Printed for personal use only

In the Line of Fire

Los Angeles sees more wildfires than any other city in the United States. Last May, when Griffith Park went up in flames and parts of Los Feliz were evacuated, a long-forgotten threat to the city came home

Photograph by Ted Soqui

The men and women whose job it is to fight Southern California wildfire perish by all possible means and in every circumstance imaginable. In the Angeles National Forest, a helicopter flying at night lands in darkness atop a second chopper already parked on the helibase. Two airmen die. In the same forest, a fire chief and his crew are surprised by advancing flames. The crew flee in one direction; the chief escapes in another, until—worried of his men's fate—he returns through fire in search of them and is killed. Above the northern San Fernando Valley, 12 firefighters are caught in a wind shift along a steep ravine that swirls superheated gases over them, raising the tiny canyon's temperature to 2,500 degrees. A fire engine driver racing to a call flies over a San Bernardino rail crossing and is smashed to death by an oncoming train. Beside Bryant Canyon, high in the Angeles Crest, a burning rat runs at two men in heavy brush, surrounding them in fire . On a San Diego blaze a firefighter pauses to talk with a passing bulldozer operator. His trousers become entangled in the machine's moving tracks, and he is pulled under and crushed. While fighting a blaze in Orange County, a fireman drops dead of pneumonia.

Wildfire in the Golden State, and especially in Southern California—the nation's maximum fire-prone landscape—is the most dynamic, violent natural event that people engage with. In sheer energy and unpredictability, a hurricane is as close as you can come to the riotous mien of a Los Angeles chaparral fire. We do not attack hurricanes, or earthquakes, or tornadoes. We do attack, however, what is essentially photosynthesis thrown into reverse, as foliage instantaneously releases stored solar energy in the form of hot gases—what we see as flames.

The 19 largest and most costly fires in 100 years have ignited within the last quarter century. Yet wildfire for Angelenos has typically remained an occurrence that happens "out there" —in the unseen San Jacinto wilderness, somewhere above lonely Morongo Valley, on a distant Los Padres plateau. Stories of firefighter deaths and injuries, or images of entire forest communities left in ashes, with lives ruined and fortunes lost, are annually beamed into the living rooms of Hancock Park, Alhambra, and Encino, like scenes out of Iraq. Usually, the smoke cannot even be spied from backyard porches. The misery on household TV screens might as well be happening in another country.

All that changed on the afternoon of Tuesday, May 8, when an 817-acre wildfire engulfed the epicenter of Los Angeles, Griffith Park. In a matter of hours one of L.A.'s oldest and wealthiest neighborhoods found itself threatened by encroaching flames. Firefighters engaging the blaze were caught on canyon slopes and gully bottoms, trapped by the chaparral inferno. Now smoke could be seen from every porch. What had always been out there was suddenly here—within walking distance of the corner of Hollywood and Vine.

Griffith Park is a hub of culture and recreation, home of the city's observatory , the Greek Theater, the Autry National Center, and the L.A. Zoo. There are four golf courses, pony and horse stables, a miniature railroad. But the park contains a wilderness as well, thousands of acres of chaparralcovered hillsides that are no different in makeup from the slopes of the Angeles Crest or the brushy savannas of Ventura County. By nightfall on May 8, city residents faced the unforeseen, or at least the long forgotten —a forest fire in the heart of L.A. That evening, winds were blowing in gusts of up to 35 mph out of the northwest. Had those same winds been from the northeast—had L.A. been under the high-pressure blast of a Santa Ana condition— Griffith Park's blaze could have jumped into the densely developed Hollywood Hills and worked its way toward the 405 freeway.

Like hurricanes, wildfires exhibit their own inherent behaviors— flashovers, eddies, flame spires, tornadoes, extreme spread rates, and erratic shifts in direction. While more chaparral fires are fought today than ever before, we understand relatively little about wildfire behavior and its chemistry. For example, we know far more about how coal ignites and burns inside a Texas power plant—or how diesel fuel interacts with the engine of an 18-wheeler on I-5—than we do about the chemistry of wildfire . "Wildfire prediction has become the back pocket I work in," says Drew Smith, who is a nationally recognized fire behavior analyst with the L.A. County Fire Department. "But fire still fools me every day. When it comes to wildfire behavior prediction in Southern California, you are always handed the test before you're given the lesson."

As far back as the 1950s, wildland firefighters were employing some form of behavior prediction. Often it is a simple set of pluses and minuses that either add up to staying put or subtract to evacuating. Moisture and calm weather are pluses; steep terrain and high winds are minuses. When a 36-year-old, heavily tattooed auto mechanic named Raymond Lee Oyler allegedly lit the Esperanza wildfire in Riverside County last fall, the crew of Engine 57 that responded to defend the Twin Pines area was led by a captain with 16 years of experience. In his career he'd attended classes like "Fireline Leadership" and "Time Pressured Simulation Assessment" and "Fire Operations in the Wildland Urban Interface." Yet the math in his head went wrong that day, the minuses converted to pluses, and all five members of Engine 57 were killed. Their deaths underscored a central question, one faced by fire agencies across the L.A. basin: How do you ensure that your crew members—even those who have received classroom training—appreciate the chaotic behavior of the fires men and women set out almost daily to halt?

Thoughts attendant to this question were on Tony Varela's mind when he sat down at a restaurant in the Mutual Threat Zone on the afternoon of May 8 and ordered a chicken spinach salad with Craisins. The MTZ, as it is known, is a multijurisdictional territory encompassing the northeast San Fernando Valley's chaparral-covered slopes from Sunland to the area where the 210 meets the 118. A fire lit in, say, Pasadena, is responded to by the Pasadena Fire Department. A fire in the MTZ can bring in the Los Angeles County Fire Department, the Los Angeles Fire Department , the Angeles National Forest's brush teams, and the Glendale and Burbank fire departments.

Varela grew up in the MTZ. He is an enthusiastic and unselfconscious man in his early fifties with shorn silver hair, a booming voice, and a personality he describes as "all type A with a sprinkling of type B on top." There are five bureaus in the LAFD: emergency services, fire prevention, training and risk management, administrative, and support services. Varela is an assistant chief in special operations, a division of emergency services, which includes urban search and rescue, HazMat, the department's helicopter fleet, 104 fire stations, and 3,560 employees.

A force that large would seem impressive for any fire agency, but each week that Varela goes to work on the 18th floor of City Hall East, he sees a new flyer taped to the wall announcing the departure of yet another chief or captain. "The guys who fought the big wildfires of the past are all retiring," says Varela. "We are losing that experience in the department." Structure fires, not wildfires, have been the LAFD's bread and butter. "We don't lack for emergencies," says Varela's boss, Deputy Chief Mario Rueda. "House fires, natural disasters, riots, car crashes, rivers that people fall into. But in the past, I don't think enough emphasis was institutionally placed on wildland fire behavior. That's one problem, in addition to the ongoing retirements, and one reason I want Varela to now participate in outside agency training and state wildfire management teams."

And spinach salad luncheons in Sunland restaurants. Sitting around the table that Tuesday were Varela's MTZ partners. There was L.A. County's Matt Gil, who also grew up in the MTZ before finding himself charged with protecting 1,800 square miles north and east of Valencia. Varela was telling Gil of a conference he had recently attended where he had been excited by talk of DC-10s that can drop 12,000 gallons of Phos-Chek retardant on wildfire. Gil smiled at Varela. "Please! You're kidding me," he said. "DC-10s are not going to handle a wind-driven fire that starts in Griffith Park and burns to the beach."

