The cabin crew had just served breakfast when Dzafran Azmir felt the first tremor. He and the other two hundred and ten passengers on Singapore Airlines Flight SQ321 had been in the air for more than ten hours. Their flight had taken off the night before from the United Kingdom, where Azmir was studying audio engineering at the University of Plymouth, and had flown across Central Europe, the Black Sea, Turkmenistan, and Pakistan. They were thirty-seven thousand feet above the Irrawaddy River, in Myanmar—three hours from their scheduled landing in Singapore—when the turbulence started. For a moment, the plane quivered around them like a greyhound straining on a leash. Then it lifted its nose and leaped forward on an updraft. Eleven seconds later—at 7:49:32 A.M. on May 21, 2024, according to the flight’s data recorder—the pilots switched on the “Fasten Seat Belt” sign and told the flight attendants to secure the cabin. They were in for some rough weather.
Late spring is the start of the wet monsoon season in Myanmar, when heavy rains and squalling winds sweep in from the Bay of Bengal. If the pilots had been equipped with better radar or forecasting software, they might have known to avoid the towering storm clouds welling up beneath them. But, aside from a few white cloud tops, the sky outside the plane was clear and bright. Those first tremors were their only warning.
In 2002, a team of researchers from NASA, the F.A.A., and six commercial airlines ran a series of experiments in Oklahoma City, in a decommissioned Boeing 747. They wanted to see how quickly a commercial jet could be secured in case of turbulence. They recruited a group of volunteer passengers, gave them all fake boarding passes and luggage, and had three of them carry around life-size baby dolls. Some of the volunteers were told to stay in their seats and act as if they were sleeping, reading, or working on a laptop. Others were told to stand in the aisles or sit in the lavatories. A crew of experienced flight attendants, drawn from the airlines participating in the study, made their way up and down the plane, serving fake food.
The research team ran nineteen drills in three days. Some began with a mild warning over the intercom: “Ladies and gentlemen, this is your captain speaking. We will be encountering a line of thunderstorms in about ten minutes.” The rest began more urgently: “All passengers and flight attendants, please be seated immediately.” The more emphatic announcement prompted a quicker reaction. Still, at best, only two-thirds of the occupants were buckled up after seventy seconds. On average, the passengers needed a minute and a half to take their seats; the flight attendants, who had to stow their gear first, needed at least four minutes. Fully a third of the occupants were still out of their seats after seventy seconds.
To the people on Flight SQ321, that would have seemed an eternity. Eight seconds after the captain’s warning, the plane plummeted. Within five seconds, it had dropped a hundred and seventy-eight feet—about the height of a nineteen-story building. There was no time to react, Azmir later told reporters. “Whoever wasn’t buckled down, they were just launched into the air within the cabin,” he said. Azmir had a window seat near the back. When a plane hits turbulence, it tends to seesaw from front to back, so the first and last rows rise and fall the most. “I saw people from across the aisle just going completely horizontal, hitting the ceiling,” Azmir said. “People getting massive gashes in the head.” Some passengers were vaulted up so violently that they dented the luggage bins, or thrust their heads through the panels where the oxygen masks were stored. Those who were standing were sent somersaulting down the aisles; those sitting in the lavatories smashed into the ceiling. It was “sheer terror,” one passenger later said. Then, just as abruptly, the plane lurched up, slamming everyone back to the ground. In just four seconds, the gravitational force on their bodies changed from negative 1.5 g’s to positive 1.5 g’s, Singapore’s Ministry of Transport later noted. It was as if their bodies went from being helium balloons to being sandbags.
“I arrived back in the airport and I couldn’t stop vomiting,” another passenger said, after the plane had made an emergency landing in Bangkok. “I couldn’t walk.” A hundred and four passengers had to be treated for injuries. More than forty of them were kept for longer stays at the hospital; six had skull and brain injuries, including a two-year-old boy. Of the seventeen passengers who needed surgery, nine had spinal injuries, including an Australian woman named Kerry Jordan, who was left paralyzed. A year later, she still couldn’t brush her teeth or use a phone.
Turbulence is the ghost in the attic of air travel—the bump and shake and rattling groan that we do our best to ignore, though it sounds like it wants to kill us. Most of the time, it hovers over mountains and in storm clouds, easy enough to avoid. Pilots can see bad weather lurking in the distance hours before takeoff, glowing like a wraith on their digital maps. If it moves, the plane’s radar can still spot it eighty miles ahead or more. But the updraft that struck Flight SQ321 was of a more sinister sort. Although it came from the storm clouds below, there was seemingly no rain in it for radar beams to reflect against. It was like an invisible speed bump in the sky.
In 1966, a Boeing 707 operated by the British Overseas Airways Corporation took off from Tokyo en route to Hong Kong. It was a sunny, cloudless afternoon, but as the plane approached Mt. Fuji a violent wind struck it from the northwest. The gust tore the vertical fin from the tail and hurled it into the left horizontal stabilizer, which broke off in turn. As the plane twisted upward, the air pressure wrenched off another tail fin. All four engines were ripped from the wings, sending the plane spinning toward the mountain’s flank. The fuel tanks ruptured, and the entire tail section fell off, along with the right wing. By the time the plane crashed, in a forest at thirty-five hundred feet, its fuselage had broken in two and a trail of debris ten miles long stretched behind it.
