Air Force meteorologist Captain Robert Miller arrived for his forecasting shift at the weather station on Tinker Air Force Base, a sprawling installation roughly five miles east of downtown Oklahoma City. After analyzing the latest weather maps and charts received via fax from Washington, D.C., Miller predicted that aside from occasional gusts of wind, the base was in for a “dull night.” It would soon be clear just how far off the mark that prediction was.
Around 9 p.m., weather stations to the west of Tinker began reporting lightning. A half hour later, even Miller’s crotchety AN-PQ-13 radar—which had been stripped from a World War II–era B-29 and repurposed as a meteorological instrument—was picking up thunderstorms that, as he later recalled, “looked vicious and were moving very fast.”
The airfield had been used as an aircraft depot since WWII, and dozens of airplanes were parked around the jetways and outside the hangars. High winds could toss them around like toys. Scrambling to protect the planes, the 28-year-old Captain Miller pushed his hands through his high and tight, and ordered his backup forecaster to issue a thunderstorm warning to base personnel despite knowing it was likely too late to secure the aircraft.
Just moments later, a report from Will Rogers Airport, a few miles to the west, stopped the men in their tracks. “TORNADO SOUTH ON GROUND MOVING NE!” it read.
By 10 p.m., a tornado, illuminated by constant pulses of lightning, tore across Tinker Air Force Base as Miller and his colleague “crouched in near panic.” Over the roar of the storm, they heard the toppling of vehicles, the snapping of airplane wings, and the crashing of glass as windows in the operations building surrendered to the tornado’s pressure. Debris fell from the sky like rain.
maj ernest j fawbush and capt robert c miller, weathermen with tinker's severe weather center, were the first to predict a tornado on march 25, 1948 photo courtesy of the oklahoma city air logistics center office of history
72nd Air Base Wing Public Affairs
Miller, left, and Fawbush, right, revolutionized the field of meteorology.
The March 20th tornado would go down as the most destructive storm in Oklahoma to date, leaving behind about $10 million in damage ($132 million in today’s money) on base, including the obliteration of 17 C-54 transports, 15 P-47 fighters, and two B-29 bombers. Eight people were injured.
The damage—as well as the vulnerability the storm had exposed—created enough consternation among Air Force brass that five generals flew in from D.C. the very next day, hoping to get to the bottom of the failed forecast. Miller and his supervisor, Major Ernest J. Fawbush, explained to this panel of generals that even the best forecasts could not predict when a tornado would strike—that they were an invisible threat until they dropped from the sky. Unwilling to wait around for the next unpredictable tornado strike, Tinker’s base commander, Major General Fred Borum, instructed Fawbush and Miller to find a better solution.
The order was highly unusual. At the time, the United States Weather Bureau still banned the use of the word “tornado” in public-facing forecasts in an effort to quell public anxiety about the storms. The misguided policy had effectively suppressed all tornado-related research since 1890. Even by the middle of the 20th century, most Americans living in tornado-prone areas relied on their own senses or folk wisdom (the restlessness of farm animals, a greenish-yellow sky) to know when to run for cover.
But Miller and Fawbush were soldiers as well as meteorologists, and they didn’t defy orders. They quickly got to work unraveling a mystery that had confounded scientists for centuries. What they found would amount to a revolution in weather forecasting, forever changing how meteorologists predict tornadoes.
Just four days later, the sky would unleash its fury again.
The unique geography of the central United States—where warm, moist air from the Gulf of Mexico meets cool, dry air from Canada—means that America sees more of these violent storms than any other country in the world.
Indigenous tribes native to or forcibly removed to the Great Plains were no strangers to tornadoes. The Cheyenne, for instance, called them ma’xehevovetaso, or big whirlwind. In illustrated calendars, the Kiowa depicted a tornado as the long, violent tail of a horse.
Tornadoes struck the East Coast, too, but with less frequency. In a diary entry dated July 5, 1643, Massachusetts Governor John Winthrop wrote of a wind that “lifted up” a meetinghouse full of people. But because of their rarity in the Northeast, tornadoes didn’t merit much scientific curiosity until the mid-1800s, when settlers charged into the Midwest, where funnel clouds stalked the horizon each spring. Even then, tornadoes were thought to be unforecastable, random bursts of turbulence.
