Skip to main content
ABC News
Tornado Town, USA

On the evening of May 3, 1999, a massive tornado tore through the Oklahoma City area. Known today as the Bridge Creek-Moore Tornado, it’s infamous for its size (a mile wide) and strength (wind speeds reached 300 miles per hour, on par with a Tokyo bullet train). It moved, as tornadoes so often do, from the southwest to the northeast, touching down in the rural plains before churning its way through the suburb of Moore and up to Midwest City, just east of downtown — which was where it pulverized my dad’s truck.

My dad, Howard Koerth, moved to Oklahoma in 1994 to teach art at Rose State Community College in Midwest City. He was there May 3, right in the tornado’s path. Instead of going to the storm shelter, he opened the back door of his building and watched the fat funnel tear apart an auto dealership. The tornado was gray, tinted with red from the layers of clay-filled topsoil it had peeled off the Earth. If you watch video of it today, you see it surrounded by a haze of confetti. When the camera zooms in, the ticker tape turns out to be, instead, a blizzard of two-by-fours, siding, whole trucks. Sixteen years later, Dad has yet to exorcise that image from his mind and he’s still asking me about the Bridge Creek-Moore tornado. Or, rather, he asks me about its sister storms — tornadoes that, to him, seem to follow the same path, flattening the same places over and over. Especially Moore. Always Moore.

He called me in 2003, when a slightly less powerful tornado — and, by “less powerful,” I mean one classified as “devastating” (an EF4) rather than “incredible” (an EF5) — hit Moore. He called me in 2010, when another EF4 struck the town. He called me in 2013, when Moore was hit — improbably — by a second EF5. He always asks the same question: “What is going on here?” One town. Sixteen years. Four big, powerful tornadoes. It’s a hell of a coincidence. Can it really be just the work of random chance?1

My dad isn’t the only person vexed by this question. And the question isn’t limited to Moore. Instead, asking about Moore is really asking a bigger question: Why do tornadoes strike some places and not others? About 1,000 tornadoes touch down in the United States every year, and the majority of them happen in one of two areas — a vertical swath running from roughly Nebraska to Texas and a horizontal swath from Oklahoma to Georgia. Within that, there are places where tornadoes seem to cluster, such as Birmingham and Little Rock, said Tom Grazulis, a researcher who, in the 1980s, compiled records of American tornadoes back to the 17th century for the Nuclear Regulatory Commission. But those clusters usually happen over longer periods of time, say, 40 years or 100 years. He couldn’t think of any other place hit as hard in such a short period of time as Moore.

EF1 71%
EF2 21
EF3 6
EF4 1
EF5 <1
Distribution of U.S. tornadoes, 1994-2012

Source: Environmental Research Letters

Nobody knows how likely it is that a given town would be hit by four violent tornadoes in 16 years; if we knew that, then we’d also know whether Moore really is especially tornado prone, or just suffering a streak of bad luck. But we do know big tornadoes, themselves, are rare. Devastating EF4s made up 1.37 percent of all the tornadoes that hit the U.S. from 1994 to 2012.2 Just 0.14 percent were incredible EF5s.

And that’s enough to make Moore’s recent history turn heads. People who live in the Plains states, as I once did, have a special relationship with tornadoes, wary but familiar, like your grandma’s dog that’ll bite if you aren’t careful. This is a part of the country where little kids dream about growing up to be storm chasers. Where tornado sirens go off every Wednesday at lunchtime, just as a test of the system. It’s a part of the country where art professors like my dad duck outside for a peek at one of the most powerful tornadoes in recorded history.

But this thing with Moore even weirds out Oklahomans. “For years people have asked me, ‘What about Moore?’ ” said Gary England, a retired TV meteorologist who shepherded generations of Oklahomans through more than 40 tornado seasons. “People talk about topography. They talk about geomagnetic forces. I think it’s very unusual. But I think most scientists would probably tell you it’s just a roll of the dice.”


The first big tornado recorded in Oklahoma happened on April 25, 1893. Witnesses claimed it was more than a mile wide. It hit Moore, which had just been incorporated that same year. Yes, one of the first things that happened in the town was the destruction of the town.

