Firestorm: Douglas Fox on the Reporting Behind His Award-Winning Story

The devastating wildfires of 2018 provide a potent reminder of the impulsive behavior of such blazes. Fire tornadoes can ambush firefighters without warning, updrafts can toss embers ahead of the flames, and firestorms can produce gale-force winds.

These phenomena are powered by the turbulent plume of hot gases that rises above the flames. Such behaviors are hard to predict — in part, because the plume itself is poorly understood. It represents the most ephemeral and inaccessible part of a fire.

In January 2016, I began reporting on a team of researchers who were mounting an unusual effort, using new tools, to monitor the inner physiology of wildfires and the extreme events that they can trigger. My reporting eventually resulted in “Firestorm,” an article in High Country News that won a 2017 AAAS Kavli Science Journalism Award.

The pathway to that story – like the fire behaviors it chronicled – was unpredictable. The following is an account of the story behind the story.

# # #

Craig Clements and Neil Lareau, of San José State University, were using two technologies — Doppler lidar, and radar — to monitor the fire plume’s air currents in a completely novel way. While many stories have tackled the science of wildfires, I loved the idea of writing a deep narrative about nothing more than hot air — a force both mundane and deceptively destructive.

By following Clements to fires, I hoped to gather dialog, sights, sounds, and smells that I could weave into vivid scenes. But that’s not how things went. By July 2016, I still hadn’t visited a fire with Clements, due to logistic snafus — prompting me to ask my editor, Sarah Gilman, whether we should kill the story. We decided to continue with it, recognizing that I would have to re-envision it. The story that was eventually published on April 3, 2017 was far from what I’d originally imagined — but ambitious reporting projects rarely go as planned.

Without access to Clements in the field, I had to reconstruct scenes from interviews, maps, photographs, fatality investigations, and even military reports from the Library of Congress. The scenes needed to be rich in sensory detail — so the reader could truly experience events.  

#  #  #

Early in the process of any story, I sketch a “wish list” of scenes, representing pivotal moments to capture in my reporting. For this story, I researched over 20 potential scenes — of which 11 or 12 ended up in the story. Researching the story’s lead scene — the flight of a research aircraft through the plume of the Pioneer Fire in Idaho, on August 29, 2016 — posed a particular challenge. I was not aboard the aircraft that day.

To reconstruct the flight, I conducted two hours of interviews with David Kingsmill, the University of Colorado scientist who sat next to the pilot. I viewed his photos and listened to his detailed sensory recollections - the sound of pens and notebooks clattering against the windows - and the deceptively innocuous “cool blue” of the plume as it first appeared on standard cockpit radar, during the plane’s initial approach. I interviewed another scientist who often flies on the plane, to better understand its instruments. I viewed photos of the plane’s interior. And I studied aviation weather radar, to confirm Kingsmill’s recollections.

I also worked hard to understand events outside the plane. I obtained GPS flight lines showing its path over the fire. I obtained images that the plane’s scientific radar recorded of the plume. I super-imposed these onto four maps: a map of the fire’s boundaries at specific time points; an infrared satellite image showing the most intense areas of burning; Google satellite imagery showing the landscape, roads, and buildings; and U.S. Geologic Survey (USGS) maps showing the topography of Clear Creek Canyon, which the fire traveled 10 miles up that day.

I reviewed daily fire reports, plus photos and videos taken from other aircraft. I interviewed a firefighter who worked lower down in Clear Creek Canyon that day. And I checked his position on my topo maps, to verify what his line of sight had been. I devoted so much effort to understanding things outside the plane because I wanted to confirm that I was correctly pairing up information on the ground with information in the air. A large fire can have multiple plumes miles apart. I wanted to be sure that Kingsmill’s narrative, and the radar images, actually corresponded with the fire front that surged up Clear Creek Canyon that afternoon.

I was also hoping to get lucky, and discover an even more vivid scene – for example, a building in the path of the fire, which crews were working to protect. I could have interviewed the people who were there — and potentially, captured an amazing firefighting scene that I could pair up with the plane’s simultaneous flight overhead. It would have allowed me to juxtapose the plume with the momentary fire behavior that it was driving on the ground — an important connection to make, since this was the point of the story.