One of the first Americans to challenge this notion was John Park Finley, who worked for the Signal Corps, a branch of the U.S. Army that was tasked with creating the country’s first weather service in 1870. After receiving a degree in science from the University of Michigan, Finley, a broad man with push-broom eyebrows, arrived at the Signal Service’s school in Arlington, Virginia, in 1877. (In 1863, the Signal Service became the Signal Corps.) There, he took weather observations and telegraphed a report to the Service’s central weather office in Washington, D.C. Finley was just one of hundreds of servicemen gathering meteorological observations, which were then amalgamated into some of the United States’ earliest forecasts.
By 1878, the Corps was receiving weather reports from 284 stations and distributing 24-hour forecasts to rural post offices across the country. The reports contained the most detailed information available—temperature, pressure, humidity, and wind speed—but were rarely accurate. They mainly relied on statistical prediction; in other words, forecasting the future weather by using past data. If certain atmospheric conditions today yielded a thunderstorm, the thinking went, then similar conditions tomorrow would portend another. But without the ability to characterize the atmosphere, early statistical forecasts were little more than educated guesses.
Two years after he had joined the Corps, Finley was sent to the Great Plains to survey the damage from a recent tornado outbreak. There, the young scientist worked like a crime scene detective, interviewing eyewitnesses, analyzing tree-fall patterns to see which way the wind blew, and inspecting ravaged buildings to establish the sequence of events before, during, and after the tornado. From these observations, Finley was able to establish the general atmospheric conditions that often led to a tornado; namely, that when warm and moist air from the south meets cool and dry air from the north, “the marked contrasts of temperature and moisture invariably foretell an atmospheric disturbance of unusual violence,” he reported in a Signal Service publication.
It was a start.
we had intercepted a tornado warned storm near waynoka before we headed south towards enid this remarkable supercell was heading towards enid, oklahoma when we caught up with it this storm gave some amazing structure with lightning bolts coming straight out of this rotating mesocyclone
john finney photography/Getty Images
Tornadoes are moving into regions that haven’t traditionally seen the storms, leaving communities unprepared.
Finley would spend the next five years traveling through the Plains states, enlisting a network of “tornado reporters” who gathered observations during the violent storms. He would pore over hundreds of accounts of tornadic events and reproduce countless weather charts to eventually create a list of six rules for forecasting tornadoes, which he published in Science magazine in 1884—and updated in simpler terms for the American Meteorological Journal four years later. Tornadoes need “high temperature gradients,” he correctly surmised, as well as the “presence of a well-defined low-pressure area.”
Finley began issuing experimental tornado predictions in the spring of 1884, dividing the eastern two-thirds of the country into districts and advising whether conditions were favorable or not favorable for a tornado to form in each of them. But many of his colleagues questioned the accuracy of these reports, and some critics suggested Finley was too lax in his identification of tornadoes, mistaking derechos, or straight-line winds, for twisters.
Still, Finley advocated that his forecasts should be used to issue tornado warnings in official releases from the Signal Corps. In 1885, the chief signal officer began allowing a special warning when “violent storms,” were likely, but explicitly forbade use of the word “tornado,” fearing it would create pandemonium over forecasts that were still too imprecise to be useful. “The concern was that the forecasts were not specific enough to be of value, and the use of the word tornado tended to cause panic,” says Russell Schneider, director of the National Weather Service’s Storm Prediction Center in Oklahoma. “They might cause people to be overly nervous when the risk wasn’t that concentrated near them.”
Over 10,000 Americans were killed by tornadoes during the Weather Bureau’s long ban on the word...in 1925 alone, 794 people died when the country experienced the deadliest tornado outbreak in its history.
It was a blow for Finley, but things would soon get much worse for the scientist. During the latter half of the 1880s, Congress debated whether the Signal Corps should continue handling weather service, ultimately nesting the agency within the War Department, which showed little interest in Finley’s work.
The end of Finley’s tornado-forecasting project was spelled out in the 1887 Report of the Chief Signal Officer, which stated: “it is believed that the harm done by such a prediction would eventually be greater than that which results from the tornado itself.”
After a decade of progress, Finley’s belief that meteorologists could and should predict tornadoes was snuffed out.
What Is a Tornado?
diagram showing how a tornado forms
Photo illustration by Alyse Markel using Getty Images
Tornadoes are swirling funnel clouds that often form within large thunderstorms when unstable pockets of cool, dry air (blue arrows) and warm, moist air (red arrows) collide. These mighty twisters sweep across the landscape, sometimes for miles, causing catastrophic devastation along their path.
1 / As large-scale supercell thunderstorms gain energy, columns of warm, moist air, known as updrafts, are drawn upward from Earth’s surface into the storm.