But that still doesn’t mean that a tornado in Moore is anything more than a roll of the dice, as England put it. Even Grazulis, who was surprised by what had happened to Moore recently, thinks the events of the last 16 years reek of random clustering. That’s because, all things considered, it’s no big surprise that a place in central Oklahoma is being hit by a lot of tornadoes. There’s a mystery about the risks associated with Moore, but it’s a mystery that’s complicated by matters of scale. If you zoom out — look at our hemisphere or our continent — the part of the country Moore is in really is more likely to be hit by tornadoes than most other places, that’s not random. But the fact that Moore, specifically, is being hit over and over … that could still just be bad luck.

To understand why, you need to know a little about how tornadoes work. All tornadoes that touch down in central Oklahoma start their lives in two places: the Gulf of Mexico and the Rocky Mountains. Warm, moist air comes up from the Gulf in the south. From the west, air ripples over the mountaintops, losing moisture and heat as it goes. This odd couple meets on the downward slope into the plains. The air currents from the south tend to be at a lower level of the atmosphere than those from the west, which creates an opportunity for naturally buoyant, hot, moist air to rise up through layers of cool, dry air. That produces condensation, just like water droplets form on the outside of a cold can of soda on a hot day. Now you have the ingredients of a thunderstorm: moisture, rising air currents, and the instability that happens when Gulf air and the mountain air jockey for position.

This is why the infamous Tornado Alley of the Plains states is Tornado Alley. It’s the place where the Gulf air and the mountain air meet. “The central part of the U.S. is incredibly well designed to produce tornadoes,” said Harold Brooks, senior scientist at the National Severe Storms Weather Laboratory in Norman, Oklahoma — a suburb just south of Moore. There are a few other places on Earth with similar profiles, but they have limitations the Plains states just don’t have, such as a mountain range like the Andes, which is thinner and can’t dry or cool air as well as the Rockies. The central U.S. is the most likely place for tornadoes to form, on this continent and anywhere in the world. Insomuch as it sits right in the middle of that, yes, Moore is at a higher risk.

But that’s Moore in comparison to Cleveland or Buenos Aires. What about at the smaller scale: Moore in comparison to, say, Tulsa? That’s a question that hasn’t been explored as much as the science of tornadoes themselves. Researchers at the National Severe Storms Laboratory say there isn’t much emphasis placed on the question of whether a specific region or town might be more prone to tornado activity than another. Instead, they’re more interested in how the storms form, how to track them and how to get more accurate warnings out faster.

But some scientists are trying to find out more about the distribution of tornadoes. Brooks, along with fellow meteorologists Patrick Marsh and Gregory Carbin3 are among the scientists who are fascinated by the possibility that Moore (and certain other places) really could be tornado magnets. They’ve published research relating to it and written about it on blogs.4 But none of them do that work as their main job. “What about Moore?” is a question guys like these talk about over beers at the end of the day, Brooks told me. The research they are doing might one day make it easier for them to answer that question. Right now, though, they can’t.

There are three big problems. First, tornadoes are really complex systems. They only form if a storm begins to rotate vertically, a corkscrew of air rising high into the sky. Scientists think that rotation starts because of wind shear, quick changes in wind speed or direction at different levels of the atmosphere. Imagine holding a piece of Play-Doh between your flattened hands. If you move them past each other, in opposite directions, the dough in between rolls up into a tube. Similarly, wind shear creates horizontal columns of spinning air. When those get caught by rising warm air, they can tip up, become vertical, and turn a thunderstorm into a supercell. A tornado happens when that spinning supercell touches the ground.

If hurricanes are nature’s nuclear warhead, tornadoes are its smart bomb.

Each of the steps in the storm’s formation – from the meeting of the Gulf and mountain air currents, to the moment the supercell stretches down and scrapes its fingers through the dirt – involves forces scientists don’t totally understand and elements of random chance. Add it all together and you have a dark, churning mass of mystery and probability.

For Gregory Carbin, that reality sank in as he watched the Bridge Creek-Moore tornado. From Carbin’s vantage point, just outside the Severe Storms Laboratory, the tornado itself wasn’t visible, but the supercell was. It rose up, black and boiling, a chimney belching angry water vapor 50,000 feet into the air. And Carbin thought, “It’s so fragile.”