The Mt. Fuji crash was one of a series of plane accidents in Japan that year. One commercial jet careened into a seawall while landing in heavy fog; another plunged into Tokyo Bay for unknown reasons; yet another, into Japan’s Seto Inland Sea, also for unknown reasons. It was one of the deadliest years in commercial-aviation history—three hundred and seventy-one passengers and crew were killed in those incidents alone—and it changed the way that airplanes were built.
When new employees come to work at the Boeing production facility in Everett, Washington, one of their first stops is often an exhibition at the company’s Safety Experience Center. It opens on a sombre note: a memorial for famous air disasters, including the successive crashes of two 737 MAXs, in 2018 and 2019, in the Java Sea and Ethiopia. Then, gradually, the tone grows more hopeful. At Boeing, as throughout the aviation industry, disasters led to innovations. Oxygen masks and electronic anti-skid brakes were introduced in the nineteen-sixties, along with bird cannons at airports, to shoo off Canada geese and fellow-fliers. Overhead bins got latched doors that same decade, to keep luggage from toppling onto passengers’ heads. Satellite communication came along in the seventies; automated flight-management systems, capable of plotting a plane’s course, speed, and altitude, in the eighties. Radar systems got more accurate; planes grew stronger, sleeker, and more flexible. Pilots got better at skirting turbulence—or, if they couldn’t, at slowing down and “riding the bumps.”
“Once I finally found him, I didn’t have the heart to harpoon him.”
Cartoon by Mick Stevens
“When people get hurt around our products, we need to take every ounce of education from it that we can,” Jacob Zeiger, a senior air-safety investigator at Boeing, told me, when I visited the Boeing facility this past December. “These technical findings are sacred.” Within hours of any major incident involving a Boeing plane, Zeiger and his team are notified, and they may spend a week to ten days studying the damage and interviewing the crew, piecing together what went wrong. In 2008, for instance, the engines on a 777 stopped responding, and it crash-landed short of a runway in London, shearing off its landing gear. Afterward, a team of investigators re-created the plane’s fuel system in a Boeing lab. A small heat-exchange unit, they found, had accumulated ice during the flight. Every 777 has since been retrofitted with a new version of the unit. Thanks to decades of such refinements, today’s jets may be the world’s most reliable machines. Flying in them is less likely to kill you than walking on staircases.
It’s the sky that’s grown more unreliable. Fierce storms and erratic winds are increasingly common with climate change. But the rise in clear-air turbulence, often far from storms and undetectable by radar, is especially alarming. Since 1979, clear-air turbulence has increased by as much as fifty-five per cent over the North Atlantic and forty-one per cent over the United States. If temperatures continue to rise unabated, it could more than double by the middle of the century. Death by turbulence is still vanishingly rare, but Flight SQ321 did have one fatality. Geoffrey Kitchen, a retired insurance salesman from Bristol, England, on holiday with his wife of fifty years, died before the plane landed. Its sudden plunge had come as such a shock, it seems, that it gave him a heart attack.
I’m not an especially fearful flier, but I dread turbulence all the same. I avoid seats in the back of a plane, check radar maps before a flight, and keep my seat belt buckled even when the light is off. I remember a conversation I once had with the forensic anthropologist Clyde Snow, who spent years investigating plane crashes for the F.A.A. In his experience, he said, when a crash did have survivors, a disproportionate number of them were men: they were the first to shove and claw their way to the exits. As he put it in 1970, in a study that included two plane crashes in which the passengers had to flee from burning cabins, “It appears that younger males were definitely favored . . . where speed, strength, and agility would be expected to play a dominant role.” In those two crashes, even the old men survived at a higher rate than the adult women and children.
That image has always stuck with me, both as a sobering comment on my sex and as a grisly worst-case scenario. So it was strange, this fall, to be looking for a bumpy ride. Some sixteen million flights crisscross the United States each year. Of those, roughly one in every two hundred and fifty gets hit by moderate-or-greater turbulence—strong enough to make passengers feel “a definite strain against their seat belts,” as the National Weather Service describes it. One in every three thousand flights encounters severe turbulence: “The airplane may momentarily be out of control. Occupants of the airplane will be forced violently against their seat belts.” By that scale, the worst turbulence I’ve felt could only qualify as light: “slight erratic changes in altitude.” To definitely experience more, I would have to fly in a very small aircraft.
The most turbulent flight routes in North America are over Colorado, where the prevailing winds from the west barrel into the high peaks of the Rockies and tumble onto the High Plains below. One morning this fall, on a stubbly brown field in Boulder, a glider pilot named Dan Swenson stared up at the sky and shook his head. A vast, lens-shaped cloud hung above us like an alien mother ship. It stretched from the foothills of the Front Range, in the west, to the Laramie Mountains, in the north, its pale upper reaches darkening to a gunmetal gray along the bottom. “So, what’s with this?” he said. He glanced over at Jordon Griffler, the scraggly young pilot who would tow Swenson’s glider into the sky with his single-prop plane. Griffler shrugged and took a bite of a bagel. “You can ride that all the way to Wyoming,” he said. Swenson shook his head again: “Holy cow!”