I didn’t succeed in finding that gem, so I had to write a general description of the flames rushing up the canyon. But I often undertake such fishing expeditions, and sometimes they pay off.

This happened as I researched a scene centered on Project Flambeau - a series of experimental fires conducted in the 1960s, to predict the effects of nuclear war on American cities. Government reports provided detail on the fires, which were set up using meticulously-spaced wood piles, to mimic the layout of a suburban neighborhood. But I wanted to go beyond this and place the fire in a vivid landscape.

Reports listed the fire’s location in opaque terminology: “Township 2 North, Range 32 East, Section 26.” I traced that language to USGS maps from the 1890s. Those maps showed only dirt wagon-roads — not features like mountains that would pinpoint the locations on modern maps. But I managed to convert this 19th-century coordinate system into latitude and longitude.

As I called up these coordinates on Google Maps satellite view, I feared it would reveal a generic spot of landscape — which I’d just have to accept (with nagging doubts) as the correct location of the Flambeau fire. But I was pleasantly surprised. The landscape was smudged with faint dots arrayed in 19 tidy rows, 18 dots per row. It showed that the 342 piles of wood that were set aflame on September 29, 1967, had actually left scars still visible after 50 years! Even the scientists who study data from Flambeau had been unaware that this evidence still existed.

This was a small moment of triumph — an example of why I enjoy this kind of digging. It allowed me to introduce the Flambeau scene with a dramatic pan of the landscape, as one would see in a movie, before zooming in to the fire itself:

In the wrinkled, sage-covered mountains of Nevada near the California border, 30 miles east of Mono Lake, there is a meadow that seems to lie in shadow even on sunny days. Spread across it are hundreds of dark patches, where the soil is mixed with charcoal. These spots lay row upon row, like the ghostly foundations of a dead city. In a sense, that is exactly what they are.

# # #

The Pioneer and Flambeau fires represent a tiny sliver of the research that I did for this story. This attention to detail didn’t just let me paint nice mental images for readers. It also allowed me to be extremely selective in choosing scenes.

While researching wild fires, I often read fatality reports laying out the circumstances that led to deaths of firefighters. Those reports sometimes provided important material for scenes, as in the 1994 South Canyon Fire in Colorado, where 12 firefighters died, likely due to a fireball from a vapor explosion. But just as often, those reports prompted me to exclude a fire from my story. One example is the 2013 Yarnell Hill fire in Arizona, which killed 19 crew members. It is sometimes cited as a case of a plume collapsing onto a fire and splashing its flames in unexpected directions. But reading the fatality report muddied that picture, revealing multiple factors, including communication gaps between firefighters, that may have contributed to the tragedy.

Yarnell Hill was one of at least 16 fires that I researched in some depth — and one of eight that I excluded from the story. Wildfires are chaotic affairs - obscured by imperfect observations that make it hard to reconstruct a fire’s behavior. My job as a journalist was to be selective — to choose the rare cases in which a fire’s behavior rises above that noise to reveal a clear insight that I can communicate to readers.

# # #

As I wrote the story, one major problem emerged. The flight through the plume of the Pioneer Fire was an obvious choice for the opening scene. Smoke plumes can appear slow and lazy from a distance. But putting the reader on the plane would shatter that misconception and illustrate Clements’ innovation: using remote sensing to see a fire’s internal anatomy of swirling hot air. There was just one problem. The story also needed a payoff at its end — a surprising revelation that the reader would still find worthwhile after wading through 5,000 words.

I had no other scene that could fill those shoes. So I realized I’d have to split the Pioneer scene — use it as both the opening and closing scenes. This can be challenging. If a scene isn’t substantial enough, then dividing it can prove futile — like splitting a baby aspirin — at which point, neither remaining piece is big enough to do its job. The Pioneer scene turned out to big enough — but just barely.

The opening scene let the reader feel the invisible hand tossing the plane around. The closing scene provided a coda to that experience: it unveiled the radar image that the plane had captured as it traversed the plume. That radar image revealed the chaotic swirl of air currents which had buffeted the plane. This was the very essence of the story: the unveiling of an invisible beast — the moment when a destructive force of nature finally became visible to humans for the first time ever.

(Associated homepage image: Pioneer Fire plume, 2016. Photo credit: Kari Greer)