“It occurred to me that, you know, what would it take for it to just be a rain shower or nothing at all? Everything needs to come together just right, and if you don’t have those conditions, if something is off — and we don’t even know what that something might be — you don’t get a tornado,” he said.

The second problem is that tornadoes are pretty rare. One thousand a year, scattered across the continent, does not produce many data points at the scale of an individual city. Most days, there aren’t tornadoes anywhere. That problem is exacerbated by the third issue: Scientists really only have about 50 years of really good tornado documentation. Essentially, Brooks told me, scientists can’t tell us whether what’s happened in Moore is abnormal because they don’t know what a “normal” amount of violent tornadoes is. With all of that, Brooks said, there’s not a good way to clearly tell the difference between patterns and pareidolia. After all, the human brain is primed to find significance in the random. In the creaky corners of our neural pathways, a jumble of rocks can become an old man, a coat hanger can become a drunk octopus, a bunch of craters on the moon give us a friendly smile. It’s so easy for a few random events to make one small town look like a tornado magnet. It would be harder not to see it.

And Moore, itself, facilitates that pareidolia. Located about 11 miles due south of downtown Oklahoma City, Moore is a town for which Interstate 35 serves as a virtual Main Street, running through the middle of town. Businesses cluster on either side: A movie theater, Hollie’s Flatiron Steakhouse, Furr’s Fresh Buffet, the skating rink, Leon’s Pharmacy. Even the public library, community center and the Chamber of Commerce abut the frontage roads.

The town may have been incorporated in 1893, but until suburbia dropped out of the sky and landed on it, Moore was so small that there was no real historic center to anchor businesses to. Stretching away from the highway, on either side, streets of tidy, middle-class homes wind around parks and curve into cul-de-sacs. Many have brick facades and a stubby look, hugging the ground like Corgis. It took me a minute to realize that this was because the construction is almost uniformly slab on grade. Central Oklahoma is tornado prone, and the National Weather Service recommends basements and storm cellars as first-line tornado shelters. But few buildings in central Oklahoma are built with either one.

Like many places with this kind of history, Moore is somewhat amorphous, its 22 square miles bleeding into Oklahoma City to the north and the more well-known (and well-off) college town of Norman to the south. It’s easy for even longtime residents to be unsure of where their city ends and another begins. The official size is misleading in other ways, as well. That’s because Moore’s school district is 159 square miles, encompassing parts of the southern end of OKC, itself. The result is a colloquial Moore that is much larger than what the census might tell you. “The largest high school in Oklahoma City is Westmoore High School. So people think of all that southwestern Oklahoma City as being Moore,” Brooks said.

Oklahoma, 2003 YouTube / KOCO

Keep that in mind while you think about the tornadoes that hit the Oklahoma City area on May 31, 2013. This was 11 days after an EF5 destroyed large chunks of Moore, grinding houses, parks, churches and two grade schools into rubble. This storm dropped at least five individual tornadoes all over the region. Sirens went off in Moore that night. Plenty of people who lived there fled for their lives. Among them was Chris Fox, his wife, two kids, his mother and his grandfather. That night, a local TV news anchor advised people to get out of the tornadoes’ way by any method possible – including by car. So they did.

“Which, had I been in my right mind, we would have stayed put and would have been fine,” Fox said. “What we ended up doing, we drove into the path of this smaller spin-off and we had to pull out of traffic into a church parking lot. As we’re pulling off, trees are coming out, roots are coming up, rain is going sideways, my kids are crying and screaming. We end up arriving at this church with 20 other people who have come from different directions to get here. The inner doors are locked. We’re in a vestibule. And this guy whips out a crowbar from his truck.”

Fox, who went on to found a community volunteer organization called Serve Moore, survived his brush with both the fury of nature and breaking and entering. But the tornadoes he and his family were fleeing never hit the place they were fleeing from. According to the National Oceanographic and Atmospheric Administration, those tornadoes touched down in El Reno, southwest Oklahoma City and other suburbs … but not Moore.