Swenson is seventy-four, with a silver mustache and a well-upholstered body, and has been piloting gliders since he was thirty. Today’s flight would be his eight-thousand-eight-hundred-and-sixty-third, but that hardly made it predictable. “Forecasts are forecasts,” he said. “Most of the time, they don’t even get them right the day after.” Our previous day’s flight had been cancelled because of dangerous winds. Did today’s conditions worry him? “The answer to that is no,” he said. “As a soaring pilot, this is the kind of stuff that can get you really excited. Just think about water. If we’re out on the ocean in a sailboat, and there are these little waves and everything’s nice, and all of a sudden there’s this great big wave coming through, the experienced captain will go, ‘Rogue wave!’ It’s just part of the deal. ‘That scared the shit out of me!’ Part of the deal.”
This was less than reassuring. As was the sight of Swenson’s glider—a dinged-up, silver-and-orange craft that looked like some kids had built it in their junior-high rocketry club. Nevertheless, we were soon in the air. Griffler’s plane towed us high above the foothills on a thin gray rope. There was a bang, the rope fell away, and we were floating free at ten thousand feet.
For a moment, I forgot myself in the sheer, giddy wonder of it, lofting like a blown leaf above forests, lakes, and an old railway winding into the mountains below. Then the first wave of turbulence hit. It felt like skidding on black ice—the sudden, sickening drop, the bewildering weightlessness—only I had no brake, no steering wheel, no hope of bailing out. The closest thing to grab was a large copper-colored knob on the dashboard, labelled “RELEASE.” But I’d been told not to do that under any circumstance, unless I wanted a very short flight. (I later learned that it was meant for the towing cable.) “We’re in that rotary stuff now!” Swenson shouted from the seat behind mine. When strong winds blow over a mountain range, they’re bent into huge, oscillating waves up high, while the air below spins down the slopes in ragged gusts, known as rotor clouds. It felt like driving over a row of logs. “I once had a passenger who threw up five times before we even released the rope!” Swenson said.
Swenson’s glider was an exquisite instrument for detecting turbulence. A jumbo jet can weigh upward of half a million pounds and fly more than five hundred miles an hour. It charges through the air like an ocean liner, barely registering most winds. This glider wasn’t much heavier than a Harley and it was moving about as fast—fifty to seventy miles an hour. It felt every bump. Small aircraft account for many of the injuries caused by turbulence and, essentially, all of the deaths—about forty a year. They fly at the mercy of the wind.
“You doing O.K.?” Swenson shouted, after a while. “The more you’re talking, the better I know you’re doing.” I tried to say I was fine, but it came out funny—a high, choked-off sound. I waited a moment, then covered for it by asking what he did for a living before he retired. He laughed. “I worked for Northwestern Mutual in Boulder,” he said. “I sold life insurance.”
Far below us, at the base of the Front Range, I could see a group of tall red towers rising from a meadow of yellow bunchgrass. They looked ancient from this distance—like the cliff-dwellings of Mesa Verde—but also vaguely futuristic. Woody Allen, in his sci-fi spoof, “Sleeper,” used them as the setting for a sinister cloning institute. In fact, they housed the National Center for Atmospheric Research, or NCAR, one of a group of federally funded centers that were created, after the Second World War, to apply the most advanced research to the most urgent practical problems. It was the Los Alamos of turbulence. When I went there later that morning, I half expected the buildings to be dark. The government was shut down, and the National Science Foundation was facing huge budget cuts. But NCAR was still open, for now.
Inside the main building, a large glass globe was perched on a steel stand. It was filled with an iridescent blue liquid that moved in ever more complicated patterns when I spun the globe, like clouds above a rotating planet. It was meant to show the chaotic nature of the atmosphere, but it was a drastic oversimplification. On Earth, air heats and rises along the equator as the planet turns, then cools and sinks when it reaches the poles. Mountains and valleys shape the paths of the wind; volcanoes scorch the air and shade it with their ash; ocean currents absorb heat and then evaporate into the sky, churning the air with their vapors. And, everywhere, warm and cold fronts rub against each other, setting off still more swirling changes. To truly capture that complexity, I would have had to shake the ball like a snow globe.
“Turbulence is one of the great unsolved problems in classical physics,” Larry Cornman, a senior researcher at NCAR, told me, when we spoke in his office. “You have to predict where these things will happen and when, but the equations are inherently nonlinear.” Cornman is sixty-eight, with brown hair, streaked with gray, that hangs below his shoulders. He was dressed in a T-shirt and a tracksuit jacket, and spoke with an offbeat affability—a holdover from Boulder’s hippie days. Before earning degrees in math and physics from the University of California, Santa Cruz, Cornman lived in a Buddhist commune in Northern California for three years. When he moved to Boulder, in 1983, he took a job at NCAR as a part-time computer programmer, and never left. He has since earned eight patents and devised some of the most widely used systems for detecting turbulence.