Next to Gil sat Tracy Pansini, Burbank's fire chief. Pansini grew up in the MTZ as well. He and Varela became best friends as boys. They knew every trail and canyon in Sunland and the Big Tujunga Wash. When the 47,000-acre Mill Fire swept out of the Big Tujunga's mouth in 1975 and down onto Pansini's neighborhood, the two teenagers, a year out of Verdugo Hills High, opted to stay the night on Rim Canyon Road. "We were in shorts and tennis shoes," says Pansini. "The Santa Anas were blowing, and we were young and dumb and running on adrenaline, and we spent that night jumping from roof to roof with hoses in our hands."

The Mill Fire drew the two men into fire fighting, though it's safe to say that more communication existed between the two "dumb" teenagers with hoses than between Southern California's fire agencies in the '70s. "I love my L.A. City guys," says Gil. "But County Fire 's culture back then was that wildland guys did not play with structure guys." County Fire thought of the LAFD as suits who sat around polishing brass all day. "I know that even up to ten years ago," says Deputy Chief Rueda, "County still jokingly thought of City as buffoons in the brush." On the other hand, says one firefighter, "city crews thought of wildland guys as basically knuckle-dragging ex-inmates who sleep on the dirt."

"What that led to back then," says Gil, "was no communication between agencies responding to the same fires." Agencies were jurisdiction bound, defending their turf, but fire ignores jurisdiction. "Before strategic planning," says Pansini, "incident commanders and operations chiefs of different fire agencies were not talking, so fires were being put out by a phantom organization. It was the folks on the ground that had to figure how to put out the fire." A worst-case scenario of poor strategic cooperation came about in 2003's Cedar Fire, which burned more than 280,000 acres and 2,232 homes and killed 15 people. "In a megafire like that," says Varela, "every agency in the fire's path was out to protect their jurisdiction, their town. But by remaining loyal to their jurisdictions, they could never piece together a communal defense plan and get ahead of the fire. The fire was always ahead of them."

So Pansini came up with the idea of a monthly multiagency lunch inside the MTZ. "We have unified command training now," he says, "building fires on maps, running them jurisdiction to jurisdiction, figuring out unified responses." True, not every one of Pansini's luncheons ended on a high note—nearby La Tuna Canyon had caught fire just hours after a previous lunch. For Varela, however, the MTZ lunches were class time: "Those guys question and critique our operations to see if we're measuring up in brush."

That Tuesday's critique focused on the Griffith Park fire of March 29, which had started at Barham Boulevard's Oakwood apartments, where the actor Telly Savalas had once lived, then roared up Mount Lee, before coming to rest not far from where Savalas is buried in Forest Lawn. Pansini was pleasantly giving his old friend shit. "Burbank is just 140 folks," he said. "But we laid 1,000 feet of progressive hose on that fire for L.A. City—and there are 3,500 of you."

Something else about the fire of March 29 was bothering Pansini. If you had watched Fritz or Dallas or Johnny that day, you knew L.A. was entering what could be the worst fire season on record. Rainfall was at a near-record low, said the weathermen, the humidity down, and the vegetation moisture level hovering around 60 percent— what LAFD deems "critical" status. "Low rainfall is cyclical," Pansini was thinking . "But for two nights in February we sustained an under-30-degree freeze. Anything already half dead on L.A.'s hillsides from lack of rain would have been decimated by that freeze." A second fire in Griffith Park could be driven from the result of just two cold nights in February.

About the time the check was called for in the MTZ, a marine on leave from Camp Lejeune in North Carolina stumbled out of a Griffith Park canyon named Aberdeen. His seared abdomen exhibited a sickly yellow hue, and behind him the canyon, filled with cheat grass, manzanita, toyon, Aleppo pine, and walnut trees, was ablaze. By midnight he would be in a bed at the Grossman Burn Center in Sherman Oaks, under police watch and suspected of arson. Lunch ended in Sunland, and Varela climbed into his city-issued Crown Victoria, exiting the MTZ by way of the 210. Dropping into L.A. on the Glendale Freeway, he looked out his passenger window and saw a chocolate-colored cloud over the park.

All across the city, LAFD battalion chiefs were catching view of the same sight . Patrick Shanley, who would lead the fight to save the park's nursery, was visiting Station 89 in North Hollywood when he spotted the smoke. Ray Gomez, who would spend the night with a dozen or so engines on a park ridge, was on Sunset Boulevard downtown when he spied what he describes as a mushroom cloud. Mark Saxelby, who would take command of Los Feliz's adjacent streetscape, was at the department's Elysian Park training center. And Craig Fry, the Hollywood battalion chief who would initially assume operations command, was busy counting out 840 feet—the distance LAFD wanted to move a station house.

Inside his Crown Vic, Varela reached Rueda by phone. It was 2:15 p.m. "Head over to the staging area at the Greek Theater and meet up with Fry," Rueda instructed him. Varela took another look at the spiraling header . Jesus. First La Tuna Canyon, now this. They were going to need a new system for picking out luncheon dates.

In the Greek's parking lot Fry had just finished building an ad hoc command center on the tailgate of his red-and-white LAFD Suburban. That day's brush burn index, a calculation based on local flora moisture and weather, was at 212—deemed "critical fire weather"—and LAFD resources had been positioned across the city in case wildfire broke out. Additional engines sat waiting in Sunland, Woodland Hills, Mission Hills, near Coldwater and Laurel canyons. Two bulldozers rested off Sepulveda Boulevard in the Valley. None was close enough to do Fry any immediate good. He glanced at the drifting header . Already two engines from Los Feliz's Station 35 were climbing Commonwealth Canyon Road into the park, searching for the blaze. "It could be in Aberdeen Canyon," Fry thought. "It could be a canyon to the west." He wasn't sure, and he was less sure of the fire's size.

Depending on the emergency, LAFD sends out a predetermined number of vehicles. A grass fire alongside the freeway ordinarily receives two engines and one battalion chief. A small house on fire receives two engines and two ladder trucks, and a large commercial building initially gets three engines and three ladder trucks. Fry had ordered 20 engine companies before he even reached the Greek. Any LAFD request for more than 15 companies is a "major emergency," meaning the chief of special operations is quickly notified. In his 18th-floor corner office, Rueda was already clearing his schedule to get to Griffith Park.

Fry is a cousin of Rueda's. He has 28 years on the job (6 as Hollywood's battalion chief), is small of stature for a firefighter, and he is haunted by a nightmare concerning Griffith Park. He envisions a conflagration igniting under Santa Ana conditions, one that burns west and out of the park into Beachwood Canyon, crosses the Cahuenga Corridor, then jumps Laurel Canyon, Fryman Canyon, Coldwater and Mandeville. He has no faith in Varela's DC-10s and their 12,000-gallon Phos-Chek loads. "Griffith Park is directly aligned with the Santa Anas," Fry says. "There is brush 50 feet high in the park." Nothing can stop a fire front propelled by Santa Ana winds above 40 mph. Agencies are helpless. "A fire like that," says Fry, "would be a catastrophic emergency of the likes the city of Los Angeles has never seen."

