The biggest problem with tornado forecasting today is not that weathermen offer too few warnings. In fact, they give too many.
For every tornado forecasters get right, they predict five more that never develop. The ratio may be as bad as 1 in 20. If they don't stop crying "wolf," meteorologists concede, people will pay them little heed.
That is one reason weather scientists are poring so closely over the information gathered from the tornado system that swept through Oklahoma and Kansas a week ago. If a storm that claimed more than 40 lives and caused some $500 million worth of damage can have a silver lining, it is this: It gave scientists one of their best glimpses yet at how such storms form.
If they can crack that meteorological mystery, there's a good chance they'll be able to forecast tornadoes much more accurately, scientists say.
"We have great data," says Erik Rasmussen, a research scientist at the National Severe Storm Laboratory in Norman, Okla. "But I've said that before and when I've started to pick it apart, there's been one key piece missing. Maybe this time, we have it all."
The storm stayed on the ground so long and created so little rain that researchers at the storm lab got more than three hours of observations and data. And since it tore through areas with plenty of roads, researchers were able to track it close-in with their instrument-laden cars.
Just as important, they were able to position those vans on three sides of the tornado: front, back, and to the south - a key area of spiraling curtains of rain that storm chasers call "the bear cage."
In a few weeks, lab researchers hope to be able to make preliminary conclusions about the key mystery surrounding such storms: How do spiraling winds at high altitude reach the ground and become tornadoes?
Thanks to a federally funded project called Vortex begun in 1995, researchers already know a good deal.
During storms, winds at high altitudes often rotate at enormous speeds, but they spin horizontally. (Imagine a long column of air, shaped like a flexible rubber hose, spinning like a rolling pin over a bread board.) A huge downdraft of air starts the process, pushing the middle of the rotating "hose" down to the ground. This forms two columns of air, now spinning vertically.
But those aren't tornadoes - not yet. For the real twister to form, the accompanying air from the downdraft has to hit the ground but stay concentrated under one of the spiraling columns. The twisting column picks up that air and sends it shooting up, "stretching" the column upward. The air shoots up so rapidly that it draws in the column, making it much thinner at ground level and rotating much faster (the same way a spinning ice skater accelerates when she pulls in her arms). A tornado is born.
What researchers don't yet understand is how the downdraft pushes down the rotating "hose" of air in the first place. Does it drag it down by sheer force or is the downdraft cold air that naturally sinks?
Another mystery: How does a separate small spinning top of air at ground level hook up with the larger rotating column? And is that crucial to forming a tornado?
By understanding these mechanics, researchers will be able to identify the key elements that separate wanna-be tornadoes from the real thing. And, they hope, that will lead to better forecasts.
"I'm excited to see what comes out of it," says Louis Wicker, a meteorologist at Texas A&M University in College Station. "If ... we understand what's going on then I think we could reduce the false-warning events."
Concerns remain. Unlike hurricanes, tornadoes form so quickly and move so unpredictably, that people often only have minutes to react. And today's radar systems, while much improved, still can't give forecasters the entire picture of what's going on. So, even though they may discover the key elements that form tornadoes, researchers can't guarantee they'll be able to see those elements every time a storm whips up.
Because of those technological limits, Dr. Rasmsussen says, forecasters will probably still cry "wolf" occasionally and get it wrong.