Moore’s tornado problem exists both as data and as mythology. There are the tornadoes that hit Moore and then there’s the pervasive sense that Moore gets hit by tornadoes.

Consider the suburb of Norman, which sits just to the south of Moore. It’s the next set of exits off I-35. Patrick Marsh, of the Severe Storms Weather Laboratory, told me initially that Moore had been hit by more tornadoes in recent years than Norman. But, as we continued talking, he went through that recent tornado history and ended up stopping and correcting himself. Actually, Norman probably had been hit about as frequently as Moore, he said. It’s just that Norman had avoided the big EF4s and EF5s that everybody remembered, and so Norman hadn’t taken on the status of being tornado prone. “I don’t think you can statistically prove that your risk is any lower than what happened two miles up the road in Moore,” he said. “But everybody in Norman thinks, ‘Oh, I’m safe. Because the tornado will hit Moore.’ ”

Oklahoma, 2013 YouTube / Michael Lynn

Between the large-scale likelihood of tornadoes, the pareidolia and the self-mythologizing, I was ready to believe that the Moore mystery wasn’t really that mysterious. Bad luck and supposition seemed to account for everything. That’s certainly the sense you get from a cursory glance at the historical data. Harold Brooks showed me a map of Oklahoma City sprinkled with multicolored tracks of all the tornadoes that had gone through the area since 1880. There’s no obvious confluence over Moore. The whole region is littered with tornado tracks. Bethany, a town on Oklahoma City’s northwest side, has been hit seven times since 1930. On the map view, it looks about as beleaguered as Moore. Meanwhile, there’s a chunk of northeast Moore that’s never been hit, at all.

But then Brooks brought out one more data set. In the late 1990s, he’d been a part of an effort to quantify what was normal and what wasn’t about the distribution of tornadoes. As I already mentioned, this is difficult work and it’s made even more complex by fudging and inconsistencies in the historical documentation.

National Weather Service records go back to 1950. Tom Grazulis’ data set, which is based on newspaper accounts and records kept by local postmasters, picks up the trail back to the 1600s, and Brooks considers it reliable to about 1870. But both of those are likely missing a lot of smaller tornadoes and tornadoes that landed in lightly populated places. Perspective matters. In 1880 a mapmaker promoting colonization of the Oklahoma Territory claimed the area was virtually tornado free, Grazulis told me. Even when these records don’t miss a tornado, they have clearly been fudged in many ways. It’s unlikely, as Brooks pointed out, that so many tornadoes would start punctually at the top of the hour, the way they tend to in the historical records.

But these records can still tell us something useful about the statistical probability of a tornado’s touching down in one place and not in another. At the very least, it tells you what is possible. Brooks analyzed the data to find the times of the year and places in the country where tornadoes seemed to be more likely to happen. It wasn’t exactly a prediction of the future – more a detailed observation of the past. Using this, he came up with the most likely place and time for a big tornado: a town called Pauls Valley, Oklahoma, on May 2.

“So, has a tornado ever hit Pauls Valley on May 2?” I asked him.

“No. I don’t think so,” he said.5

But when I laughed, he explained. That location comes with a caveat. It’s got some built-in margin of error to make up for all the poorly collected reports and missed tornadoes of decades past. Scientists call this smoothing the data, and Brooks’ estimate is smoothed to within 50 miles or so.

Moore is 46 miles north of Pauls Valley.

Brooks presented this data at a conference on April 30, 1999. Three days later, the Bridge Creek-Moore tornado came to town. “In some sense,” he said, “That tornado on May 3 was about as likely of a violent tornado as you could imagine.”

Michael Bewley was 11 years old in 1999, when the Bridge Creek-Moore tornado flattened the house he shared with his mother on the outskirts of Moore. One day, they had a small, neat home on three acres at the end of a long dirt road. The next day, they had rubble.

Bewley and his mother had no basement. They could have crawled into the bathtub, pulled a mattress over themselves, and hoped for the best. Instead, they ran for the car. “She was a waitress and she grabbed her time cards, I grabbed the dog, and we left,” he told me. When they came back later that night, everything was gone. Their belongings had been crushed, thrown and rained on. Bewley is certain they wouldn’t have survived if they’d stayed.