Mark Saxelby, driving in from Elysian Park, pulled next to Fry and was assigned reconnaissance duty. "Get up on the other side of the fire and tell me where it's at" were Fry's parting words. Saxelby circled along Commonwealth Canyon Road onto the back side of Roosevelt Municipal Golf Course , passing the mouth of Aberdeen Canyon, and then turned onto Vista del Valle Drive, a twisting spine that climbs the park's eastern slopes.

Ray Gomez arrived at Fry's post, followed by battalion chief Gerry Malais. Right there Fry split the fire down the middle in two divisions , Alpha and Zulu. "Find the hot flank and take that," he told Gomez. "You're Zulu." Malais and Gomez drove off , also passing Aberdeen's mouth. "That fire was up that canyon," Gomez says today. Ideally, you attack a wildfire by locating its origin, or anchor point, then work forward, flanking and fighting it in pincer formation from either side—in Aberdeen's case, Malais's Alpha and Gomez's Zulu. "But when we were all driving past Aberdeen," says Gomez, "it was still green. The topography was such that you couldn't see smoke—nothing. You had no idea that the fire's anchor was up in there."

Edging along Aberdeen's eastern ridge, Saxelby sighted the fire from a cliff-side road that led to a footbridge straddling the canyon's apex. The flames rolled in lazy waves around Aberdeen's bowl. Saxelby is in his late forties, has a fondness for plug tobacco, and exudes the kind of calm, weird, militaristic machismo of Robert Duvall's Colonel Kilgore in Apocalypse Now. "Shit," he thought, "a few water drops and we are going to hook this thing and be home in two hours." Gomez, climbing Vista del Valle and watching the smoke trail off in white wisps, had a similar thought: "Fry is calling in an awful lot of resources for nothing. This is going to be a piece of cake."

At the Greek an official-looking car skidded to a halt in front of Fry. "Which chief is this?" Fry was thinking. City Councilman Tom LaBonge sprang onto the pavement. "Craig! The park's on fire!" La- Bonge blurted. After six years in Hollywood, Fry knew LaBonge well and was fond of him. But he also knew how excited the councilman could get.

"Tom," Fry said. "I can see that the park is on fire."

"Okay, okay," said LaBonge. "Okay. I'll go out and I'll tell you where it's at."

"All right, Tom," said Fry. "You go out and you tell me where the fire's at."

As LaBonge careered off , Fry couldn't help noticing that the councilman's wife, Brigid, and a young legislative analyst were hanging on for the ride.

Saxelby had shut down the cliff-side road—too narrow and dangerous for trucks. Without that access, engines 35 and 235 had to follow Vista del Valle around to the back of Aberdeen, then shimmy their hose lines up a nearly vertical, 150-foot rockfall to position themselves in a saddle under the footbridge. William Heritier, holding an inch-and-a-half line that sprays 125 gallons a minute, gazed into Aberdeen. "I was looking down a chimney, down a chute," he says. "That's a bad spot to be in. But we wanted to get water on it and knock it down." Beside him stood Brian Walker, who clutched a second hose line. The two watched as the fire slowly made its way toward them.

Everything was beginning to speed up around Fry. At the Greek, engines were arriving, ready for assignment—not only LAFD engines but companies that would come from Glendale, Burbank, San Marino, and L.A. County. This was exactly the kind of interagency jurisdictional response that had been MTZ chatter an hour earlier.

As if on cue, Varela appeared. "Craig— where do you want me?" he asked Fry.

Varela is Fry's senior. But as first on the scene, Fry held incident command for the moment. By now he had ordered 40 engine companies, some 400 men, to the Greek. Sensory studies show that perception narrows in stressful moments like the one Fry was experiencing. Situational awareness drops off radically. An individual communicating with three to seven others can keep track of what is going on around him and those he is speaking to. When that number climbs above seven, a commander becomes distracted.

The LAFD's emergency response system is designed so that when a commander's lines of communication become too numerous, his responsibilities are split among others, dropping him back into the ideal awareness zone. This is part of the Incident Command System created in the aftermath of October 1970, when 773 fires raged across California. Following the attacks of 9/11, Presidential Directive Number Eight, signed by President Bush, made California's ICS the official national emergency response model. When Hurricane Katrina flooded New Orleans and Minneapolis's Bridge 9340 collapsed into the Mississippi River, the shapes of those emergency responses were born out of chaparral flame.

"You're liaison," Fry told Varela, halving his own duties. "Start taking down the names and numbers of those engines."

Golfers began running off the Roosevelt course into the Greek Theater's parking lot. "Hey, we got burned-up people over here!" each one was telling Fry. In fact, only one individual, the marine, had suffered burns, and an EMS truck had already responded to him. Fry didn't know this. Golfers kept running off the course. Others approached the arriving engines on Vermont. Every report off the green concerned the same man, but Fry imagined swells of hikers staggering out of the charred brush. It was some time before he got a handle on what was really happening, and then LaBonge was back, zooming into the parking lot.

"Craig! Craig!" the councilman shouted. "The merry-go-round is about to burn!"

"Tom, you're crazy," Fry responded. The park's merry-go-round was more than a mile by foot and a mountain away from where he knew the fire was burning. "It can't be at the merry-go-round yet."

"No!" LaBonge replied. "It's burning, it's burning. It's going to burn!"

Fry looked at LaBonge and said nothing. But he began to think, "Did the park have an arsonist at work?" Three years earlier, someone had set fires by the merry-go-round week after week after week. Was the arsonist back at play?

* * * * *

Ninety-eight percent of fires in Southern California are caused by human activity. Occasionally a lightning strike ignites the brush. In 1967, in one storm, 13 separate fires were started by lightning strikes on the Angeles Crest. But a storm like that is unusual. The Santa Monica Mountains have seen just two lightning fires in the last 30 years, and Griffith Park has not recorded one in a century.

A small percentage of chaparral fires are the work of arsonists. In the 1960s and '70s, a man named Pat Ruis claimed to have started some 2,000 wildfires, though a confession like that is almost rarer than lightning on the Hollywood sign. After a 70,000-acre blaze in Ventura County, arson investigators said that many of the tips they received came from women who were mad at their boyfriends that day. The motives of arsonists can be as mysterious as their ignition patterns. In San Diego County, as an example, one arsonist lit a fire on the same curve of the same road within the same week for 12 consecutive years. He was never caught, so the curve's fascination remains unexplained. Sometimes, arsonists feed off each other's behavior. For instance, between 1970 and 2003, whenever one or more fires were burning in a Southern California national forest, there was a nearly one-in-four chance that another forest would light up. Most depressing, however, are firefighters, or would-be firefighters, who involve themselves in arson for work or recognition. In 1997, a young man named Steve Robles, who visited public schools dressed as Sparky the Fire Dog during fire safety presentations, was linked to seven blazes. Robles had attempted unsuccessfully to become a firefighter. But he did appear at fires he ignited to help evacuate residents. He eventually received a prison sentence of 18 years. Throughout his incarceration his mother is keeping a candle burning for him 24 hours a day on her fireplace mantel.

Most fires are accidentally lit. The 280,000-acre Cedar Fire was caused by a lost hunter who set a signal fire for help. Yet 95 percent of wildfires never grow larger than ten acres in size . No systematic study has been conducted that might explain why the remaining 5 percent grow larger. One thing, though, is known for certain: There is a direct link between population and fire starts. Jon Keeley, a research ecologist with the U.S. Geological Service, discovered that as the population of Southern California doubled, and then doubled again, during the last century, fire starts increased at nearly the same rate. More ignitions led to more acreage going up in flames. "What this means today," says Keeley, "is that something like 20 to 30 percent of Southern California burns every decade. So the equivalent of the entire region burns over every 40 years or so."

