Pillar of fire
In 2011, a New Mexico wildfire went from normal to nuclear, kicking up a 45,000-foot column of tornadic winds and burning debris. Three local scientists set out to learn why.
A WORRIED HOMEOWNER NAMED MARK Winkel stood on his porch and pointed his telescope at a wildfire ripping through the forest several miles from his home outside Los Alamos, New Mexico. The blaze had started 12 hours earlier when a strong gust knocked an aspen into a power line in Las Conchas, a hiking trail along a 13-mile-wide caldera called the Valles Grande. Already it had torched about 7,000 acres, an impressive rate of spread, but predictable given the heavy winds and the onset of fire season, which would last until July’s monsoons finally saturated the tinderbox.
Having faced three big wildfires in the span of 20 years, locals in this part of parched and drought-stricken New Mexico knew enough to consider this one dangerous. But it was now 1:30 a.m., an hour when most fires, faced with cool air, calm for the night. As Winkel pointed his scope up one of the eight canyons that radiate like spokes from the caldera, he saw something unexpected: a yellow-orange wall marching down the southern face of the Jemez Mountains that surround the Valles Grande caldera.
Wildfires don’t typically burn downhill. They climb upward, their flames drying and igniting the fresh vegetation above. This one was racing downslope, at night, directly at Winkel. Worried, he scrambled uphill for a better view. Near the top, a hot wind struck his chest, and he watched to the northwest as the blaze’s front rolled like barrels in 35-foot-high flames. He had never seen this effect before— few people have. Winkel was witnessing a blowup, an intense and sudden force, second in power to a nuclear explosion, able to boil stream water, melt dirt, and crack boulders. This one would spawn a horrific 45,000-foot furnace of smoke and soot, spin up 400-foot-high fire tornadoes, generate powerful updrafting and downdrafting winds, create lightning high in the plume, and send embers flying almost 25 miles away.
Fire-behavior experts had predicted Las Conchas would expand to 12,000 acres overnight. Instead, by the time the sun rose, it had rendered 43,000 acres to white ash. Now the conflagration was advancing on the towns of White Rock and Los Alamos. Afraid it would burn homes, authorities ordered roughly 18,000 residents to flee. Some experts went on TV to express concern that the approaching flames might reach the Los Alamos National Laboratory, which houses tons of nuclear waste. Firefighters from state, local, and federal agencies descended on the area to stop or at least slow the blaze. It took weeks for them to halt its progress.
Today, the eastern caldera remains a scorched moonscape. Six miles from the Los Alamos lab, as Highway 4 climbs steeply into the caldera, the ancient ponderosa forest suddenly gives way to a sprawling expanse of dead trees, a landscape so barren that it takes luck to find a stick thinner than 4 inches. Even pines just beyond the inferno’s reach were baked to death, their needles kiln-dry. In the end, Las Conchas proved one of the most violent blowups in recent history. But what triggered it?
EXTREME WEATHER DRIVES EXTREME FIRE. A STRONG AND RELENTLESS wind in a dry area can stoke even smoldering trash into a runaway inferno. The Las Conchas Fire ignited in a dense forest during the worst drought in millennia, with winds blowing up to 40 miles per hour, 20 feet off the ground, driving the flames forward.
Extreme wildfires, in turn, create their own weather. As intense heat lofts smoke into the air, it forges a convective column that generates powerful updrafts. It carries fuel-rich hydrocarbons, a byproduct of burning vegetation, that can ignite like gasoline vapor. The heat also propels moisture that condenses into pyrocumulus clouds. These anvil-shaped thunderheads perch atop smoke columns and spawn extreme turbulence, downdrafting winds, and even hail that, rather than cooling flames, stokes them by churning out even more erratic winds.
For 22 years, Rod Linn has studied wildfires, building computational tools to unravel the mysteries of their behavior. Linn heads up fire and atmospheric research at Los Alamos National Laboratory.
“Some fields have gaps of understanding that are this big,” says Linn, pinching thumb to forefinger. It’s a mid-April afternoon, and we are sitting in the lab’s expansive research library. “With fire,” he says, stretching his hands far apart, “the gaps are this big.”