Cornman arrived in Colorado just as the airlines were confronting a series of mysterious crashes. On June 24, 1975, an Eastern Airlines plane was blown to the ground half a mile short of its runway in New York. It collided with some approach light towers and burst into flames, scattering its pieces across Rockaway Boulevard. Seven years later, in New Orleans, a Pan Am flight was forced down by the wind just after takeoff. It plowed into an adjacent neighborhood, killing eight people on the ground and all its passengers. In the U.S. alone, similar wind-shear accidents killed more than four hundred people between 1973 and 1985.
“People were dying and we didn’t know why,” Cornman told me. “We didn’t understand the physics of why the planes were crashing.” The deadly gusts were thought to be blowing in from the ocean or from thunderstorms outside the airports. But the danger turned out to be right above them. In the late seventies, researchers at NCAR and the University of Chicago discovered that the crashes were caused by microbursts—sudden, violent downdrafts. In a microburst, a storm cloud dumps cool air and rain straight down, like water from a broken awning. The air spreads horizontally after it lands, so the pilot thinks he’s flying into a headwind at first. He lifts the plane’s nose slightly and decreases the engines’ thrust. Then the downdraft hits, followed by a vicious tailwind, sending the aircraft to the ground.
The NCAR team spent the next ten years working on the problem with researchers at airlines, universities, the F.A.A., NASA, and NOAA—the National Oceanic and Atmospheric Administration. “It was a national imperative,” Cornman said. Luckily, the beginnings of a solution were already in place. The team at NCAR had used sophisticated new Doppler radar systems to detect microbursts. When those were added to the wind detectors already installed at many airports, and the two systems were integrated with software that Cornman developed, microbursts could be detected as they were happening. “A problem where hundreds of people were dying suddenly stopped,” Cornman said. The last time a commercial flight was downed by a microburst in the U.S. was in 1994.
Turbulence is rarely that simple. It’s too scattered, too mercurial, too easily triggered by weather patterns that trigger other patterns in an endless cascade. “It’s not just one thing that’s going on,” Bob Sharman, an atmospheric scientist at NCAR, told me. “It’s not just atmospheric convection. It’s not just wind flowing over mountains. It’s everything going on all the time and interacting.” Sharman is one of the country’s preëminent authorities on turbulence prediction. The computer models that he has built can predict where rough air is most likely to arise. “The problem is,” he said, “when we go to meetings with the airline industry and suggest a probabilistic approach, a pilot will stand up and say, ‘No! I want you to tell me if there will be turbulence at this place, at this time.’ ” Sharman threw up his hands. “Nobody knows that. I understand that, in theory, you would want that. But, in practice, that is just not possible.”
Early in October, I took a red-eye flight from New York to Santiago, Chile. I’d been reading a website called Turbli, run by a turbulence-obsessed engineer in Stockholm named Ignacio Gallego-Marcos, who has a Ph.D. in fluid dynamics. Gallego-Marcos had gone through a year’s worth of forecasts from NOAA and the Met Office—the U.K.’s national weather service—and combined them with flight-tracking data from around the globe. In 2025, he concluded, three of the five bumpiest flight routes in the world flew into Santiago.
Chile is like a giant laboratory for unstable winds. When the jet stream crosses the Pacific, it slams into the Andes at an almost perfect right angle. On the Chilean side, the climate is sunny, stable, and dry. On the Argentinean side, fierce winds rise and fall above the peaks in mountain waves, tossing up rows of thin white clouds like foamy crests. In winter, the air below the peaks can plunge into the Uco Valley, in a stifling zonda wind. The sudden compression generates terrific heat, sending temperatures above a hundred degrees in Mendoza, Argentina. “It is very psychologically difficult,” Roberto Rondanelli, an atmospheric scientist at the University of Chile, told me, when we met in Santiago. “People get crazy in Mendoza.”
Things get even worse to the northeast. As the dry valleys of Argentina give way to subtropical rain forest, the humidity rises in tremendous thunderheads and rushes back down in torrential rains as it cools. The convection drives huge, lashing winds through the atmosphere, with violent updrafts and downdrafts. “We don’t get this in Chile,” Rondanelli said. “As a meteorologist, I really resent that. We are the boring part. In Argentina, they have all kinds of crazy weather. Hail that is almost baseball-size! The biggest storms in the world—a thousand kilometres wide! I would love to cross one of those storms.” He sighed. “I have three kids and two stepkids, so I may not do this. But I would love to feel how it is to be inside them—without dying, hopefully.”
Cartoon by Carolita Johnson
A modern jet has an array of sensors to help it contend with bad weather. Magnetometers track the plane’s direction; gyroscopes help calculate its pitch and roll; accelerometers detect its changes in speed; Doppler radar measures the distance to the storm and how quickly it’s moving. In the early nineties, Larry Cornman developed a program that could sift through some of that sensor data to measure the air’s turbulence in real time. Atmospheric scientists call this the “eddy dissipation rate,” or E.D.R. It’s usually scored between 0 and 1—calm to severe.
The flight from Mendoza to Santiago is the bumpiest in the world by that measure. It has an average E.D.R. of .23. That’s nearly a third higher than the most turbulent routes in North America—from Denver to Jackson Hole and from Albuquerque to Denver—but still far from severe. On a Boeing 737, Cornman told me, an E.D.R. of .23 would register as moderate turbulence—“uncomfortable, especially for long periods, but people won’t hit the ceiling.” Then again, averages can be deceptive. A roller coaster might average only fifteen miles an hour if you include the slow climb up the hill. But that first drop is all you remember.
