On a narrow road in southern Iowa, my phone shrieked.
Tornado warning. Take shelter now in a basement or an interior room on the lowest floor of a sturdy building.
Again, the alert sounded. Then again. As if shelter were a choice.
I peered through a peephole of scratched glass, wedged into the lead storm-chasing truck with tornado researchers Josh Wurman and Karen Kosiba.
On either side of us, only fields. No houses. No basements. No sturdy buildings. The dark belly of a storm dragged across a horizon gone an eerie green.
For tornado chasers, it was a near perfect sky.
My son Cheney, a photojournalist, was somewhere behind us in a rental car alongside two other research vehicles. The convoy was strung out across miles of highway. He had been tracking my location, but now I had lost cell service.
I couldn’t reach him. I kept dialing. No answer.
When a call finally went through, Cheney told me the wind was driving the rain so hard the road had disappeared.
“I’m scared,” he said.
Blinded, he had turned one way, then the other. Finally, he stopped. Either direction, he feared, could put him in a tornado’s path.
Cheney was no stranger to danger. He has covered Ukraine during wartime, spent a New Year’s Eve alone in a car reporting on Chicago homicides, and crisscrossed the country documenting violent weather, including the devastating 2021 tornado in Mayfield, Kentucky.
I hadn’t heard him express fear since childhood.
Then he said it again. “I’m scared. I don’t know what to do.”
About 1,200 tornadoes strike the United States every year. Between 2020 and 2025, they killed roughly 410 people, injured nearly 3,800 and caused about $8.5bn in reported property damage.
The tornado map is changing too. A 2024 study found 25% fewer moderate-or-stronger tornadoes in the western US and 12% more in the east, where mobile homes are more prevalent, basements to get to safety are scarcer, and twisters more likely to strike at night.
Sometimes the change from ordinary to deadly takes only minutes. Several months earlier, physician Jason Persoff had told me about chasing near Joplin, Missouri, in May 2011. A little after 5pm, he and his partner began looking for hotel rooms, figuring they were done for the day.
“Suddenly, it was as if someone vomited on the radar,” Persoff remembered. “It was incredible.”
A violent tornado seemed to emerge from the sky almost fully formed. Its sound lodged deep in Persoff’s body. “I felt irrational fear,” he said. “Like a primal warning.”
They pulled over to help stranded truckers as police arrived. Persoff told one officer he was a physician.
“I hope you don’t work at St John’s,” the officer told him. “That hospital’s been destroyed.”
“That’s still one of the most terrifying things I’ve ever heard,” Persoff said.
More than 150 people were killed in that tornado.
Wurman, whose team I traveled with, is among the researchers trying to understand why some storms produce tornadoes while others do not, and what tornado winds do close to the ground – knowledge that could improve warnings and help engineers design buildings to withstand them.
But the systems built to study and forecast tornadoes are under strain.
In the first four months of 2025, the National Weather Service lost roughly 13% of its workforce – more than 560 employees – to firings, retirements, resignations and attrition.
By the 2026 storm season, the agency was hiring again, but staffing gaps remained, and some stations were unable to maintain the twice-daily weather-balloon launches on which forecasters and researchers rely.
Scientists I spoke with said they had lost confidence in forecast guidance. Chris Weiss of Texas Tech, who led a parallel chase team with the National Severe Storms Laboratory, was blunt: “I don’t remember when a suite of models has performed as poorly as this year.”
Frustrated by the blurred, distant view from fixed-site radar, Wurman invented the Doppler on Wheels – mobile radar laboratories known as DOWs – three decades ago. Filling one with diesel costs about $1,000, and that is the inexpensive part: the greater investment lies in the trucks, trained graduate students and experienced crews that take years to build.
But funding has become more uncertain. By the end of this field season, neither Wurman’s team nor Weiss’s could count on support for the next one.
Cessily Skinner, whose family has lost two homes to tornadoes in west-central Georgia, closely monitors every approaching storm. Now the cuts give her another reason to fear: “As someone who lived through a storm with all the data in place,” she said, “it’s terrifying to think, ‘OK, so now we really are guessing.’”