2 / Strong winds that blow in different directions and speeds at different altitudes within the storm, known as wind shear (white arrows), contribute to the formation of tornadoes within supercell thunderstorms. Wind shear forms a rotating pocket of warm horizontal air that is turned vertical when it comes into contact with a storm’s updraft. This forms a mesocyclone.
3 / As that warm updraft rises, it smashes into sinking cold air. This temperature difference is thought to cause the funnel to narrow, spin faster, and draw even more air from the ground.
4 / A tornado will typically weaken and dissipate when its surface supply of warm, moist air is used up.
Miller and the 33-year-old Fawbush were young but experienced meteorologists who had both served in WWII. Fawbush was an expert on the Arctic climate of Alaska while the California-born Miller had studied the tropical atmosphere of the South Pacific. The pair surely knew that the project of tornado prediction was fraught with political and scientific difficulties.
Still, the duo obeyed their orders, working around the clock to identify patterns in the atmospheric conditions that had led to the March 20th tornado as well as an even more destructive twister—the infamous Woodward, Oklahoma, tornado of 1947—an F5 tornado that killed 116 people in the Sooner State.
They studied reams of data and as many accounts of the storms as they could get their hands on. Like Finley before them, they quickly found that the presence of certain weather conditions did seem to foretell tornadic activity.
“We listed several weather parameters considered sufficient to result in significant tornadic outbreaks when all were present in a geographic area at the same time,” Miller later wrote in a memoir titled The Unfriendly Sky.
Among those parameters, according to Howard Bluestein, a professor of meteorology at the University of Oklahoma, was wind shear, or the way winds change direction and speed with altitude. Fawbush and Miller were probably the first to determine that “vertical wind shear is very important for tornado production,” he says. The Air Force meteorologists also zeroed in on the significance of the potential atmospheric buoyancy, or the tendency of warm air to rise rapidly into the atmosphere.
While earlier scientists may have identified the significance of these conditions first, Fawbush and Miller would have the opportunity to test their findings just days after they put them to paper. “That,” Bluestein says, “was luck.”
In 1890, a few years after John Park Finley was unceremoniously dismissed from his tornado studies, Congress moved the weather service to the civilian-controlled Agriculture Department and renamed it the U.S. Weather Bureau. Five years later, Willis L. Moore was named its head. A former newspaper reporter, Moore believed that the Corps’ tornado reporters had long been misclassifying windstorms as tornadoes, arguing that “in almost all cases of great disaster there is a pronounced tendency to exaggerate the actual facts,” and suggesting that journalists and insurance agents regularly trumped up accounts of violent storms for their own ends. True tornadoes, he said, simply didn’t happen all that often.
Soon, other meteorologists were speaking out against efforts in the previous century to predict the storms. Writing in 1899, former weather service director Cleveland Abbe asserted that the proliferation of daily newspapers and telegraph lines had created the illusion among the public that the occurrence and intensity of tornadoes was increasing. The agency “had no right to issue numerous erroneous alarms,” he said, and that “the stoppage of business and the unnecessary fright would…be worse than the storms themselves.”
Meanwhile, Oklahoma meteorologist J. I. Widmeyer argued that tornado predictions had created chronic anxiety in the state, sending residents “fleeing to caves and cellars whenever thunderstorms appeared,” where “exposure to the dampness…resulted in more deaths than all the tornadoes that had ever occurred.” Former weather service chief Mark Harrington assured readers of his popular 1899 book, About the Weather, that the chance of being injured by a tornado was about one in a million.
seventeen c 54's were victims of the first tornado this one was picked up and slammed on its back photo courtesy of the oklahoma city air logistics center office of history
72nd Air Base Wing Public Affairs
A Douglas C-54 Skymaster was among the many aircraft destroyed on March 20th.
It was no surprise, then, that the 1905 regulations of the National Weather Bureau contained the statement, “forecasts of tornadoes are banned.” The ban was reiterated in subsequent reissuing of the regulations, and during this time, research on tornadoes by the meteorological establishment all but ceased.
The tornadoes, however, did not.
On March 25, 1948, just five days after the initial storm, Miller and Fawbush were back in the weather station on base, studying the morning weather charts, when they came to a troubling conclusion. Based on the pair’s recently transcribed parameters, Miller recalled, “central Oklahoma would be in the primary tornado threat area by late afternoon.”
