plains chasing supercells during the May-June tornado season. Now VORTEX2 is in a data-analyzing phase that meteorologists hope will help them better understand the wind funnels.

Forecasters now issue tornado warnings when someone, usually a trained storm spotter, reports a tornado to the National Weather Service or when an NWS Doppler radar indicates one has formed or is on the verge of forming. Josh Wurman, one of the eight scientists on the VORTEX2 steering committee, says that in addition to increasing the warning time from today’s average of 13 minutes before a tornado hits a particular community to 20 or 30 minutes, VORTEX2 might help push the false alarm rate from the current 70 percent down to 50 percent or lower.

The major reason for the high false alarm rate is that the radar indications are ambiguous, says Russ Schneider, director of the National Weather Service’s Storm Prediction Center in Norman, Okla. Forecasters are torn between waiting for the radar to give a strong indication that a tornado will occur, which could make the warning too late to save lives, or issuing an earlier warning that could be a false alarm.

The purpose of VORTEX2 can be summed up as looking for ways to resolve the ambiguity — to find previously unknown indications that a tornado will soon form.

Wurman says approximately $10 million of the $13 million that the National Science Foundation and the National Oceanic and Atmospheric Administration are spending on VORTEX2 is going into data analysis, which could take until 2017 to finish.

He estimates that in 2009 and 2010 researchers collected 30 terabytes of data about the tornadoes. This compares with the 18 terabytes of data in the entire 1940 U.S. census that the National Archives put online on April 2.

One highlight of VORTEX2’s field phase was the tornado that cut across Goshen County, Wyo., on June 5, 2009. The Weather Channel was broadcasting live, and crews were shooting a “Storm Chasers” show there for the Discovery Channel and for an Imax movie (“Tornado Alley”) as scientists collected the most data ever recorded on the birth, life and death of a tornado.

Karen Kosiba, a scientist helping to organize the data collection, was in one of 11 trucks capturing close-up views inside the tornado and its parent thunderstorm. “I was so busy, I think I opened the door only once to look at it,” she recalled. Later, when she saw the Weather Channel video from Goshen, she thought: “Wow, that was the tornado we got data in. It is very picturesque.”

The tornado didn’t cooperate with the data collectors. At the last minute, it changed direction and missed all the mobile weather stations that had carefully been placed where the wind funnel might hit, a good example of the frustrations of measuring tornadoes. But the thunderstorm and its tornado were within range of the radars and other instruments for a half-hour. The scientists collected more useful data than from any twister in history.

VORTEX2 scientists expect to find examples of tornado precursors, or triggers, in the data they collected in Goshen and elsewhere. Such precursors have thus far proven elusive.

“The triggers for tornado genesis must be subtle, or we would have noticed them already,” Wurman says. But “I never expected to collect some data one afternoon, then go back to the lab that night and say, ‘Eureka! So that’s how tornadoes form!’ ”

VORTEX2 is the second major recent effort to collect data about tornadoes. In a project now called VORTEX1, scientists roamed the Plains in 1994 and 1995 and collected data, but not nearly as much as they amassed in 2009 and 2010.

VORTEX1 researchers “learned some very fundamental things about how storms interact with the surrounding air [and] how environmental conditions can influence the formation of tornadoes, which we use in forecasts,” says Schneider. “I’m expecting similar nuggets to come out of VORTEX2.”

Increased computer power has allowed scientists to create better models for predicting tornado formation — if the researchers can find the right data to feed into the computers.

Two kinds of information seem to hold the most promise.

Kosiba and others are examining Doppler radar data and other measurements involving supercells’ descending cool air and rising warm air that could indicate a strong tornado is likely to form.

Other teams are examining data from VORTEX2 mobile radars that show changes in the amount and location of ice and water drops in different parts of supercells. They’re looking for patterns that could signal an early stage of tornado formation.

Forecasters won’t have to wait until the scientists finish analyzing data to begin receiving useful information. Schneider expects to begin nibbling at some useful VORTEX2 nuggets when the American Meteorological Society’s annual Conference on Severe Local Storms occurs in November.

The tornadoes that moved across the southern and central Plains last month illustrate the forecasting gap that the scientists hope VORTEX2 can close.

Schneider’s Storm Prediction Center began issuing alerts a week in advance that the region might face a tornado outbreak because large-scale weather patterns were forecast to be favorable. In subsequent days, the center refined these alerts and stressed the danger. In the minutes before the tornadoes hit, forecasters watching their radars wrestled with the conflict between warning too late or issuing a false alarm. Schneider and others are convinced that hearing about the tornadoes for a week in advance encouraged people to be alert and to take shelter, helping to hold the death toll to six.

Harold Brooks is a scientist at NOAA’s National Severe Storms Laboratory who isn’t part of VORTEX2. He expects the project to produce important advances, but he said they “won’t be as great as [what happened when] the Doppler network” was deployed across the country in the 1980s. “That was sort of like the Wright brothers inventing the airplane. Lots of improvements have been made in airplanes, but it’s hard to top that first big thing.”

He thinks one of the first results of VORTEX2 could be better training of storm-spotting volunteers.

“Fifty years ago, we didn’t even know which part of a thunderstorm to tell spotters to look at for a tornado. Every time we learn more about how storms form, we can train spotters to look for different things. There’s a good chance we could get things that allow us to use spotters more intelligently.”

Brooks says VORTEX2 results could go beyond tornado forecasting. “If we understand the environment better, we could produce better forecasts for renewable energy. Wind-turbine operators want to know whether the wind will shift in 30 minutes. Solar-energy operators want to know when clouds will move in. This could be one of those positive unintended consequences of VORTEX2.”

Williams is a freelance writer specializing in weather and climate; he is a regular contributor to The Post’s Capital Weather Gang.