“I can assure you that the scariest flights of my life were crossing the Andes,” a Chilean pilot told me, when I flew from Santiago to Mendoza. He’d been flying over the mountains for more than fifteen years, he said, sometimes in a four-person plane. The high peaks present a Scylla and Charybdis problem in winter, when the jet stream intensifies and storms roll in from the Pacific. Fly high and get tossed around by mountain waves. Fly low and get driven toward the rocks by the zonda wind. “It can throw down a small airplane,” he said.
Sitting on the plane in Santiago listening to the safety instructions, I imagined how they might sound if they were more like the stories and flight-accident reports I’d been reading: “Should a passenger hit the ceiling twice, do a flip in the air, and land on his stomach . . . Should a service cart topple onto a flight attendant and fracture her ankles . . . Should people start screaming and calling to Jesus . . .” In the U.S., turbulence causes more than a third of all accidents on commercial flights, the National Transportation Safety Board found. Those accidents tend to hurt people in predictable ways. Passengers usually get injured near the back of the plane, for instance, often as they are walking to the lavatory, sitting inside it, or waiting in line. But the total number of injuries is hard to determine. One major airline estimated that it receives two hundred turbulence-related injury claims a year, but the N.T.S.B. doesn’t keep track of “minor injuries”—including those which require a hospital stay of less than forty-eight hours. “These things happen all the time, but because they don’t cause death or serious injury they’re swept under the rug,” an N.T.S.B. senior meteorologist and investigator told me. “We only have about a hundred aviation investigators for fourteen hundred accidents a year.”
If there’s one takeaway from the N.T.S.B. statistics, it’s this: flight attendants are in a hazardous business. When planes hit turbulence, the pilots and passengers are usually in their seats. The flight attendants are often still on their feet. They get nearly eighty per cent of the serious injuries caused by turbulence, and, for every one of those injuries, a 2001 study found, another seventy flight attendants get minor injuries. Nearly a third of the time, they have no idea the turbulence is coming.
“It was relatively calm and the three of us were working in the aft galley,” a flight attendant recalled in an N.T.S.B. crew statement last March, about a flight over the Philippines. “A young little boy entered the galley asking for sandwich snacks. Suddenly, unexpectedly, clear-air turbulence struck. . . . It all happened so fast. I remember flying up and then dropping down hard, hitting the side of one of the carts and then being thrown backwards, hitting hard flat onto the galley floor. My head and back slammed onto the floor and so did the little boy right beside me. His screams of fright and agony when he landed hard on his back still haunt me.”
My flights across the Andes, as it turned out, were among the calmest, most serenely beautiful of my life. On the return trip to Santiago, the sun was setting, and the high peaks floated by soundlessly below, lit amber and violet like corals beneath a glass-bottomed boat. “There are no words to describe it,” a flight attendant from Santiago told me later. “Where else do they have mountains like these?” Still, she said, on a flight across the Andes a year earlier she was shaken up so badly that she couldn’t fly over them again for a while. “I was scared for three months,” she said. Another flight attendant insisted that she wasn’t afraid of turbulence at all. “I’m super used to it,” she said, then added, “But we should not lose the fear of turbulence. If you get too much used to it, you can make mistakes. You can be, like, ‘No, it’s nothing,’ and then paff! ”
Thirty years ago, atmospheric scientists began to notice a worrisome trend. They knew that turbulence from storms increases in summer, when hotter and more humid air rises to form thunderheads. And they knew that clear-air turbulence increases in winter, when the temperature drops dramatically at the poles but not in the tropics. That temperature difference drives the jet stream, which creates turbulent eddies and wind shear as it rushes through slower bodies of air. But, on top of these seasonal swings, there seemed to be an over-all trend as well: Earth’s atmosphere was getting rougher.
This made intuitive sense. Temperatures had been rising across the globe for nearly a century. The more heat and energy there is in the atmosphere, the more turbulent it ought to be. But the climate tends to frustrate expectations. If temperatures at the poles rise more than temperatures at the tropics, for instance, the difference between them will decrease, and the jet stream could slow down. Nevertheless, on average, turbulence seemed to be rising everywhere. The surprise was how much. Between 1958 and 2001, the weather data suggested, clear-air turbulence increased between forty and ninety per cent over Europe and North America. The British atmospheric scientist Paul Williams found similar increases when he looked at data from satellites, weather balloons, and aircraft from 1979 to 2020. If carbon-dioxide emissions continue apace, Williams estimates, moderate or greater clear-air turbulence could rise by as much as a hundred and seventy per cent on flight routes over the North Atlantic by the middle of the century. Turbulence from storms and other sources could also nearly double, a study co-authored by Bob Sharman found.