With Cheney still on the line, a team member plotted his location against the storm. The radar could see what he could not. A tornado we were tracking was east of him and moving east. Cheney should be safe, she said, as long as he stayed put and waited it out.
I repeated this to him.
As we spoke, the rain loosened. The sky lifted. The road reappeared. Cheney pressed the gas and began closing the distance between us.
The second radar truck arrived first. Fifteen long minutes later, Cheney’s rental car came into view.
Relief and disappointment arrived together. The convoy was whole again. But the day had yielded none of the data the scientists had come for.
Several tornadoes touched down across the region that afternoon. None crossed the roads where our instruments were waiting.
The storms left behind a sky ribboned with dark clouds. Some team members stood outside the trucks and applauded its beauty, lifting their phones for photographs. Wurman and Kosiba were not among them. They were already studying the models, deciding where the team should spend the night to be best positioned for morning.
Tornadoes offer enough spectacle to draw film-makers, thrill-seekers, and carloads of chasers with cameras. But scientists who chase for data need more than drama. They need geometry: an open horizon, a road in and a road out, a storm that forms where the instruments can reach it.
Even when everything lines up, a mobile radar crew can spend weeks following one promising storm after another, only to watch each weaken and dissolve before the instruments capture the measurements they came for.
Why do they keep doing it?
For data analyst Paul Robinson, the pull started with an image on television: Dorothy clutching Toto, calling for Auntie Em as a black twister writhed in the distance.
“The Wizard of Oz is still my favorite movie tornado,” he said. “For me, it captures the feeling.” Growing up in West Virginia, he began keeping a notebook, recording the weather whenever it turned severe.
But 31 May 1985, turned that interest into purpose.
Everyone on this team seems to carry a storm they cannot forget. Robinson’s was not a single tornado but an outbreak: 43 twisters tearing across Ohio, Pennsylvania, New York and Ontario, killing 89 people.
“That convinced me that maybe I could do some good by going into tornado research,” Robinson said.
Research scientist Trevor White’s defining storm came later, on 24 May 2016, near Dodge City, Kansas.
“Every chaser and their mother came out,” he recalled. “We were on it from the time it was a small cumulus cloud. I had never seen a storm that could effortlessly produce tornadoes. I counted three on the ground at the same time.”
In two hours, a dozen tornadoes formed beneath the supercell. It was nearly perfect in the way that mattered most: no one died, and the tornadoes remained mostly over open country.
But even experienced researchers cannot always remain beyond a tornado’s reach.
In 2013, veteran tornado researcher Tim Samaras was killed near El Reno, Oklahoma, along with his son Paul and colleague Carl Young. Samaras was Wurman’s friend. Wurman’s team was tracking the same tornado that day.
Asked by a reporter whether the deaths made him reconsider his work, Wurman noted that covering extreme weather “isn’t dangerous statistically”. Samaras, he said, was among the most responsible chasers he knew. Wurman would not speculate about what had gone wrong but said his own team would review its procedures.
Wurman and his colleagues then returned to the data, reconstructing the motion of the nearly impossible-to-see subvortex that had overtaken Samaras and his team.
He pursues the sky at its most theatrical without surrendering to drama himself. He would rather leave a question open than force an answer the data cannot support.
“I wish this research were more of a eureka thing,” Wurman said, “but it’s not. We can’t do laboratory-type experiments. We can’t put 10,000 patients into a study. Meteorology is horrible that way. We’re trying to understand what’s special about the circumstances where tornadoes form and what it looks like when they do. But we can’t control the environment.”
“Every time we’re in a disaster area, I carry with me the trauma of people who survive it,” White told me. “I don’t know if we can do anything to ease the impact of that experience, but I’ll never stop searching. The only thing worse than a human tragedy is a human tragedy where you can’t learn anything.”
A mile outside Good Hope, Illinois, the deployment order came. Our truck slid to a halt on the shoulder of a two-lane road slick with rain.