L.A. burns with such efficiency because of the development patterns of its outlying real estate, which, in fire science terminology, are called either "interface," where housing abuts wildland, or "intermix," where homes are built inside wildland. The Hollywood Hills have been called the intermix of all intermixes. There is an odd correlation between the populations in intermixes and fire ignitions. A study this year in the journal Ecological Applications found that an intermix landscape with 45 individuals living in a square kilometer will experience a higher number of wildfire ignitions than any other landscape. Raise or lower the population per square kilometer, and ignitions drop off . This is because, theoretically, increased population and housing density leave less landscape to ignite. On the other hand, a population below 45 means there are fewer people to potentially start fires.

Griffith Park is 4,210 acres of wilderness surrounded by development—an "occluded interface" filled with bobcat, deer, raccoon, fox, coyote, skunk, and at least one mountain lion. Its highest mountain, at 1,870 feet, is Cahuenga Peak. There are more than a dozen canyon systems in the park, the largest of them being Coolidge, Vermont, Royce, and Fern. Springwater flows year-round out of Western Canyon. It is the only wilderness of its type in the nation, centered as it is in a city. On average it sees 30 ignitions a year, most caused by trash fires, car fires, cigarette butts, or the cooking fires of transients. Griffith Park, however, is also a historically popular cruising area for gay men seeking anonymous sexual encounters. Twenty percent of fires in the park are attributed to gay men waiting off trails and smoking. Two canyons are the most popular for cruising. One is Western Canyon, home of Fern Dell. The other is Aberdeen.

Flames nearing 50 feet in height were now charging up Aberdeen's chaparral and scrub oak, closing in on William Heritier and Brian Walker at the bridge. The once-lazy bowl fire was about to slop over the canyon's crest, blow through the two firefighters like they were twin stands of manzanita, and spill into the northern park. Gomez, positioned on Vista del Valle, was on Tactical Channel 11, calling for helicopters: "I need water drops on E35's location, now!"

But it was too late. The fire slammed into the saddle under the Aberdeen bridge. Heritier and Walker, clutching their hoses, retreated down the rockfall and onto the pavement where Gomez stood. Nearly a dozen engines were idling on Vista del Valle, a midslope road cut along the higher banks of Fern Canyon that overlook the park's merry-go-round. In Fern, Gomez saw that there wasn't just one fire, as LaBonge had told Fry. There were a half-dozen fires scattered across the undulating slopes of Fern's five gullies. Gomez had fire above him, on Aberdeen's ridgeline, and fire sweeping up from below. His men were in one of the most life-threatening positions an engine crew can find itself in.

Gomez scanned the constellation of fires surrounding him. This wasn't the work of an arsonist. Aberdeen was somehow lighting up the park.

* * * * *

How did it happen? How did Aberdeen jump from being among the 95 percent of fires that are contained before they reach ten acres and into that small category that take off? After all, two battalion chiefs with a combined experience of 56 years had concluded that the blaze would be extinguished within a few hours. Why did Aberdeen explode?

The origins of how we understand wildfire's behavior, why it creeps over one landscape and sprints across another, date back to 1972, when a research engineer named Richard Rothermel at Missoula, Montana's Northern Forest Fire Laboratory wrote out a series of 13 mathematical equations. Rothermel had served in the U.S. Air Force in the '50s as a special weapons aircraft development officer, and in the early '60s, he worked at GE on aircraft nuclear propulsion. In Missoula, Rothermel divided North American foliage into 13 fuel models. There was a grass group, a shrub group, a timber group, and so on. The equations related fuel type to wind speed and slope gradient. If you wanted to know how long it would take, say, a grass-and-timber fire pushed by 15 mph winds to climb a hillside with a 22 percent gradient, you plugged the variables into a Rothermel equation and out came the answer.

Simulation models like BEHAVE , based on Rothermel's 13 equations, were created in the 1990s. Ideally, they would help predict a fire's movement using a computer. But there were drawbacks to Rothermel's early work. To begin with, far more than 13 fuel types exist in North America. Fire scientists improved on Rothermel and came up with equations for as many as 40 fuel types. Limitations still persisted: For example, there is no fuel model for "18-foot chaparral mixed with Spanish-style architecture." In addition, Rothermel's equations couldn't account for a fire that moved from grass to timber to shrub, or from a 20 percent gradient to a 70 percent gradient.

More adjustments were made, but no amount of tinkering could address Rothermel's inability to factor in weather conditions. "Everyone acknowledges that Rothermel's equations are the guts of fire-spread modeling today," says Max Moritz, a biogeographer in UC Berkeley's Department of Environmental Science, Policy & Management. "And they do an okay job modeling fire behavior. But they break down under extreme conditions. They're not designed to take in everchanging weather and topography elements, atmospheric feedback and turbulence—all these noisy stochastic elements that make fire behavior really hard to predict. There are some pretty sophisticated models out there, but we don't have the Holy Grail of real-time modeling yet. With the amount of computer processing needed to model real-time fire, it might take 24 hours to model just 1 hour of a wildfire's behavior."

Aberdeen Canyon faces southwest. Slopes with southwest aspects receive more sunlight and less moisture than slopes facing any other direction, leaving their chaparral drier and increasingly susceptible to fire. Aberdeen also sits in alignment with midafternoon winds that blow in off the Pacific each day and channel along the canyon's walls. When Saxelby first saw the fire, Aberdeen's daily winds had not yet hit. A slope fire with no wind will die down when it reaches a canyon's ridgeline. But a fire driven by a wind of 7 mph or more has the potential to top the ridge. By 2 p.m. winds of 12 mph were blowing into Aberdeen, moving the fire upward through 15-foot-high chaparral. If those conditions had stayed static—if wind speed, fuel load, and slope gradient remained unchanged—Rothermel's equations could have predicted the fire's behavior.

But winds and slope were intensifying the fire's energy release, helping Aberdeen create its own weather—a convection funnel that sucked flaming chaparral brands into the air. This was the dead plant material left over from the February freeze that had worried Tracy Pansini. At the canyon's lip those embers shot out in every direction due to Aberdeen's random wind patterns. "I've been on top of that canyon," says Saxelby, "and the winds are blowing northward on one side of the bridge and south on the other side." Aberdeen, it turned out, was the perfectly designed fire maker that a math equation couldn't explain. "That mushroom cloud I first saw on Sunset Boulevard," says Gomez, "picked up all those embers and then dropped them across the park. The chaparral moisture was so low that wherever an ember landed, a spot fire began."

The spotting activity that drove the blaze from then on had not been witnessed before by the LAFD. "On most fires," says Patrick Shanley, who teaches fire behavior and who took over Alpha by midafternoon, "you see spotting 100 feet in advance. On this fire, everything was outside the envelope. I saw a spot fire 1,600 feet out, off the wind line, in shaded brush."

"In 27 years," says Saxelby, "I have never seen what I saw that day. I saw a spot fire a half mile out. How cool was that ember when it finally landed, and how dry did the brush have to be for it to ignite? That's why that friggin' thing spotted around Gomez."