The weathermen summoned Major General Borum, who arrived and peered into a radar screen with them as the room crackled with a nervous uncertainty. For the last few days, Borum had also been hard at work, designing a new “Tornado Safety Plan” for the base; now he wanted to know if it was time to set it into motion. “Are you planning to issue a tornado forecast for Tinker?” he asked.
What followed, according to Miller, was an “uneasy quiet.” Both Fawbush and Miller knew that a tornado warning was taboo in the weather business—and that the odds of another tornado hitting Tinker just a few days after the first were less than 1 in 20 million. When Borum suggested a forecast of heavy thunderstorms, the men were relieved. “Far better we should take such odds rather than actually issue a tornado forecast and be laughed out of Uncle Sam’s Air Force,” Miller wrote.
The storms, however, continued to grow in intensity. Alarming reports poured in from stations to the base’s south and southwest that warned first of growing cumulus clouds, then isolated thunderstorms, and finally a burgeoning squall line. Soon, that squall line appeared on the meteorologists’ radar. Worse, it was headed straight for base. By the middle of the afternoon, Borum was back in the weather station, asking again: “Are you going to issue a tornado forecast for Tinker?”
According to Miller, he and Fawbush tried to talk their way out of making “such a horrendous decision,” highlighting the statistical near impossibility of a second twister and emphasizing that, in fact, “no one has ever issued an operational tornado forecast.”
Borum was unimpressed. The general, an imposing man known for his unconventional management style, gave orders for Fawbush to type up an alert. “You are about to set a precedent,” he said.
Over 10,000 Americans were killed by tornadoes during the Weather Bureau’s decades-long ban on the word.
Every spring, churning black storms barreled down the middle of the country like a bowling ball spinning toward a set of pins. As authorities watched and did little, Plains states residents turned to the Farmers’ Almanac and their own observations to predict when the next tornado would strike. Many families had a dedicated weather observer (usually the father) who would keep watch for omens before leading the way to the storm cellar. But this individualized and unscientific approach was no match for the weather’s capricious wrath.
In 1925 alone, 794 people died when the country experienced the deadliest tornado outbreak in its history. That year, on the afternoon of March 18, a once-serene Midwestern sky turned sinister when a tornado touched down in Ellington, Missouri, shocking unprepared residents before taking off on a three-and-a-half-hour tour of destruction. The twister, which at times carved a path nearly a mile wide and stirred up winds topping 300 mph, would tear across 219 miles of Missouri, Illinois, and Indiana, blitzing whole towns and killing hundreds.
Especially tragic storms like the Tri-State Tornado, as it came to be known, as well as the development of meteorological tools like the piloted weather balloon and radiosondes, would inspire bursts of tornado research during the 1920s. But it was WWII—and the proliferation of valuable military infrastructure around the country—that forced the National Weather Bureau to reconsider its ban.
this newspaper clipping shows tinker's ace forecasters, maj robert c miller and lt col e j fawbush, of the severe weather center, looking over the massive stack of teletype rolls they collected in 1953 to continue their research on destructive storms photo courtesy of the oklahoma city air logistics center office of history
72nd Air Base Wing Public Affairs
Miller and Fawbush continued to study tornado formation for years after their famous forecast.
In 1938, the agency began allowing the word “tornado” to be used in weather reports issued only to emergency personnel. During the war, the Bureau once again established a network of volunteer storm spotters near munitions plants, airfields, training camps, and other military posts to warn of oncoming storms.
But when the war ended, the network disbanded and urgency around the task of predicting severe weather faded. By the time Fawbush and Miller glumly typed up their tornado warning on March 25, 1948, they knew issuing such an alert was more than a scientific gamble—it could be career suicide.
As Tinker base personnel got to work diverting air traffic and securing aircraft and loose objects, Fawbush and Miller sat alone in the weather station, surveying what seemed to them like a hopeless situation.
“I wondered how I would manage as a civilian, perhaps as an elevator operator,” Miller wrote. “It seemed improbable that anyone would employ, as a weather forecaster, an idiot who issued a tornado forecast for a precise location.”
Their forecast had indicated only that the odds were good there would be a tornado within a 20,000- to 30,000-square-mile area. “That’s roughly half the state of Iowa, and here the General is making preparations for a tornado that’s going to hit the base,” Schneider says. “They thought their careers were over.”
A couple hours later, when nearby Will Rogers Airport reported just a smattering of weak thunderstorms, Miller packed up for the night, leaving a grim Fawbush to “go down with the vessel.”