Robinson jumped out to unleash the radar dish. It began sweeping the sky, each rotation shuddering through the vehicle.
Back in the driver’s seat, he talked to himself – “c’mon, c’mon” – as he worked the controls to lower a stabilizing pad into the mud, trying to keep the vehicle’s right side from sinking. He kept the left wheels on the pavement, he told me later, in case we needed to move fast toward – or away from – a tornado.
In the back, White balanced one keyboard on his lap and kept several others within reach. Four screens glowed in front of him.
“Frustrating,” Robinson murmured. “We’re getting data, but I want to see it; I want to see the fury.”
The horizon closed around us. From the front passenger seat, I watched a low shelf of black clouds over the cornfields, its ragged underside reaching down in unfinished gestures.
The walkie-talkie crackled, then Wurman’s voice came through. “What do you see?”
No one answered immediately. White and Robinson studied the screens and conferred quietly in technical shorthand. They were looking for a hook curling in the rain, a couplet tightening in the wind field, some sign that the atmosphere had made up its mind.
The clouds kept darkening. The dish kept turning. The truck kept shuddering. Robinson waited through one scan, then another, watching new data appear on the screens over White’s shoulder.
Finally he spoke, his voice tight. “It’s falling apart, isn’t it?”
After a few more minutes, Wurman confirmed he saw the same. He called out a new waypoint and then we were speeding again down a country road, pursuing the uneasy hope at the center of this work: a storm dangerous enough to teach but merciful enough to spare.
When things align, the value of a single intercept can endure for years. On 18 May 2013, the team captured data from an EF4 tornado near Rozel, Kansas. It remained on the ground for about 30 minutes, long enough for the researchers to hold position and gather unusually rich measurements.
“That gave us a chance to dig in deep,” Kosiba said. “And that data is still being used by engineers today. That’s why you have to keep going out, even with the misses. We’re in it for the long haul.”
A 2024 study coauthored by Wurman and Kosiba analyzed mobile-radar wind profiles from 36 tornadoes. Those findings helped shape new national standards for designing buildings to withstand tornado winds, showing the potential impact of readings collected in the path of storms.
This is the scale on which science advances: one measurement clarifying another, a dataset enduring long after the tornado itself disappears.
Behind every chase is a life forced to accommodate the forecast. We had driven 2,000 miles in three days, from dawn until after dark, crossing and recrossing North Dakota, Minnesota, Wisconsin, Iowa and Illinois. Wurman’s team had gone weeks without the intercept they needed. Weiss’s team had endured an equally disappointing season.
“We put in the same effort for fewer results,” Weiss told me. “We have lives and families, and we’re making a lot of sacrifices. So when it doesn’t work out, it eats at you.”
White has a soft spot for hummingbirds. He keeps several feeders at home, and goes to great lengths to find someone to refill them when he’s gone. His absence is measured out in sugar water. He carries a small stuffed hummingbird on the road.
“I told my wife a week ago we’d be home tomorrow. Instead, we are still out here, and there’s no end in sight,” he said. “But there’s something fundamental about this research that draws me in. The best version of me is the version that does this work.”
Before this trip, I had imagined storm chasing as a form of appetite, the desire to see the most violent thing the sky could make. I hadn’t understood its strange rhythm: tedious hours beneath empty skies, then sudden races down country roads, decisions pressed into seconds, and moments of stark fear.
The work demands patience when nothing happens and clear judgment when everything happens at once. Inside the truck, it looked less like appetite and more like devotion: painstaking, deliberate, largely invisible.
In a culture that rewards quick verdicts and emphatic claims, these scientists let the sky contradict them. They reverse direction and answer to evidence rather than personal conviction. That discipline feels radical.
Years from now, the data they gather may become a stronger roof, an earlier warning, hospital walls that hold. The researchers may never know whose lives their persistence reaches.
That last night with Wurman’s team, I wrote three lines in my notebook:
We went storm chasing.
Nothing happened.
We learned a lot.