How much turbulence can an airplane bear? Every year, the question is asked and answered by a group of Air Force and NOAA pilots and researchers known as the hurricane hunters. The initiative began, unofficially, in 1943, when Lieutenant Colonel Joseph Duckworth flew into the eye of a hurricane near Galveston, Texas. Duckworth made his flight on a dare, but the programs have since taken on a more serious role: to report on hurricanes as they develop and to study their inner mechanics. Last year, Joshua Wadler, a hurricane hunter and a meteorologist at Embry-Riddle Aeronautical University, in Florida, went through the turbulence data from every NOAA hurricane flight since 2004, and two infamous ones from the nineteen-eighties. He measured how much each flight was thrown around along six axes of motion: roll, pitch, yaw, surge, sway, and heave. (The words alone can induce vertigo.) Then he made a list of the bumpiest flights ever recorded.
Wadler’s own worst flight was through Hurricane Ian, in 2022—a trip famous for its furious lurches from side to side. But that flight came in second on his list. The bumpiest of all was through Hurricane Hugo, east of Barbados, in 1989. “It was like riding a roller coaster inside a car wash,” Frank Marks, a NOAA meteorologist who has flown through more than a hundred hurricanes, told me recently. “Within less than a minute and a half, the wind increased to a hundred and seventy miles an hour. And then, just as we’re breaking into the eye, starting to see a little clearing ahead, a little sunlight, someone says, ‘There’s a flame coming out of engine three!’ And we think, Oh, crap.” Their plane, a Lockheed P-3 Orion, nicknamed Kermit, lost power in engine three and fell more than six hundred feet, nearly into the Atlantic. It pulled out of the dive and climbed back into the eye of the storm. Then it circled there for about an hour before an Air Force plane escorted it through a soft spot in the eye wall.
Kermit and its sister plane—another P-3 Orion, nicknamed Miss Piggy—have flown into hundreds of hurricanes since 1976 and never lost a passenger. The Air Force, in that same period, has flown thousands of hurricane flights in its Lockheed WC-130s, also without a single fatality. Both sets of planes were designed for the military: the P-3 is like a flying tank, Marks told me; the WC-130 is more like a flying barn. Commercial jets are built for lighter duty, and they try to steer clear of hurricanes. But how will they fare in a windier world?
“I talk to the airlines, and they realize there is a looming threat,” Paul Williams told me. “But it’s quite distant for them. They see it as decades in the future. The people really taking notice are the designers. The airplanes they’re designing now will still be flying in 2050, so they can’t wait. And here is a bit of a shocking thing: the design and certification standards for air worthiness are based on turbulence measurements from the nineteen-sixties. It’s reasonable to ask, Are those standards still fit for the purpose? Will they still be fit for the purpose in 2050?”
When I visited the Boeing facility in December, the company was going through a rough patch. The two 737 MAXs that crashed in 2018 and 2019, killing all their occupants, were brought down by the same faulty sensor: it kept telling the flight-control software to tilt the plane’s nose down. After the second incident, the F.A.A. grounded the entire 737 MAX fleet for twenty months. Then, in 2024, another of the planes ran into trouble. It was climbing to sixteen thousand feet over Portland, Oregon, when one of its side panels blew off, leaving a gaping hole in the plane’s left flank. The cabin pressure dropped so quickly that a fifteen-year-old boy had his shirt ripped off his back. It was later found that the panel was missing four bolts that should have kept it in place.
“The past five to seven years have been very hard on us personally,” Darrel Larson, a vice-president of operations, told me. Larson, who grew up on a farm in North Dakota, has a rangy frame and a folksy manner. We were standing on the floor of the building where the 777 is constructed—the largest factory in the world by volume. It’s a vast, open hangar, two-thirds of a mile long and a third of a mile wide, subdivided by rows of glass-walled offices. A system of tunnels runs beneath it, and a network of rails above, hung with gantry cranes that can lug an eight-ton engine across the floor. The factory runs around the clock, yet it was eerily quiet. The planes are mostly assembled from modules made at other sites nearby, and pieced together in a strictly orchestrated sequence. Shrink-wrapped airplane lavatories were clustered along the perimeter, next to carts full of bins with numbered parts. It was the world’s most sophisticated Lego set.
Critics have blamed outsourcing for the company’s recent lapses: some twelve hundred suppliers design and manufacture parts for Boeing planes. “There were a number of issues that we had,” Larson said. “But we’re coming out of that. We can’t build airplanes fast enough.” He walked me over to a half-finished 777 that was about to have its wings attached. “I remember my very first fuselage join,” he said. “The mechanics drilled about eight hundred holes, then started inspecting the skin around them. I thought, This is going to take forever!” The wings have to be mated to the body with gaps of five- to ten-thousandths of an inch. Then they’re bolted on with just enough torque to flex and hold fast, even in the most severe turbulence. In Boeing’s stress tests, they can bend almost forty-five degrees and snap back, and withstand years of flexing without structural fatigue. “It’s not a piece of farm equipment,” Larson said. “It’s a life-support system. At thirty-five thousand feet, you can’t pull over.”
When disaster strikes on a flight these days, it’s almost never the way we fear. The wings won’t rip off in a gale. The plane won’t get thrown into a mountain. In the seven decades since the first paying passengers flew on a commercial jet airliner—from London to Johannesburg in 1952—the number of commercial flights has increased exponentially, while the risk of dying on one has grown incredibly small. “It works out to a probability of fatal injury of one in forty-six million flights on U.S. and E.U. airplanes,” Jacob Zeiger, the air-safety investigator at Boeing, told me. When an accident does happen, it’s usually because of human error or a ground collision or some combination of factors, including the simple act of walking around on a bumpy flight.
