The vans and compact cars looked like refugees from Mystery Science Theater 3000 as they took to the highway in search of tornadoes. Wind gauges spun, and weather vanes swung erratically from poles anchored to rooftop racks as the caravan headed out.
The scientists inside the vehicles aimed to use the instrument-laden squadron to take close-up readings that would help them pry open the secrets of how such destructive wind storms form. In the year since their final foray - part of a two-year effort known as the VORTEX project - researchers who roamed from Texas to Kansas in search of twisters have begun poring through the data they've gathered.
VORTEX, which involved more than 75 researchers from 10 universities and several federal agencies in the United States and Canada, returned a wealth of observations that participants estimate will take several years to interpret. The ultimate goal: to help save lives by improving the lead time and accuracy of tornado warnings.
North America is not the only tornado hot spot. Australia and Bangladesh also account for a large number of the world's tornadoes. Last week, for example, a twister packing winds estimated at 125 miles an hour spent half an hour leveling 80 Bangladeshi villages, leaving an estimated 615 dead and 34,000 injured.
The United States has the dubious distinction of being the tornado capital of the world. Last year set a record, with 1,233 confirmed sightings in states ranging from Texas to New Hampshire and Florida to Oregon. Since 1950, no state has been spared. Indeed, Southern California experiences as many tornadoes as the Oklahoma City area, says Roger Wakimoto, a professor of meteorology at the University of California at Los Angeles. The difference, he quickly adds, is that California's twisters are much weaker and shorter-lived.
Although tornadoes can occur during any season, they become most active from late winter to early summer, starting near the Gulf of Mexico in March and moving north and east as the seasons progress. Outbreaks typically peak in May and June in Kansas, Iowa, and Nebraska.
Weaker tornadoes, with wind speeds of around 110 miles an hour, last less than 10 minutes. They cut a swath roughly 100 yards wide for about a mile before they disappear. Strong tornadoes have been known to last for up to two hours or more, cutting a path up to 1,000 yards wide and more than 100 miles long. Estimates of the maximum wind speeds in the strongest tornadoes have reached as high as 280 miles an hour.
Because they can develop suddenly and vanish just as fast, tornadoes pose a particularly thorny problem for forecasters.
VORTEX, which stands for Verification of the Origin of Rotation in Tornadoes Experiment, was designed to test current notions of how tornadoes form, as well as provide new information that could be fed into computer models that try to simulate and forecast tornadoes.
During the springs of 1994 and '95, a small army of scientists and graduate students chased storms and tried to surround tornadoes with their armada of specially instrumented cars. They dotted the landscape ahead of tornadoes with small instrument packages called "turtles," in hopes a twister would inhale one and give researchers the inside story. Mobile Doppler radar yielded unique close-up images of storm circulation patterns, while sounding balloons and the National Oceanic and Atmospheric Administration's hurricane hunter aircraft rounded out the assault force.
In the end, the team bagged 10 tornadoes - including a severe twister that hit Dimmitt, Texas, and became the most monitored tornado in history. They also probed numerous thunderstorms with tornado potential.
"VORTEX was able to get more information about tornadoes from more points" than ever before, says Charles Doswell, a scientist at the National Severe Storms Laboratory at the University of Oklahoma in Norman. From a scientific standpoint, he says, the VORTEX data are exciting - and humbling.
Going into the project "we had some sense that we knew what was going on," he explains. Instead, "we saw enough variability in the way tornadoes form that ... each one was unique.... As one of my colleagues put it, we've defined the depths of our ignorance."
As daunting as the VORTEX data may be, the understanding of conditions that lead to tornado-generating thunderstorms has come a long way since Benjamin Franklin saddled up to chase dust devils. "Even before radar and traditional meteorology started after World War II, people thought that a tornado's rotation developed within a storm and then fell to the ground," explains Dr. Wakimoto, a member of the VORTEX team. After all, that's what people saw - a dark funnel cloud descending from the base of a thunderstorm.