Each year, up to 100,000 wildfires burn between 1 million and 11 million acres across the United States, much of it in the drought-afflicted West and Southwest (though a higher percentage of fires burns in the East and Southeast). They claim dozens of lives—both civilians and firefighters—and destroy several billion dollars’ worth of property. The federal government spends up to $2.2 billion annually fighting them—roughly four times what it did 25 years ago. Experts cite three reasons for the inflating trend: a warming and drying climate that turns forests into kindling; the ill-advised practice of extinguishing natural forest-thinning fires, leaving more to ignite later; and 140 million Americans who now live in vulnerable places. That figure was close to zero some 150 years ago, and it grows each year as cities and towns sprawl farther into wildlands. “Fires have probably always blown up,” says Linn. “But the consequences weren’t dire when it happened on a sparsely populated landscape and firefighters didn’t have to fight them.”
It sometimes seems like wildfires are getting more frequent. They’re not. But they are getting more intense, giving rise to an age where megafires raze towns, claim lives, and bring camera crews running. California’s 2013 Rim Fire burned 402 square miles. Arizona’s 2013 Yarnell Hill Fire destroyed a town and killed 19 elite firefighters by trapping them in a canyon. These conflagrations are outliers, monsters that color America’s fear of flames. But they all started the same way, as tiny, insignificant blazes that, for reasons science has yet to describe, transitioned in a few devastating moments into savage blowups.
About four months after the Las Conchas blaze, Linn and some scientists visiting from around the Southwest toured the site. What they saw had all the tells of an exceed-
ingly rare phenomenon known as a column collapse; experts have long believed this event occurs when a tower of smoke and soot gets so heavy, it falls back to earth, creating a wind so powerful that it blows on the surrounding fire like a bellows.
Linn was intrigued. Whatever the exact cause of the blowup, he now had an excuse to study, in his own backyard, one of the rarest events in wildfire science. And he had access to what almost no other scientist did: supercomputing power, courtesy of Los Alamos National Laboratory. He quickly assembled a team of specialists and landed a lab-sponsored, three-year research grant. It was the same type of grant that had launched Linn’s research more than two decades earlier. Back then, his task was crisis forecasting. He used supercomputers to try to predict when and where the next great inferno would burn. Now he would explore the forces that generate them.
ONE FRIDAY IN APRIL, I MEET LINN’S TWO TOP researchers on the Las Conchas project: Jesse Canfield, a slow-talking Chicago-born fluid dynamicist who wears threadbare Carhartts and hiking boots, and Jeremy Sauer, a fast-talking geophysicist and Montana native. We gather around a conference table in the library. Across the street, past a bank of lockers where lab employees stow their cellphones to prevent hacking, a fire-station radio pops with nonemergency traffic.
Sauer stands at a whiteboard, doodling a picture of the Jemez Mountains with flames racing down from its peaks. As the fire burns, he explains, it releases energy that pushes soot, smoke, and hydrocarbon gases into the atmosphere. Take away its fuel— like if it encounters water or rocks—and its energy suddenly disappears. All that stuff that rose, now heavier than the air around it, can now fall back on itself, creating a forceful wind that explodes the inferno.
The technical term for such a wind is a density current, heavy air plowing into lighter air. “In this case, it was a layer of dense air that generated a wind as it flowed downhill,” Sauer says. Canfield had pored through the wildfire science literature, turning up other blowups, including one in 1871 that burned a million acres in a few days. Density currents most likely caused it. But experts had blamed only one fire on a possible column collapse. That was the 1990 Dude Fire in Payson, Arizona, that killed six firefighters. Like Las Conchas, it had raced downslope.
To test the hypothesis that a collapse had stoked the Las Conchas Fire, the researchers headed to the supercomputers. For six months, Canfield coded a simulation of the blaze, mapping the terrain and the smoke column. They wanted to know, theoretically, the highest wind that a collapsing column would produce. In his simulation, you can see an ink-purple cloud rising above Las Conchas topography, but with no fire beneath it, the cloud falls. When the column strikes land, it’s like a detonated building billowing dust, the force of the density wind rushing out in every direction, generating surface winds of up to 131 feet per second, plenty strong to trigger a blowup.
“It was proof of concept,” Canfield says.
Trouble was, the results were misleading. And Canfield knew it. He pulls out a notebook filled with math symbols. “Nerd hieroglyphics,” he calls them. The calculations show that at peak intensity, Las Conchas sent an estimated 2.3 tons of soot into the atmosphere every second. “That’s equivalent to the weight of about 9,000 Honda Accords launched into the sky in two and a half hours,” Sauer says, marveling at the enormity of the fire’s power.