Airplane designers have done their best to minimize the pain. Since the nineteen-thirties, cabins have evolved from wicker chairs in bare metal tubes to mood-lit cocoons full of curved surfaces. The air has been pressurized so that pilots can fly up to smoother skies. The service carts have been lightened with composites and braced with built-in brakes and stay-closed drawers. Should the plane suddenly drop, the passenger seats can bear a force of sixteen g’s; the jump seats have harnesses to hold down the crew. Should a fire break out, the walls, seat backs, and tray tables are made with self-extinguishing polymers that won’t give off toxic fumes. Floor lights will guide the passengers down darkened aisles, to evacuation slides that inflate and unfurl to the land or water below.
Still, for all the clever innovations that have made cabins safer, the economics of flight have also potentially made them more dangerous. Planes once furnished with plush recliners and wide-open aisles are now packed with more passengers each year. Those in first class lounge in lie-flat seats with built-in air bags in their seat belts, while those in economy sit shoulder to shoulder and knee to seat back, hoping they won’t crack heads in a storm. In 2021, the F.A.A. published a study on emergency evacuations which concluded that current cabin-seating requirements “can accommodate and not impede egress for 99% of the American population.” But last year a review panel from the National Academies of Sciences, Engineering, and Medicine found that the study was flawed. The F.A.A. trials had included only volunteer passengers with no physical limitations, and none of them was over sixty years old.
“It’s a low-margin industry, so it’s always a cost-benefit analysis,” Larry Cornman told me, in Boulder. “But year in and year out, people are getting hurt.” The best way to prevent injuries, Cornman believes, is to avoid turbulence altogether. Air-traffic control will warn pilots of any storms on their weather maps or radar systems. But for clear-air turbulence pilots still rely on reports from other pilots who have already passed through it—an unsatisfactory arrangement for both parties. “A pilot is an analog, qualitative sensor,” Cornman said. “They get bounced around and say, ‘It’s moderate turbulence.’ Well, what does that mean?” A bone-rattling drop on a small plane might feel like a bump or two on a jumbo jet. Or a pilot might be too busy trying to control the plane to comment on the weather at all. “By the time they make the report, it’s happened way back there,” Cornman said. “So the locations aren’t very accurate, either.”
“Up until a minute ago, it was an honor just to be nominated!”
Cartoon by Pia Guerra and Ian Boothby
Turbulence doesn’t have to be a matter of word-of-mouth. Commercial jets have the capacity to measure and transmit it automatically—using the software Cornman developed at NCAR in the early nineties—and have it relayed to other pilots. The software is freely available to airlines, but most of them balk at the cost of sending and processing the data, and sharing it with others. “It’s not much, but it adds up,” Sharman told me. “A lot of them are saying, ‘Wait a second, we’re barely making it. We can’t afford another hundred thousand on transmission!’ While the ones that are willing to pay for it say, ‘Why should I share my data with somebody else who doesn’t?” So far, only around two thousand planes have been equipped with the software—about one in four planes in the American fleet.
A few years ago, Cornman found a way around the problem. In the U.S., commercial flights served by air-traffic control—some twenty-seven thousand a day—are required to transmit their position, altitude, and velocity. By tracking those transmissions and the planes’ motions over time, a new program that Cornman developed at NCAR can create a moment-by-moment snapshot of turbulence as it’s happening. The F.A.A. is planning to test the program this year. Together with NCAR’s earlier software, Sharman’s forecasting models, and data from radar arrays on the ground, this system could start to give pilots the advance warning they need. “That is the future for me,” Cornman said. “All these operations get integrated in a seamless network. The pilots don’t have to talk to air-traffic control and say, ‘Should I go up or down?’ They just get a display with a color-coded flight track on it. And they see that it looks better a few thousand feet up.”
There’s just one hitch: the system still needs guinea pigs. Even the best weather models can’t pinpoint where clear-air turbulence will occur. So the NCAR programs continue to rely on firsthand reports from planes that have already been tossed around. New technologies could change that in coming years. A plane equipped with a lidar sensor—which uses lasers to detect much finer particles than radar can—could pick up on turbulence even in a cloudless sky. But lidar systems are still too bulky and expensive to fit into a plane’s nose cone. And the government and the airline industry have been slow to invest in improving them. For now, the best hope for a flight heading into turbulence might be to program the plane itself to ride the bumps.
“This is sort of choose your own adventure,” Ryan Pettit, a technical fellow with Boeing’s flight-controls division, told me. We were sitting in the pilot seats of a multipurpose simulator cab. From the inside, it looked like the flight deck of a 777, complete with banks of gauges, switches, and digital screens, and a view of Mt. Rainier through the windshield. From the outside, it looked like a giant, one-eyed robot: a cabin perched on three mechanical legs more than two stories tall. In months of chasing turbulence, the closest I’d come to it on a commercial flight was in Texas, when a thunderstorm struck my plane just as it was preparing to land in Austin. “Folks, it looks like it’ll be smooth sailing for the first hour and forty-five minutes,” the pilot had warned, as we left New York. “Then it’s all downhill from there.” But this simulator was nothing if not reliable. It was turbulence on demand.