By the mid 1970s, researchers at the University of Oklahoma were developing radar that takes advantage of the Doppler effect to detect wind circulation in clouds. Doppler radar measures the shift in frequency that results when its signals bounce back off of moving water droplets in the clouds. If part of the reflected beam rises in frequency compared with the outgoing beam, that part of the cloud is moving toward the radar site; if the frequency falls, that section of cloud is moving away from the site.
This tool gave researchers their first glimpse at the large-scale motion that takes place in the vast, violent thunderstorms whose broad central cloud columns are dominated by winds rotating counterclockwise. This circulation, which meteorologists call a mid-level mesocyclone, begins about a mile above the ground and can reach diameters of up to 12 miles. Such "supercell" thunderstorms have spawned some of the most violent tornadoes. Researchers suggested that perhaps over time the mesocyclone drops, and when it hits the ground, tornadoes can form.
Yet that left scientists puzzling over why supercells rotate. One suspect was Earth's rotation, which puts a large-scale spin on storms ranging from hurricanes to nor'easters.
In 1978, researchers from the National Center for Atmospheric Research in Boulder, Colo., and the University of Illinois did what only computer modelers can do: They turned off Earth's rotation. In their simulations, the rotating supercells still formed. Further research showed how low-level winds that changed direction or rapidly changed speed depending on their height generated horizontal whorls of air. These horizontal vortices were drawn into a supercell's central cloud column by strong updrafts, giving the storm its rotation.
This still left scientists up in the air about how rotating air develops near the ground. More field observations and computer time led to the discovery that as rain-cooled downdrafts to the north of the storm sink, they wrap around the warm central updraft, forming a vortex of air along the boundary between them.
By the time these vortices reach the ground, they have evolved into low-level mesocyclones roughly four miles across that tend to get drawn back up into the storm on its southwest side - the region of the storm where tornadoes often form.
"What we don't understand," Dr. Wakimoto says, "is how you get from five or six kilometers to a tornado."
Nor has VORTEX cleared the air.
"Our models don't suggest the kind of [tornado] evolution we saw in VORTEX," says Louis Wicker, a meteorologist at Texas A&M University. For example, he notes that the June 2, 1995, Dimmitt tornado - "a violent, long-lived tornado" - grew amid conditions thought to be twister-killers. "That's a significant wrinkle. It tells us there's something about formation we're overlooking."
Doswell notes that a 1994 VORTEX observation tracked an intense tornado that evolved with a fledgling updraft that occurred well away from a preexisting mesocyclone. In that case, a swirling vortex of air near the ground likely was "stretched" by the updraft. As it stretched upward, the vortex narrowed and gained speed, much as a twirling ice skater speeds when she draws her arms close to her body. Indeed, tornadoes are now thought to rise from the ground to the cloudbase; the funnel "cloud" forms only after the tightly whirling column of wind has established itself.
The 1995 observations, Doswell adds, suggest that immediate "triggers" to tornado formation may lie in subtle low-level conditions in and around the storm and may be very difficult to measure. Unfortunately, he says, unless the storm is within 20 miles or so, Doppler radar won't pick up the low-altitude activity.
Meanwhile, as the National Weather Service deploys more Doppler radar, researchers are discovering that more thunderstorms contain mid-level mesocyclones. But they haven't seen as many tornadoes as the number of storms with mesocyclones would suggest.
Where once such storms were thought to generate at least half of all tornadoes, some researchers say this figure could drop to 20 percent or less. Thus, one of the key tornado precursors that Doppler radar was designed to detect seems to be linked to fewer and fewer twisters.
"The challenge is that you're looking for very subtle effects in a storm with 100-mile-an-hour downdrafts, 50-million-volt lightning bolts, and baseball-sized hail," says Tom Grazilus, founder of the Vermont-based Tornado Project, a clearinghouse for tornado information.
Yet for all the uncertainties, he adds, "This is the most exciting time I can remember" in tornado research. Digging into the VORTEX data may well sustain that attitude for years to come.