But those Hondas actually disprove the hypothesis. In order for all that weight to fall at once, the plume would have to lose heat faster than it shed weight. Physics dictates that process is a near impossibility. So for this column to collapse so suddenly, the fire would have to have flicked off like a light switch. But Las Conchas did not go out like that. Not before and not after the blowup. In fact, it burned on for five more weeks.
Canfield’s simulation not only disproved that a column collapse triggered the Las Conchas blowup, the lesson likely holds true for the 1990 Dude Fire—and any other blowup blamed on column collapse. “Firefighters have long believed that column collapses exist because that’s what eyewitness accounts tell them,” Linn says. “But eyewitness accounts are notoriously unreliable. The science doesn’t verify what people see on the ground.”
So the team’s primary—and most compelling—suspect turned out to be a boogeyman. After a year of work, it would’ve been understandable if Linn’s team was disappointed. They weren’t. “That’s science,” Canfield says with a shrug. “You generate a hypothesis, then you set out to disprove it.” With one of firefighting’s most persistent myths dispatched, they moved on.
Extreme wildfires create their own weather, forging convective columns that can rise 45,000 feet, spawning turbulence, wind, and even hail.
BY NOW THE TEAM HAD IDENTIFIED A FEW OF LAS CONCHAS’ MAIN characteristics. Among them, something Sauer had spotted from the front stoop of his house: a pair of counterrotating vortexes that churned inward along the entire axis of the 45,000-foot-tall plume of smoke and ash and fire. “I was blown away; seeing that column was to witness all my theoretical science in real life,” Sauer says.
It’s Saturday morning, and I’ve joined him on that stoop as he looks west from the small ranch house toward the burned trees lining the Valles Grande, the caldera about 12 miles away where the plume once towered. The vortexes he saw had created a vacuum that very likely sucked up flames and generated 400-foot-high tornadoes of fire. The same “massive vertical velocity,” as Sauer put it, had ripped pine cones from branches, ignited them, and shot them into the air, where winds carried them as far as 2 miles away, igniting smaller blazes ahead of the main fire. Lighter debris, like pine needles, rose to the column’s full height and rained out miles farther. One witness reported ash falling 25 miles to the west.
At the column’s top, moisture sucked up from trees and brush condensed into water and ice. When these plummeted to the surface, they created a massive downdraft pushing the column toward the ground. “That’s what makes people think they saw a column collapse,” says Sauer. “But really it’s just those high winds blowing downward.”
Sauer goes on about counterrotating vortexes like someone in a midlife crisis talking vintage Rolexes. We enter his home office: a dark room with three computer monitors, and Frisbees stacked alongside dense physics books. Sauer knew the vortexes probably didn’t cause the blowup. He’d also seen them earlier, at daylight, and the fire exploded between 10 p.m. and 3 a.m. “We started asking ourselves,” he says, “what kind of meteorological phenomena produces enhanced winds at night?”
Early in their research, the team considered a number of suspects but set them aside to pursue the column-collapse hypothesis. Now they returned to the Weather Research and Forecasting Model. This uses an atmospheric data set at mesoscale—weather systems ranging from 3 to 60 miles, or roughly the size of sea breezes and ocean squalls—and helps meteorologists create daily forecasts. Using the lab’s supercomputers, Sauer and the team plugged in the wind and local-temperature measurements from nearby weather towers for June 26 and 27, along with topography, such as mountains and canyons, and scaled the forecast to the fire’s size.
What the computer spit out resembled a topographical map, but for the atmosphere, with pressure, wind speed, and direction plotted at several elevations. For the night of June 26, Sauer noticed a curious pattern. Between 400 and 600 feet above the fire, in a boundary between two atmospheric layers, a sine wave representing oscillating pressure began at the crest of the Jemez and rolled and broke in the exact direction in which Las Conchas had blown up: a signal for a strong wind blowing downcanyon.
As he looked at the chart, Sauer recognized that the pass between the Valles Grande and Frijolos Canyon was a textbook example of the topography that can shape mountain waves. These winds form on the leeward side of a peak when low and dense air squeezes between the peak and another air mass above it. As pressure builds, air speeds up—like water in a crimped hose— and becomes a heavy wind. Highly turbulent air above mountains has been known to cause airplane crashes. But almost no research has been done on the surface-level currents that mountain waves create.
Could one have created the density current that blew up Las Conchas? Sauer thought it possible. So he dug in. He spent a year developing a simulation of air moving from the caldera down the canyons. He pulls up the resulting animation. It looks like a rainbow bred with a lava lamp. “Check out right here,” he says, pointing to a river of digitized air squeezing through a mountain pass.
