Tornado Alley, USA: new map defines nation's twister risk.

Residents of Pontotoc County, Okla., take note: You're sitting smack dab at the center of a meteorological bull's-eye. Once every 4,000 years or so, a tornado strong enough to rip the roof off a house will sweep across the very spot in southeastern Oklahoma that you call home. That doesn't sound very frequent, does it? Truth be told, it isn't. Fewer than 1,200 tornadoes are spotted in the United States each year. A typical twister damages an area of only about 13 square kilometers. In a country that boasts almost 9.4 million square kilometers, that leaves a lot of undamaged real estate.

Even in Tornado Alley--the nickname for the swath of the central United States where twisters are most common--tornadoes are a rare phenomenon. New models suggest that in this L-shape region, which stretches from western Iowa down through Nebraska and Kansas to southern Oklahoma and then over Arkansas and Louisiana to southeastern Mississippi, any particular spot can expect on average to wait from 4,000 to 10,000 years between roof-ripping twisters.

TWIST AND SPOUT Tornadoes are violently rotating columns of air suspended from so-called cumuliform clouds, which are dense, tall, and characterized by rising mounds, domes, or towers of condensed water vapor. Twisters are home to the strongest winds on the face of the planet (SN: 5/15/99, p. 308), and they come in various sizes, from tightly whirling funnels just a few yards across to mile-wide behemoths.

By definition, a swirling vortex of wind isn't a tornado unless it's in contact with the ground. Also, a twister's size isn't correlated with its strength. Large tornadoes can be weak, and some of the smallest funnels are the most destructive. A tornado doesn't even have to have a visible funnel, says Charles A. Doswell III, a meteorologist at the University of Oklahoma in Norman. The funnel cloud associated with most tornadoes results from moisture condensing out of humid air as the vortex accelerates and the air pressure inside drops.

Because tornadoes can't be characterized by their apparent size--and because some are never seen at all--they're rated according to the damage they cause to humanmade structures. The Fujita scale is named for the University of Chicago meteorologist, Ted Fujita, who developed it in the 1970s. The damage ratings run from F-O, which causes slight damage to chimneys, to F-5, in which well-built homes are demolished, steel-reinforced concrete structures are badly damaged, and automobiles or objects of similar heft are lofted distances of more than 100 meters.

Meteorologists have reliable, comprehensive statistics only for tornadoes during the past 50 years or so, says Joseph T. Schaefer, a meteorologist at the National Weather Service's Storm Prediction Center in Norman, Okla. The number of twisters reported from year to year is highly variable. Over the decades, however, the average number has been increasing by a dozen or so each year. In the 1960s, there were about 650 twisters reported annually. In the 1990s, the average was more than 1,100. Schaefer discussed this finding at the annual meeting of the American Meteorological Society in Orlando last January.

What's causing the proliferation? "It's probably not climate change," says Schaefer. "It's people change."

As people move out of cities and disperse over larger areas, they are more likely to spot tornadoes. As increasing media coverage makes travelers and residents of suburbs and exurbs more weather-aware, they are more likely to report twisters. The growing popularity of cellular phones, which enable people on the road or in areas of downed phone lines to call in sightings to emergency centers, will probably continue to fuel the apparent increase, Schaefer speculates.

HEAD FOR COVER! The few decades of reports now available simply don't include enough twisters to permit a reliable analysis of the nation's tornado risk. Therefore, meteorologists such as Harold E. Brooks of the National Severe Storms Laboratory in Norman, have turned to computer simulations. In their most recent study, Brooks and his colleagues started with weather and damage data from 10,000 tornadoes that occurred between 1921 and 1995. Using just five parameters, their simulation generated a realistic mix of 4 million tornadoes over 30,000 years that the team then subjected to statistical analyses.

First, the model calculated the probability of a tornado occurring on a particular day. If a tornado appeared, the model computed the probability of companion twisters being spawned that day by other thunderstorms. Then, by reference to data from the original group of twisters, each tornado was assigned a Fujita scale rating, a path length over which damage would occur, and a path width.

The simulation mapped the area damaged by each tornado into the 80-km-square grid box where the twister first touched down. The greater the tornado damage in a box at the end of the simulation, the higher the probability that any point within it would have been hit by a twister, says Brooks. From that probability, a simple calculation tells the average length of time that passed before a tornado rated F-2 or larger will hit any given spot. An F-2 rating signifies winds strong enough to rip the roof off a frame-built home.

The researchers presented their analyses at the meteorological meeting in January.

Southeastern Oklahoma fared worst in the simulation, with each point in the area getting hit once every 4,000 years on average. A large portion of the central United States--stretching from the Colorado-Kansas border to western North Carolina and from the Gulf Coast to southern Minnesota--suffered a twister approximately once every 10,000 years. Nevada suffered tornadoes so infrequently that points there might get damaged only once every 5 million years.

Brooks suggests that the exact probabilities of being hit by a twister in extremely low-risk areas such as Nevada should be taken with a grain of salt. The number of tornadoes from those regions in 75 years may have been too small to provide accurate input to the model. Tornado hazard in such regions is probably somewhat higher than the model predicts, Brooks speculates.

Other meteorologists think so, too, for slightly different reasons. Doswell points out that many areas of the U.S. West, and even parts of the Great Plains, are so sparsely populated that many tornadoes are simply never observed. It's often 100 km or more between the small towns in these areas, he notes. Even if tornadoes are spotted, they're probably underrated on the Fujita scale because there's little for the twisters to damage except crops, rocks, and the occasional tree or telephone pole.

"All we have to go on is the data we have," says Doswell. "We know that it's inaccurate, but it's the best that we've got."