But, in one sense, he did stay. Bewley still lives in Moore. Today he manages Chris Fox’s Serve Moore foundation. Bewley has an infant daughter whom he plans to raise in Moore and who has already taken her first turns in the storm shelter. These are the realities of his life: Has it been shaped by chance, or something else?

We can’t completely discount the possibility of something else. In 2004, Brooks’ colleagues Chris Broyles and Casey Crosbie published a paper that analyzed the locations of all the EF3, EF4 and EF5 tornadoes that touched down between 1880 and 2003. They focused on these three classes of tornado because that filters out the smaller type of tornadoes that more easily get left out of records. By smoothing the data in this way, the researchers saw some places where larger tornadoes really do seem to be more common.

Looking at it this way is looking at tornadoes on a zoomed-in scale, regional instead of national. At this level, Moore still isn’t unique. But it is part of a clique — a gang of cities and counties marked by the invisible target painted on their backs. Broyles and Crosbie drew a frequency map of the 979 big tornadoes to touch down in 123 years, showing the number of tornadoes per 1,000 square miles. Plotted out this way, they found clusters. There are dark blobs – tornado alleys within tornado alleys – scattered across the continent. One of those blobs sits over central Oklahoma, north of the Canadian River, stretching from Oklahoma City to Tulsa. Moore is a part of that blob. Other places, including Fillmore County, Nebraska, and Union County, Mississippi, appear to be even more prone to big tornadoes.

America’s mini-tornado alleys, as identified by Chris Broyles and Casey Crosbie of the National Severe Storm Laboratory. This map is reprinted from their 2004 paper. Moore is part of the handgun-shaped blob hovering over central Oklahoma.
America’s mini-tornado alleys, as identified by Chris Broyles and Casey Crosbie of the Storm Prediction Center. This map is reprinted from their 2004 paper. Moore is part of the handgun-shaped blob hovering over central Oklahoma.

Courtesy of Chris Broyles.


This study wasn’t perfect. For one thing, Brooks said, it’s probably no coincidence that the highest frequencies were east of the Mississippi River – where the population density, even in rural areas, is higher than in Oklahoma and other Plains states. That higher population density probably means more thorough reporting of tornadoes. It’s also possible that there are differences between locations in how tornado damage is recorded — and, thus, in how the tornadoes, which are classed based on the damage they cause, get counted.

Recently, a Severe Storms Laboratory research scientist named Corey Potvin teamed up with Brooks and Broyles to re-evaluate the mini-tornado alleys data. They tested out some new ways of accounting for flaws in historical records and calculated the probability that these mini-alleys occurred randomly was just 3 percent. In October of 2015, they presented the results as a poster at the National Weather Association Annual Meeting. Their conclusion: “At least some of the mini-tornado alleys likely are real.” Potvin now thinks that may be a bit premature to say and there are a lot of caveats that go with it, but he is confident they aren’t just a relic of the sampling: garbage data produced by flaws in the way the tornadoes were documented and categorized. What’s happened in Moore is shaped by chance — but it’s also, probably, more than that.

Unfortunately, this is where tornado science dusts its hands and wanders off for a beer. Meteorology can tell us about how tornadoes form at the continental scale. Detailed study of the historical records can tell us about regional probabilities. But when you get to the hyper-local level — the real question of, what is up with Moore? — scientists go mute.

Luckily, we have insurance agents. (If anybody would know about the risks of natural disasters, it’s the insurance industry, right?) And from their perspective, Moore just isn’t that special. People who live in Moore don’t pay any more in home insurance premiums than people in nearby communities around OKC and central Oklahoma, said Robert Hartwig, president of the Insurance Information Institute. Frankly, he told me, insurers are more concerned about the thunderstorm that moves across the whole state than they are about the tornado that drops from it to wreck part of a county or two. Oklahomans have the fourth-highest property insurance premiums in the nation, and much of that is tied up in risks that might be tornado related, but aren’t tornado specific, such as hail, straight-line winds, tree branches crashing through the roof.