Where his finger rests, a wind current shoots down the Jemez mountain slopes. Near the fire, the air roils like froth at the end of a breaking ocean wave. It was just as Mark Winkel, the only known eyewitness to the blowup, had reported seeing that night: air rolling like burning barrels.
Finally, all of the pieces fit snugly into place. All except one. Their model put out wind measurements as strong as 114.8 feet per second. But the night of the blowup, local weather towers measured winds gusting only as high as 26 feet per second—far too low to be a mountain wave. “We really had thought this was the mechanism,” Sauer says. He clicks away from the animation. “But now we knew it had to be something else.”
ONE AFTERNOON I JOIN CANFIELD, SAUER, and Linn at Los Alamos’ only decent bar, Bathtub Row, a microbrewery busy with scientists who have accepted life beside the moonscape left by the Las Conchas Fire. Beer helps.
From somewhere in the Jemez, a cool wind sinks down on us. “In the summertime,
The culprit, it turns out, may have been slyer and quieter than anyone expected.
every night around 9 p.m., we can hear the wind screaming through the trees,” Canfield says. “Anybody who lives here will tell you about it.”
After determining that the blowup wasn’t caused by column collapse, counterrotating vortexes, or a mountain wave, the team returned yet again to step one, and this time looked inward. They investigated an effect that they’d observed on evenings just like this, talking fire and drinking beer on their back porches. One night, Keeley Costigan, an atmospheric chemist in the lab, overheard them venting their frustration. She had been studying the molecular makeup of smoke particles in the Las Conchas Fire. It turned out that she’d been pulling data from a 150-foot research tower that stands in an unburned canyon just north of the blowup. The team had not included it in their mesoscale modeling.
Intrigued, they pulled the tower’s data. Once again, a compelling pattern emerged. On three of the five nights prior to June 26, precisely between 10 p.m. and 3 a.m., a downcanyon breeze had blown at a rate they recognized with amazement: roughly 26 feet per second. “The fingerprint matched,” Sauer says.
In the nine hours preceding the blowup, 40-mile-per-hour gales had stretched the Las Conchas Fire into a thin and narrow footprint about 6 miles long by 1 mile wide. After the sun dipped behind the caldera’s 11,000-foot Redondo Peak, the temperature cooled, the winds dropped, and the fire—as expected—began to calm.
But then, as the nighttime air cooled and became denser, it began to pool inside the caldera like water filling a 13-mile bathtub. At around 10 p.m., this pool of dense, oxygen-rich air spilled over, generating 26-foot-per-second winds that sloshed down the canyons. They struck the flames perpendicular to the fire’s path. And just like that, what had been a 6-mile-long simmering southern flank woke in a dragon’s breath.
The culprit, it turns out, may have been slyer and quieter than anyone expected. By the time the team figured this out, their grant funding had nearly run out. They couldn’t pursue any other suspects, but to them, the bathtub hypothesis remains the most plausible. It’s also, possibly, the most useful. “Most fire blowups are probably best explained not by the rare or unpredictable,” Sauer says, “but by the relatively common effect of localized meteorology.”
That means future eruptions might be predictable. And that could inform how we fight wildfires. “We can’t predict to the hour or the minute when a fire will blow up or even if it will blow up,” Linn says. “But knowing local weather patterns could tell firefighters that when a fire is burning at a certain time in a certain place, a blowup is possible.”
So firefighters—and the meteorologists who sometimes work alongside them—can deploy more-sophisticated models to predict whether a wildfire will go nuclear, endangering lives and property.
Linn drinks from his beer. Behind him, the setting sun casts numberless torched trees in silhouette. If Las Conchas has a silver lining, it’s a grim one. The risk of another blowup striking Los Alamos is thin. There’s nothing left to burn.
The Fire Detectives The scientists who investigated the Las Conchas inferno, debunking a decades-old fire myth: Rod Linn, Jesse Canfield, and Jeremy Sauer.
Up in Smoke The wildfire torched more than 150,000 acres, created strong winds, and spewed a massive plume— seen here from 60 miles away— packed with smoke, debris, and burning hydrocarbons.
Fire in the Sky On June 26, fire and smoke gained a chokehold on Los Alamos. Fire-behavior experts expected the blaze to calm with the cool night air. Instead,after midnight, it exploded into a rolling inferno.