Zulu's position was terrible. A midslope road puts you between the fire below and what that fire wants: the unburned vegetation above you. On October 3, 1933, hundreds of L.A. County welfare relief workers were called into a similar Griffith Park location to fight a small canyon brushfire. Within the hour the fire ran over them, killing 29. It was the largest loss of life in a wildfire fight in U.S. history. Some bodies, so badly burned, were identified by belongings that had been piled in apple crates: a high school class ring, a Ford ignition key, a human foot.

In 1996, in Calabasas, four LAFD and three Glendale firefighters caught on a midslope road were burned. "We were lucky on that," says Shanley, "because we could have killed 15 guys on that fire." In 1998, four LAFD members were burned on a midslope road, this time in Chatsworth. More recently, on October 26 of last year, it was the crew of Engine 57 fighting the Esperanza Fire that were in a location similar to Zulu's. The day before the 26th, a storm moving from Idaho into Utah had formed a high-pressure cell over Nevada, creating Santa Ana conditions in Riverside. By 6 a.m. a wind-fed fire was climbing a dry drainage channel toward the lightly populated Twin Pines area. Engine 57 took up its defensive position at 15400 Gorgonio View Road, an octagonal cinder-block house that sat at the top of the glowing drainage. In 2002, a California Department of Forestry and Fire study of Twin Pines identified 15400 Gorgonio View Road as "nondefensible." Nevertheless, Engine 57 remained at the address.

At 7:15 a.m. radio contact with 57's crew was lost. At 8 a.m., after the fire blew through, three deceased firefighters and two others who would soon die were found by arriving crews. The engine cab had exploded. Fiberglass in the blanketlike fire shelters carried by the men had melted away, indicating ground temperatures had reached at least 1,610 degrees.

* * * * *

On Vista del Valle, a Griffith Park ranger named Patrick Joyce peered up and watched as a helicopter cut in low over the ridgeline and pulled back, letting loose a veil of water that cascaded onto Fern Canyon. Joyce has fine blond hair, the sun-worn face of a southwest aspect, and knows every curve of Griffith Park's 52 miles of back roads—most of which he will corner in his truck with alarming intimacy and speed. In a few hours he would help lead an attack on the fire's northwest flank. But on this bend of Vista del Valle, Joyce was trapped under a heat transfer. Wind blowing through flaring scrub oak on the southern ridge was streaming embers into the eucalyptus trees that lined the road's northeast side. The eucalyptus, in turn, were detonating one by one, sending their own embers into Fern Canyon's chaparral.

Fern had not burned in 50 years. In its ravines sat four to five tons of fuel per acre, every ignited acre the energy equivalent of ten sticks of dynamite. "Up in the trees, the fire sounded like a freight train," says Joyce. "But in Fern, when the fire roared up the drainages, it sounded like a series of 737s taking off over you." The jets may not have been literal, but the exhaust was. "It got to the point," says Joyce, "where you could not drive a vehicle out of there. It was that smoky. You couldn't see. But you had to be aware, because the eucalyptus limbs—huge limbs—began cracking off one by one, dropping unseen through the smoke. You could hear them crashing all around you."

A person cannot stand within 50 feet of a burning eucalyptus, so unbearable is the heat. As a rule of thumb, fire agencies retreat when flame heights are above a man's head. Gomez had realized by now, however, that he and his dozen engine companies were stuck beneath the burning Australians . "I couldn't go down Fern Canyon," he says. "I couldn't go up. We couldn't back out because there were too many engines arriving behind us. And I wasn't going forward either, because the smoke was so thick that if you were standing right in front of me, I could not see you."

The lee-side intensity of a fire's heat can be four times what it is at the anchor point. In that environment, brush will ignite 400 feet from the nearest flame. The heat pouring off Fern's lee-side was shearing over Vista del Valle in waves. Zulu's safety zone was a long pocket created by the road cut on the mountainside. "If you held your hand up in the air on Vista," says Gomez, "you would have reached into that heat and burned it." So Gomez walked in a half crouch as he moved from engine to engine. Sometimes he would just drop flat onto the pavement. All the firefighters on Vista del Valle were either crouching or diving onto the blacktop. They were negotiating a shifting, invisible atmosphere of hot gases. During San Diego's recent Pechanga Fire, two firemen were burned when a stream of hot gases coursed across a plateau. A supervisor standing 20 feet away was unharmed. "That's how it was on Vista," says Gomez. "If you moved ten feet the wrong way, you would burn." At one point Gomez spied a firefighter's protective neck webbing on fire. He put it out with his hands.

Below Gomez, Varela was leading the fight at the merry-go-round. "It was burning right down to the parking lot," he says. "I tried to walk some 'dozers up the trail to cut a firebreak, and they got run right out—the fire was coming at them too fast."

For the moment, it seemed, Griffith Park had become a helicopter fire. Glenn Smith, flying in Helco 6, the air fleet's command ship, was orbiting the park, directing the traffic flow of four choppers that were ferrying 360-gallon water loads and two larger Firehawk helicopters that could drop nearly three times as much. "Our goal became trying to control the spot fires," says Smith, "but there were just too many of them—dozens and dozens of them—and every engine crew seemed to be calling for priority drops. One spot we must have got 50 calls on."

In the parlance of aviation, Griffith Park is known as a "wire environment." Electrical towers march in concert over the park's hilltops, stringing links of power line that hang 60 feet above the ground. Sixty feet is about the height a chopper wants to level off at for a successful water drop. Scott Bowman, who'd flown out of Van Nuys Airport just minutes behind Smith, was making drops on Fern Canyon. "There were a lot of wires in that canyon," he says, "and when the smoke laid down, you couldn't see them. For a while it seemed like the fire was only burning where the wires were."

At 2:30 p.m. Helco 6 radioed operations to "advise ordering air tankers to help prevent spotting." There was no response. At 2:36, another request: "Have you considered air tankers?" and then another at 2:41. No response. "I wanted to paint some retardant lines between the spotting and the houses," says Smith, but a reply likely never came that afternoon because no one at operations could wrap their head around what Smith was asking for. The last time an airplane had been called in to fight a fire in the City of L.A. was November 6, 1961, when chaparral-fed flames burned through Bel-Air.

* * * * *

When Max Moritz was a grad student at UC Santa Barbara in the '90s, he began taking notice of a large stand of chaparral near the campus. It ran along the western end of the Santa Ynez Mountains, bordered by the Santa Ynez River to the north and the coastal plains of Santa Barbara to the south, and unlike other chaparral stands in the region, there was something curious about this one: It never seemed to burn.

"The hypothesis has always been," says Moritz, "that Southern California is plagued by these huge horrible fires because of fire suppression. The old fuels build up, and we get infernos. Well, for a whole bunch of reasons that theory didn't seem to hold true for areas where I started looking." Fire hadn't been seen in parts of Moritz's stand for more than 60 years. He wanted to know why.

Few ecosystems are more maligned than Southern California's chaparral. Known as elfin wood a century ago, the miscellany of drought-hardy plants like manzanita, scrub oak, chamise, and ceanothus has undergone an anthropomorphic shift in 100 years of wildfires, losing its elfin softness. Chaparral plants are filled with extracts that ignite easily, their leaves glossy with oils that the plants produce during hot weather to seal in moisture and that burn intensely. Thus, chaparral has been fervently described as flora that "needs to burn" or, worse, "loves to burn." Poor chaparral: from elf to nympho in a century's time.