That’s because, unlike a hurricane, which can flatten property for hundreds of miles, tornadoes are a more discrete threat. The Bridge Creek-Moore tornado left a mile-wide path of complete destruction, but houses a few blocks away went untouched. Most people who got hit by that tornado haven’t been hit by any of the others. Parts of Moore have never been hit, at all. If hurricanes are nature’s nuclear warhead, tornadoes are its smart bomb. That difference impacts individual risk. And so the hurricane-prone states of Florida, Louisiana and Texas come before Oklahoma on the list of states with the highest premiums.

Oklahoma, 2016 YouTube / KOCO

Knowing that, it becomes less surprising to learn that during the 16 years when Moore has been earning its reputation as America’s tornado magnet, it’s also been growing like gangbusters. A 2014 Census Bureau report showed a 41.3 percent increase in population between 2000 and 2013. “Our growth rate is higher than the state average and is typically one of the highest of the larger cities in Oklahoma,” said Deidre Ebrey, Moore’s director of economic development. That’s not because of the tornadoes. (If anything, it’s probably because of Oklahoma’s oil and gas boom.) But if you’re looking for a place to live near OKC, you could do worse than Moore. As many people who live there told me — the cost of living is low, the schools are good, the commutes are short. And you probably aren’t any more likely to be hit by a tornado than you are in a neighboring suburb.

Even if evidence comes along someday to prove that there really is something that draws tornadoes to Moore, specifically, that might not really matter all that much to the individual risk of the people who live there. Scale matters. And it contributes to the difficulty of figuring out why tornadoes strike some places and not others. To ask “why,” you first have to know “whether.” And whether tornado hot spots happen or not is relative. “Moore is a mystery, and you aren’t going to get an explanation,” Grazulis told me.

If that’s where we have to leave it … well, it wouldn’t be the first time tornadoes have led people on a bit of a wild goose chase. Take the case of Codell, Kansas. On May 20, 1916, Codell was hit by a tornado. It was hit again on May 20, 1917. On May 20, 1918, a third tornado tore through town. Yes, really. You can find the records with the Kansas State Historical Society. Was there something special about Codell? Maybe. And then again, maybe not.

“What I really would like to know is what it was like on May 20, 1919,” Brooks said. “That’s the story I want. But we don’t really know much. I guess the 1918 tornado just sort of ended the town.”

And after that, Codell, or what was left of it, was never hit by a tornado again.

CORRECTION (May 26, 12:28 p.m.): An earlier version of this story misspelled the name of a town in Oklahoma. It is Pauls Valley, not Paul’s Valley.

CORRECTION (June 1, 6:09 p.m.): A previous version of a map caption in this article misidentified the federal office where Chris Broyles and Casey Crosbie work. It is the Storm Prediction Center, not the National Severe Storms Laboratory. Both are divisions of the National Oceanic and Atmospheric Administration.


  1. EF stands for Enhanced Fujita, the scale by which tornadoes are measured. First developed in the 1970s by Tetsuya “Ted” Fujita, it goes from EF0 to EF5, and ranks tornadoes by the level of destruction they cause. People often talk about wind speeds in the context of the Fujita Scale, but those are almost always estimates back-calculated from the damage observations.

  2. Why since 1994? That’s the year NOAA began to have Doppler radar data covering almost the entirety of the United States. From this point on, scientists know about nearly every tornado that touches down, even if no humans are around to observe it in real life.

  3. Since I interviewed him last fall, Carbin has left the Storm Prediction Center and taken a new job at the National Centers for Environmental Prediction’s Weather Prediction Center in Maryland.

  4. For instance, they all worked together on a paper showing that there’s no evidence that the number of tornadoes that touch down in the United States is increasing with climate change. There is evidence that the prime seasons for tornadoes — which differ from state to state — have changed and become more variable. There’s also evidence that tornadoes tend to cluster more than they did in the 1970s: fewer days with any tornadoes, but more days with lots of tornadoes. It’s unlikely, though, that these trends would account for what Moore has experienced.

  5. This comes from an interview I did with Brooks in November of 2015. An EF4 tornado hit the rural county around Pauls Valley on May 9 of this year.

Maggie Koerth was a senior reporter for FiveThirtyEight.