That evening in his recliner at home, as he watched the storm roll in, Miller tried to put the whole episode out of his mind. But a radio in another room roared to life with an urgent news bulletin, warning listeners of a “destructive tornado” near Tinker Field. Sure that it was just a recap of last week’s events, Miller called up the base’s weather station to check in.
The line was dead.
“I felt a strange unbelieving excitement rising,” Miller wrote.
He grabbed his things and raced back to the base. When he arrived, Miller could hardly believe what he found: snapped power lines and truncated telephone poles; emergency crews working to clear debris from the streets and restore power; a “jubilant Major Fawbush.”
Incredibly, another tornado had struck Tinker Air Force Base. Thanks to Fawbush and Miller’s warning, this storm caused far less damage, and no one was injured. Remarkably, their forecast had been accurate; the men were instant heroes.
Two years after Miller and Fawbush made their famed forecast, the Weather Bureau effectively ended its moratorium on the word “tornado.”
In 1951, following years of additional research and successful predictions, Fawbush and Miller were placed at the helm of the Severe Weather Warning Center, a new unit responsible for predicting dangerous conditions at all Air Force sites in the Continental U.S. That year, its 156 tornado warnings were accurate 65 percent of the time. The National Weather Bureau soon adopted the military’s methods in the name of public safety.
The following year, the agency established its own severe weather unit, which would eventually become the Storm Prediction Center in Norman, Oklahoma.
“Essentially, we’re the national tornado forecasters,” says Russell Schneider, who now heads up the center. “But all tornado and severe thunderstorm warnings that we issue trace their heritage back to those initial forecasts by Fawbush and Miller in 1948.”
destroyed airplanes at tinker air force base from a tornado that struck it back in 1948
Oklahoma Historical Society
The March 20th twister, the costliest storm to hit Oklahoma at the time, caused the equivalent of $132 million in damages.
Today, predicting tornadoes is even more reliable thanks to the technological advances of the last 75 years, says Howard Bluestein, the University of Oklahoma professor. Scientists today still use a combination of data from satellites, radar, weather balloons, and buoys to make their forecasts, but they’ve got a little bit of extra help from supercomputers that can run billions of equations reliably and with breakneck speed.
“We now have computer models that can do a reasonably good job at forecasting atmospheric conditions, predicting the same parameters that Fawbush and Miller were looking for,” Bluestein explains. “But we still can’t predict the exact location a tornado will form. The best we can do is say there is a certain likelihood over a certain area.”
The Center accomplishes this through probabilistic forecasting, a method that provides a range of possible outcomes—determined through multiple simulations on the weather model using slightly different data each time—and the probability that each outcome will occur.
Forecasting a tornado, Bluestein explains, is “a combination of science and art: using what we know about the science of storms, plus what the models tell us, plus what we’re seeing with our eyes.” For example, tornado watches are issued when conditions are right for a tornado, but warnings are still reserved for actual twister sightings.
Schneider says that for the forecaster, tornadoes still pose a professional and personal challenge, not unlike the ones Fawbush and Miller faced: “Tornadoes come together very quickly—in a matter of minutes, on some occasions—and you’re forecasting something that is really consequential.”
More severe storms, which have the potential to cause a significant amount of damage and imperil lives, can weigh on the meteorologists charged with forecasting them, he explains. “These are potentially very humbling forecasts.”
At the Storm Prediction Center, where staff work beneath a composite map signed by Miller, the Air Force meteorologists and their landmark forecast loom large.
“We’re still living in their legacy,” Schneider says. “We’re still working very hard to add to it.”
Why the U.S. Is a Tornado Hotbed
map illustrating air masses and tornado alley in the united states
Photo illustration by Alyse Markel using Getty Images
The United States, which sees an average of 1,200 twisters each year, is uniquely suited to form destructive tornadoes. Cool, dry air from Canada, propelled by the jet stream, surges over the Rocky Mountains toward regions of low atmospheric pressure in the Great Plains, colliding with pockets of warm, moist air that travel up from the Gulf Coast and warm, dry air from the Southwest. The atmosphere becomes unstable, forming powerful thunderstorms and, often, tornadoes.
Recent research suggests that climate change may be pushing increasingly powerful storms from Tornado Alley—which spans from Texas to South Dakota—southeast and into ill-prepared communities.
Headshot of Ashley Stimpson
Ashley Stimpson is a freelance journalist who writes most often about science, conservation, and the outdoors. Her work has appeared in the Guardian, WIRED, Nat Geo, Atlas Obscura, and elsewhere. She lives in Columbia, Maryland, with her partner, their greyhound, and a very bad cat.