The loudest defender chaparral has is a man named Rick Halsey, a former San Diego School District Teacher of the Year who quit his job to create the California Chaparral Institute, a small operation that he runs out of the basement of his home in Escondido. "I'd been a public school science teacher for years," says Halsey, "and I was teaching this stuff that didn't make any sense to me." What he had been teaching was the generally accepted, if unproven, theory that frequent unsuppressed fires would burn away older plant matter, making room for younger chaparral stands that would be better at resisting flames. "I mean, okay, it made sense on the surface—'We suppress fires, and because of that we've created an unnatural buildup of fuels. Old chaparral becomes decadent and needs to burn. And so we get big, unstoppable fires.' But then," says Halsey, "I thought, 'Wait a minute. Almost every fire we get is human caused, like 98 percent of them. Lightning isn't doing it. So if the natural state of chaparral is that it needs to burn, then what was lighting the fires before we all got here?' "

In March 1983, Science magazine published a paper by a UC Riverside professor named Richard Minnich that off ered the most valid argument to date that chaparral's natural state was to burn with frequency in relatively small fires. Minnich had created a map using Landsat satellite imagery to plot fires that had occurred between 1972 and 1980 in the border region around Southern California and northern Baja California. On the border's Mexico side, where there was minimal fire suppression, a pattern of small fires appeared obvious even to the untrained eye. On the California side, where fire suppression existed, far more acreage looked as if it had burned. Minnich's map made intuitive sense. If we stop suppressing chaparral fires, he concluded, the landscape would return to its organic balance. Fire would create a mosaic of younger, healthier chaparral stands. The relative moisture found in those younger stands would, in turn, help suppress future fires and lead to a pattern of smaller and smaller fires.

There was at least one problem with Minnich's paper: His map included two huge California fires that had occurred before 1972. If you removed them, the fire patterns in the two regions looked remarkably similar. There also seemed to be intellectual transference at work. By the 1980s, it was accepted in forestry science that fire suppression in some conifer forests led to infernos. Fallen limbs accumulated on the forest floor, and when grass ignited, those dead limbs acted like ladders, walking the flames up into the crowns of the trees. "But a chaparral landscape," says Moritz, "is like a conifer forest laid out on its side. It's all crown. So you don't need to worry about the vertical continuity of ladder fuels leading fire into the crown."

Moritz, with the help of the USGS's Jon Keeley, studied ten large areas of chaparral that ranged from Monterey County to the Mexican border. One area was the Santa Ynez stand. The histories of fires, some going back as far as 85 years, were included in each region studied. "I computed hundreds and hundreds of fires," says Moritz. "What I found was that age doesn't have an eff ect on whether something will burn or not. Big fires burn through all ages of chaparral. I found fires that burned equally through chunks of chaparral that had burned 2, 12, or 19 years earlier."

Minnich's Landsat map had suggested that younger stands of chaparral could halt a moving fire. "Under moderate weather conditions," says Keeley, "what determines if a fire will spread is the proportion of dead fuel to live fuel. But if you have wind, it doesn't matter how old the chaparral is." In the winddriven fires that beset Southern California, all burns equally. "Except," says Moritz, "in the Santa Ynez region, where the opposite was true. There, fuel age did have an effect on the probabilities of fire. Why? It's not windy. It's a landscape that seems to be sheltered from the Santa Anas."

We live today in a firescape of our own making, one forged from human ignition. Keeley believes that thousands of years ago, before humans, Southern California's landscape might have taken a hundred years to burn over.

"The question," says Rick Halsey, "gets back to, 'How often did the huge fires occur before us?' We don't know. There's no record. But our fires in the fall are not natural. They are all man-made."

* * * * *

By 4 p.m. Rueda had assumed incident command of the fire. The operation's digs had considerably improved since Fry first unfolded his Suburban's tailgate. Command 3—termed a "Communication Command Control Module"—had been delivered and was under full steam. Command 3 is an awe-inspiring piece of disaster-management porn . In shape and size, the state-of-the-art trailer vaguely resembles a freezer Grendel might have kept out back, refitted with a push-button hydraulic lift system, ten swivel chairs, 12 UHF and VHF radios, 20 TV screens, twin DVD and VHS recorders, an Internet satellite dish, a pneumatically powered 50-foot telescoping aerial camera with weather-monitoring technology, and a black 120-volt Panasonic auto-stop electric pencil sharpener with waste-elimination capabilities. "Command 3 is nice," says Rueda, "but the people who come with it are nicer."

Rueda grew up in El Sereno, the child of an immigrant Mexican artist who put his three sons through Catholic school. He is a tactful man, observant and reflective and occasionally taciturn, a chief who is considered distant by some in the LAFD and just hard to read by others. On the 18th floor of City Hall East he can seem the superego to Varela's id.

Rueda came on the job two years before Dorothy Mae . Dorothy Mae was an aged fourstory apartment house on the corner of Sunset and Figueroa. On September 4, 1982, when a fire started in the second-floor hallway, 170 people were living in Dorothy Mae's 43 apartments. "Twenty-four people died in Dorothy Mae," says Rueda. "I started out in the MacArthur Park area, and large apartment fires like that were then very common. This was before sprinkler systems were required in old buildings. That's how I began my career—high loss of life and structure fires."

Dorothy Mae prompted city regulations for the mandatory retrofitting of older apartment housing with sprinkler systems. Rueda moved on to Watts, where the structures kept going up in flames. "Vacant houses would burn quite often," he says. "Now that they're worth money, you don't see as many fires, but all sorts of things happen in South L.A.—commercial fires, shootings, traffic accidents. I liked it down there." In 1998, he was promoted to deputy chief and in 2003 transferred to his current post. He says that the notion of a strategic view of fighting wildfire didn't occur to him until he reached a chief officer role because "as a firefighter you are so focused on laying line and the task at hand, you don't have time to step back and see the larger picture."

In Command 3 the picture, as viewed on 20 separate screens, was beginning to come into focus. Miscommunication had filled the period when the blaze moved from being a ten-acre burn to a parkwide forest fire. This is common in wildfires. A USDA Forest Service report concerning Southern California's devastating 2003 fire season found that most fire agencies had experienced communication breakdowns when wildfires shifted from small incidents to conflagrations. The efforts of operations personnel to stay on top of the fire, and the responses in the field, could not keep pace. At the Greek, as Aberdeen blew up, Fry had thought Alpha and Zulu were anchoring the fire, when in fact they were above its path. Poor field reporting was coming in from Alpha. At 2:30 p.m. Fry got on Tac 11 and demanded, "Division A , what I need you to do is tell me exactly where you are and start running that division." For some time Fry was unsure which canyon the divisions were even in. (By 3:30 p.m. Shanley would take over command of Alpha.) Most confusing of all was the report relayed to Fry that a commuter on the 210 freeway had seen a fire in La Tuna Canyon and that LAFD dispatch was going to start moving his resources out of Griffith Park and into the MTZ. "Twenty-five engine companies were being called up for the new La Tuna blaze," says Fry, "which the Valley's division was describing as a huge fire at the time." It turned out that the commuter had been watching Aberdeen's smoke in La Tuna's mouth.