Pettit and his colleague Paul Strefling, sitting in the pilot’s seat between us, are engineers in the business of ride quality. Their job is to program the movable parts on an airplane’s tail and wings—the rudder, elevators, and nearly two dozen ailerons, flaperons, and spoilers—to smooth out its flight automatically when turbulence hits. To get data for the simulator, their team takes full-size Boeing jets on research flights over the Rocky Mountains. They hunt for rough air, then loop through it again and again, like race-car drivers on a test track. They record every flutter and quake using the plane’s sensors, then download them to the simulator’s computers. The flight deck we were in could be swapped with one from a 737 or a 787, and the turbulence reprogrammed for the size and shape of those planes. Then, with the flip of a switch in the control room next door, the cab would start to shake and roll on its piston legs, as if having a seizure.
An airplane caught in turbulence moves both as a whole body and as a flexible structure, Pettit said: “It’s sort of mind-blowing, the number of ways an airplane can bend and twist.” To compensate for all that jouncing and twisting, the flight-control software uses the plane’s sensors to calculate the force and direction of the wind. Then it moves the flaps on the wings and tail to counteract the pressure. “Say the airplane is bending like a yardstick,” Pettit said. “The elevator”—the horizontal flap on the tail—“can move to oppose that. Or the rudder will stop it from moving side to side.” If the plane heaves up, the spoilers can press it back down. If it drops, the wing flaps can lift it up. If it starts to roll, the ailerons can right the ship.
A voice from the control room broke in to make sure we’d snapped on our five-point harnesses. Then the demonstration began: “Three, two, one, now.” The flight deck lurched, and we began to plunge up and down. It was simulating a moderately turbulent wind recorded over Idaho and Montana. But the control room had segregated the wind’s vertical and lateral forces, so I could feel them in isolation. We were being bounced around by the vertical ones now, and it was kind of fun—like riding a hobbyhorse. The lateral motions came next. They were only half as strong, Pettit told me later, but they felt twice as discomfiting—slow, seasick waves like ocean swells. But the worst, by far, were the motions that were both vertical and lateral. When the control room programmed the flight deck to re-create full turbulence, the steady waves suddenly turned chaotic, off-kilter, completely unpredictable. A jolt or two, an odd pause; a little jerk to the side, and then the bottom fell out.
“The trick in turbulence is that it’s all jumbled together,” Pettit said afterward. “When the airplane is bending in that noodly way, your body is bending, too. It’s designed to carry your weight vertically. But, if you’re on a seat, you might move from side to side at one frequency, and then, at a higher frequency, your head might go one way and your body in the other. It can be harder for motion sickness. And there is a frequency where your eyes begin to rattle and you can’t focus anymore.”
We ran a few more tests in the simulator—I kept telling them to crank it up—but the results were largely the same. After each set of waves, the control room would repeat the test, only with the turbulence-dampening software turned on this time. When the motion was only vertical or lateral, the effect was dramatic: big waves turned into small ones. But when the motions were merged the dampening seemed to hardly make a difference. The over-all movement decreased, Strefling assured me, and the software took the edge off a few bumps. But the sudden jolts and drops were still there, and you still didn’t see them coming.
“It’s something I’ve always struggled to explain,” Strefling said. “Sometimes you have chaotic events with multiple modes that suddenly coalesce, and those things are really hard to dampen and control. It bothers me to my core that I can’t improve them. You know, I’ve been working on this for over a decade. I will never stop working on it. I will work on it forever.”
When I flew out of Seattle the next day, the sky was roiled with clouds, threatening rain. The government shutdown had ended, but the atmosphere still seemed suspiciously under-monitored. The National Weather Service had lost some six hundred workers; the F.A.A. was short more than three thousand air-traffic controllers; and there was talk of dismantling NCAR altogether. Russell Vought, the director of the White House Office of Management and Budget, had called the research center “one of the largest sources of climate alarmism in the country.”
Still, when my plane’s turn came, the takeoff was as exhilarating as ever—the low rumble and rising thrum, the smooth detachment and sudden lift, and then the surge through clouds into a piercing blue sky. To fly is to be suspended in disbelief. Hurtling through thin air at subzero temperatures, thirty-five thousand feet above Earth, you have no choice but to trust that the atmosphere won’t kill you. That the giant machine you’ve boarded won’t fall apart. A Boeing 777 is pieced together from more than three million components, many of them essential to keeping it aloft. And the atmosphere has countless more moving parts. As the French aviation pioneer Pierre-Georges Latécoère put it, in 1918, “I’ve redone all the calculations, and they confirm what the experts say: It can’t work. There’s only one thing to do: Make it work.”
Our fear of turbulence is both a natural and a disproportionate response to that improbable feat. Our reactions to it are as chaotic and nonlinear as the winds themselves. “It’s easy to run a simulation for a few seconds and make it a little better,” Pettit said. “But if you’re in flight and you’re getting that moderate turbulence for an hour or two, the anxiety builds up. After a while, I’m, like, Turn this airplane around, I want to go home. I don’t need to go to Hawaii anymore.” ♦