Schaefer agrees: "We've seen what we've seen. We're just not sure we've seen everything."

DAMAGED GOODS? Other factors, many of them related to the Fujita damage scale and how it's applied, might significantly affect the results of the tornado-risk assessment conducted by Brooks and his colleagues.

First of all, says Doswell, a tornado's F-scale rating is based on the worst damage found anywhere along its path. In a simulation, that might tend to overestimate the hazard because the high F-scale rating would apply to the entire swath of damage. On the other hand, many tornadoes may be underrated because the strongest winds in a funnel may occur at a time and place where there's nothing much to damage.

Schaefer notes that many of the F-scale damage ratings assigned in his database to tornadoes between 1950 and 1975 may be too high. His analyses show that even while the average number of tornadoes has appeared to increase each year, the number of twisters rated F-2 and higher has dropped significantly during that period. Schaefer again suggests that people, and not an actual change in climate or tornado behavior, are to blame for the anomaly. In the mid-1970s, soon after the Fujita scale was invented, trained observers went into the field to assess tornado damage. For earlier twisters, researchers have used newspaper accounts to assign F-scale ratings. Exaggerations in reporting, among other factors, may have led to inflated damage estimates.

And maybe the Fujita scale's assessments are too simplistic, says Schaefer. For one thing, the field personnel should consider the quality of construction of the structures that have been damaged. The destruction of one flimsy building shouldn't skew the F rating of an entire tornado, he notes.

Likewise, the assessors need to evaluate whether other factors, such as windblown debris, aggravated damage. A tornado that struck Xenia, Ohio, on Sept. 20, 2000, is a perfect example, he says. The home that suffered the most damage collapsed when the roof blown off the house across the street landed on it. Houses next-door escaped nearly unscathed.

Many factors influence the extent of damage that tornadoes inflict, says Doswell. Besides the strength of the structure that's hit and the quantity of windblown debris, these factors include how long the winds blow on the structure and the rate at which the winds intensify. "We don't know a lot of the details of how winds behave around tornadoes," Doswell laments.

Although scientists know nature's overall ingredients for tornadoes (see box, page 297), mystery enshrouds the small-scale phenomena that cook up a twister. In fact, says Doswell, the real question is why there aren't more? Only a tiny fraction of the storms that occur each year produce tornadoes, and the ones that end up spawning damaging funnels aren't much different from those that don't.

Tornado Scarcity

Twister total through April is one of the lowest on record

For the first 3 months of this year, Tornado Alley--indeed, the
whole nation--caught a break. Through March 31, only 40 tornadoes
had struck the continental United States. This year's
start is tied for the ranking of third lowest since 1950, says
Joseph T. Schaefer, director of the National Weather Service's
Storm Prediction Center in Norman, Okla. The record low count, 16,
occurred during the first quarter of 1969, and only 18 tornados spun
across the lower 48 states during the same period in 1951.

As of May 1, this year is still running well behind normal with
only 154 tornados, as compared with an average of 270 for Jan. 1
through April 30. For this year's dearth of funnels, twisterphobes
can thank the high-altitude river of air known as the jet stream. Its
southerly position this spring prevented masses of warm, moist
air in the Gulf of Mexico from moving north to the U.S. plains,
where they could collide with cold air spilling down from Canada.
That mixing creates the thunderstorms that spawn most tornadoes,
explains Schaefer. Instead, the jet stream this year steered
most storms out over the Gulf or along the coast.

Could this good luck last the rest of the year? Statistically, it's
likely. From 1955 through 1999, the 11 years that tallied the lowest
January-to-March tornado totals also experienced fewer-than-average
twisters in the following 9 months, says Harold E. Brooks, a
meteorologist at the National Severe Storms Laboratory, also in Norman.

When totals were adjusted to account for the gradual increase
in tornado reports since the 1950s, each of the 11 years with the
fewest first-quarter tornadoes had an average of just 71 twisters.
That period for the other 33 years registered, on average, a
whopping 155 funnels. Brooks and his colleagues found that this
disparity carried through to the rest of the year. The 11 years
with low-tornado winters counted, on average, 943 twisters in
the period from April 1 through Dec. 31, while the other years
racked up 1,064 funnels.--S.P.

FREE-RANGE FUNNELS While tornadoes are most common on the Great Plains and throughout the Mississippi River valley, they can occur almost anywhere in the United States. Rough terrain may inhibit the formation of twisters, but it doesn't prevent them altogether. That's because the storm clouds that produce tornadoes are often 12 to 15 km from top to bottom, says Doswell. "The atmosphere doesn't know about the calendar, the clock, or a map, and an atmospheric feature that size isn't going to care about a bump on the ground," he adds.

The idea that tornadoes don't cross mountain ranges, rivers, or other geographical features simply isn't true, says Doswell. In July 1987, an F-4 tornado roared across the Continental Divide in Yellowstone Park at an elevation of at least 3,000 m. The twister that struck Salt Lake City in August 1999 dropped down one side of a canyon and climbed up the other. The deadliest tornado in U.S. history is the so-called Tristate tornado of March 18, 1925, which killed 695 people. It swept across the Mississippi River as it carved a 350-km corridor of devastation across Missouri, Illinois, and Indiana.

Likewise, it's only a myth that twisters avoid large cities. Salt Lake City, Oklahoma City, Miami, Nashville, and Fort Worth, Texas, are just a few of the downtown areas that have been damaged by tornadoes in recent years. The only thing that protected them previously had been good luck and their small size, says Doswell.

Even though tornadoes are extremely rare events, they wreak havoc when they do strike. "All the statistics in the world don't help you if you're in the path of a tornado," he notes. "Somebody's going to be unlucky every year."