As five o'clock drew near, Rueda faced a new problem. The fire was moving east, west, north, and south at the same time, a tide under the influence of several moons. It was advancing in the direction of the zoo, which had been warned to shelter the elephants and primates, toward the observatory, which had been evacuated like the rest of the park, toward east Los Feliz, which was ready to evacuate, and toward the Greek Theater and Command 3. "In 27 years," says Rueda, "I had been to dozens of fires in Griffith Park under the exact conditions, and each one typically burned up the bowl where it started, and that was it." Now Rueda had 80 engine companies arriving. The fire containment perimeters he was setting up covered an unheard of expanse, from the L.A. River to Mount Hollywood Drive to a ridge overlooking Los Feliz. Alpha and Zulu were being wed to Romeo, Mary, Paul, Oscar, Nancy, and—the designation given to Los Feliz—Yankee. Smith's original request for air tankers was also under serious consideration.

When a piece of Hollywood Boulevard vanished into the ground during subway construction in 1995, the LAFD was first to respond. "We didn't have a manual for sinkholes in Hollywood," says Rueda, who had been on other unusual calls: a deranged man who gained control of a crane and began bashing its giant steel hook into a building, a gas leak in the Miracle Mile that blew up a Ross store and sent persistent fire out of every crack in nearby sidewalks and parking lots. "Whenever I give a lecture, I ask, 'Where is the manual for sinkhole response?' How do you solve a problem you've never seen?"

This was a sinkhole moment. Rueda made the call to County and requested two air tankers carrying retardant loads to be scrambled for Los Feliz. He reflected on the strangeness of his request. Then he made a second call: "I just want to make sure you know I'm not asking for two water tank trucks to show up." County assented—requests were now out as far away as Idaho for Rueda's air fleet. Inside Command 3 the chief looked over the growing fire map. "We need a fire behavior analyst," he said to Fry. "Ask L.A. County to send Drew Smith."

* * * * *

When Fry reached Smith by phone that afternoon, he was in Camarillo, leading a fire behavior calculations course for the Ventura County Fire Department. Smith grew up in Ventura County, as his parents had. Even so, there is something Southern about him that he's unable to explain. For instance, he speaks in what sounds vaguely like a Cajun drawl. He salts his conversation with phrases like "Fire prediction— that there's my bag of chips." "Drew, in my opinion, is one of the most multidynamic fire analyzers out there," says a behavior specialist named Kenney Rogers, who sometimes works with Smith from a trailer he keeps parked in Truth or Consequences, New Mexico. "He continuously thinks outside the box of what a fire is supposed to do. That's why he's so quick and accurate." Smith couldn't get to L.A., he told Fry, so he began working on the fire in Camarillo.

Smith favors the BEHAVE simulation program. "I've seen him run a hundred models on one fire just to figure where it's going," says Rogers, who relays weather data to Smith during fire engagements. But Smith is not fond of math and uses BEHAVE only as a yardstick. "I'm more of what you'd call a ground-truther," he says. "Ground-truthing is looking at what's making the fire tick , what its footprint—or signature—looks like. Usually there are only three signatures a fire will have. It can be fuel dominated, wind dominated, or topography dominated. Sometimes, though, a fire will forge its signature and become dominated by a variety of elements." In July Smith climbed into a helicopter in the little Northern California town of Fort Jones and then flew out over the Humbug Basin to ground-truth a series of fires in its canyons. There were 16 blazes in all. Local crews wanted to know which ones they should attack and which they could leave alone for the time being. Smith hovered over the Humbug for a couple of hours before ground-truthing just two fires he deemed most threatening. The crews went to work.

Every firefighter has a story of a wildfire that forged its signatures so many times, its identity and behavior remained a mystery. For Tracy Pansini it was San Diego's Pechanga Fire. For Matt Gil it was the Pine Fire near Gorman that burned in hugger-mugger fashion across the foothills. For Rueda it was "the Topanga Fire of 2005 that jumped the 118 and then took off in spectacular runs into uncharted territory—all the canyons we didn't have lines around." Smith worked that fire as he does nearly every county wildfire. The more erratic a fire becomes, the more need there is for a behaviorist. At the same time, as a fire grows increasingly erratic, it becomes harder to predict. The simplest blazes to chart are Santa Ana-driven fires. "A Santa Ana is actually what we call a 'static' wind condition," says Smith. "We know the wind is blowing east to west. We know how hard. That's atmospheric stability at its greatest. In three days, I can tell you, that fire will be in Malibu."

The toughest fires to fight are plume dominated, where the convection energy is so great, a mushroom cloud known as a pyrocumulus builds as high as 40,000 feet into the atmosphere. Plume-dominated fires are the most aggressive and unpredictable and the most dangerous for firefighters. "Eventually," says Smith, "when that pyrocumulus reaches 40,000 feet or so, it cools, loses its buoyancy, and then drops in downdrafts 40 miles an hour straight into the canyons, blowing fire everywhere. Any nearby firefighter gets burned." During the first week of June 2002, twin pyrocumuli, fed by the Copper and Wolf fires in the Angeles National and Los Padres National forests, were looming over Southern California. Smith worked on both. The Copper Fire plume was so huge, it could be seen from the International Space Station. "When Copper finally collapsed," says Gil, "it was a historic event. We never burned up so many acres of L.A. County in an hour as we did that day."

In Camarillo, Smith called for weather reports and then helped the LAFD sketch out a behavior prediction for Griffith Park's fire. The wind was going to transition from its southwest direction to a northwest position, which would push it into Los Feliz. Increasing winds of up to 35 mph would give it multiple signatures, making its behavior more erratic. Driven by a wind that strong, the park's fire could easily spot over any retardant dropped by aircraft onto the city.

* * * * *

At 4:30 p.m. Michael Fulmis sat down in his Laguna Niguel home and switched on the TV. He is the second assistant chief in the LAFD's special operations bureau, and he had been out of the office all day in Ontario. Fulmis saw a mile-long fire front moving through Griffith Park, said, "Holy mackerel!" and walked out to his Crown Vic. An hour and a half later, he was still stuck in rush-hour traffic.

At 7:30 p.m. Fulmis finally stepped out of his Ford. Along the way he had picked up Councilman LaBonge. They were above Dante's View, a five-acre terraced garden that had covered a high slope of the park for more than 40 years. The sun had recently set, and the vista would have otherwise been beautiful; the lights of LaBonge's district were blinking on below them. Twelve hours earlier, LaBonge had hiked from his home and rested in the same place, which he felt a strong kinship to. He may be the only councilman in America responsible for a district that is 25 percent city park, and he is very fond of it. Now flames were moving toward Dante's View. LaBonge asked if helicopters could save the garden.

Fulmis reached Rueda on Tac 11. "I'm on the Mount Hollywood fire road," he said, "and we have the fire burning up to a garden here that's very important to the city. Can we get a water drop, or are the birds tied up someplace else?"

"I'll see if they can fit one in," Rueda replied.

A moment later Fry reached Fulmis, explaining the situation in the sky over Los Feliz: "I just spoke with air recon, and they have all their major commitment down to Yankee division for those structures." Fulmis turned to LaBonge and informed him the garden could not be saved. Soon, they watched the flames roll into Dante's View. "That was disappointing to him," says Fulmis, "but he understood the priorities. It was a beautiful garden, and he watched it burn, and you could see the tears in his eyes. He knew there was nothing we could do. But it still broke his heart."

To the northeast of Dante's View, down on Vista del Valle, Gomez and Zulu's men had retreated from their perilous midslope position. More than a dozen engines sat on a knot of asphalt known as McComb Helispot . Brian Walker, who had originally stood guard on Aberdeen's advance, looked up and saw a giant olive green twin-prop air tanker descending from the south onto Los Feliz, trailing an indistinct shadow of falling retardant. "Those wings are barely over the treetops," Walker thought. He was joined by Captain Ken Willahan. Both men noticed that the brush around them was coming alive. Rabbits were retreating to McComb's safety, followed by snakes, and lizards, and then a huge rat. "He'd been sucking gas and didn't look too good," says Walker. Deer were leaping onto Vista del Valle. Willahan remembered attending a fire behavior class in which Shanley had said, "When you see animals start coming out of the brush, hold on." He reached down and grabbed an alligator lizard at his feet and placed it in the pocket of his yellow brush coat for safekeeping.

Then McComb exploded. Flames 125 feet high, on the helispot's south side, rose above the engine crews. "I was looking up," says Gomez, "and I am thinking, 'Wow—the sky is on fire.' It felt like, out of the frying pan and into the fire. We were cut off again."

During the preceding hour Gomez had been trying to rescue Engine 76, whose crew was trapped on a dirt road in a canyon just north of McComb. Engine 76's street tires had lost traction on the dirt road, and Gomez had been on Tac 11 with Fry: "We have one of our rigs that cannot get out, and if they don't get a water drop, they will be at risk." But Glenn Smith, flying Helco 6, had choppers being pulled in every direction: "Guys were screaming and hollering for priority drops all over the park."

Smith reached Fry, saying, "We're aware of 76's request, and we're getting over there now." Gomez lined his engines along McComb's south perimeter, arcing water plumes down on 76 and its four-man crew. He was about to order the crew out by foot when McComb's south side flared into the night sky.

"Fire does its own thing," says Gomez. "When it blew up on the other side of us, it sucked up all the air. That blowup stopped the flames that were advancing on 76. I was thinking, 'Why isn't 76 getting run over?' Well, the other fire ate up all the oxygen, drawing it away from the flames that had been threatening the engine crew." Engine 76 finally regained its traction and withdrew from the canyon.

Shanley, in Alpha, had ridden out a quiet afternoon with his assistant, Daniel Curry. "Everyone to the north of Vista del Valle seemed to be getting the work," he says. "Up to that time it had not been the most exciting job in the world." Around 7 p.m., though, Shanley felt a breeze on his face and shouted, "Wind change!" He raised a small Kestrel wind meter in the air and watched its digital readout. "It was 20 miles an hour out of north by northwest," he says. He radioed the weather shift into operations, reporting, "We may be moving into a bad wind situation."

"Fry came back on," says Shanley, "and said, 'Yeah, we were expecting that.' I'm thinking, 'Well, I wish I'd known.'"

Shanley and Curry drove along Commonwealth near the Roosevelt golf course. "It was a scene out of Vista del Valle again," he says, "but now wind driven, moving very fast, just a wall of orange." He led his engines up Commonwealth and into the park's nursery, a collection of low-slung buildings and ancient hothouses that have sat in a hollow at the top of a thin canyon since the 1930s. For the next several hours, as Shanley's engines held defensive positions, flames—sometimes 50 feet in height—worked their way eastward around the nursery. Varela, hearing of the action near Commonwealth, pulled all ten of his engines off the merry-go-round's perimeter and moved them into the Los Feliz streets. Shifting an entire division, abandoning one section of the fire for another, is not a protocol move. It is rarely seen in wildfire engagements. "I didn't ask anyone," says Varela, "and it was the right move. But I did hear Rueda said, 'What the hell is he doing?'"

At times the smoke at the nursery dropped to just three feet off the ground, and men had to crawl . Deer were running through the rows of king palms and Chinese elms; raccoons and skunks were disappearing into the lantana and African daisies. At some point Shanley realized he was feeling every one of the 58 years of his life. "I decided then," he says, "that I'm getting way too old for this."

The fire was moving on to the eastern border of Los Feliz, where Saxelby and his engine crews waited in the dark.

* * * * *

Fry and Rueda were in Command 3 when they heard Saxelby's request come over the radio from Los Feliz: "I think we need to pull the trigger on evacuating the northern portion of division Y. It has hit the ridge." As Drew Smith had predicted, the fire had jumped the air-dropped retardant lines two ridges away from where Saxelby's crews sat waiting on Shannon Drive. The fire had also jumped a line cut out of the chaparral by crews wielding chain saws and Pulaskis. The retardant-dusted ridge was the evacuation's trigger point, set in advance by Saxelby. Once the fire hit the hilltop above Shannon, which borders the park's eastern edge, it was on the LAFD engines in five minutes.

Fire jumps from chaparral into housing typically by one of two ways. Its embers will blow into attic vents, igniting homes from the inside out. (By chance, during 2003's Old Fire a large number of houses had attic vents that were facing the oncoming Santa Anas, and 993 homes burned down.) Or embers will ignite ornamental vegetation. On Shannon, Saxelby had taken notice of the Italian cypresses. "I hate those things," he says. "They burn like Roman candles—worse than eucalyptus." Saxelby's men sprayed the candles with Barricade, a fire-resistant slurry made of diaper material that is shot onto trees. Once, true to his stoic form, Saxelby stuck his finger into a layer of Barricade that had been under a blowtorch flame for two minutes. "It felt cool to me," he says.

The fire line of the blaze descended onto Shannon at a 45-degree angle, as a surfer would ride down a wave. Power went out in the immediate neighborhood while smoke billowed past. In darkness, the hillsides folded invisibly into the sky, so that flames looked like ribbons falling toward Saxelby and his crews. All night he leapfrogged engines one by one along Shannon's western edge, as the fire scraped past the backyards of milliondollar homes.

"I never had a bad moment on that interface," Saxelby says today. "I never felt uncomfortable." At one point during the fight a shingle roof did catch fire. Two LAFD firefighters, spying the new blaze, crawled up ladders hauling axes and chain saws and cut the flames out. It was the only real estate casualty of the entire day in the Hollywood Hills—a few thousand dollars' worth of roof damage, covered initially by plastic tarp and then, later, by insurance.

* * * * *

By noon the following day the Griffith Park fire was out. Mop-up operations would continue for another two days, as engine crews were slowly drawn down and firefighters returned to their stations. The burned marine booked for starting the fire—it was suspected that he had carelessly discarded a cigarette—was released the next day. He eventually returned to Camp Lejeune, though he remains a suspect in the LAFD's unfinished arson investigation. Officially, the fire's size was set by the city at 817 acres—the largest recorded blaze in the park's history. That number, however, was based on a map that did not factor in the park's rolling topography. If you account for an uneven landscape, says Gomez, the fire's true size was closer to 1,200 acres. No one knows for sure what burned on May 8. No comprehensive study of the park's flora has ever been conducted. But by May 11 a new Griffith Park was revealed to the city—the scorched skin beneath the nowvaporized chaparral. Things long hidden were made visible: catch basin dams that had been forgotten for decades, two beat-up Fords of 1930s vintage found in collapsed repose along separate ravines. "We didn't find any dead bodies," says park ranger Patrick Joyce. "But most of the time you find bodies after the rains, because they've been buried and the erosion uncovers them." Raptors and coyotes began hunting down those animals that, having survived the fire, discovered they had no place to shelter.

The evening of May 8 would turn out to be the LAFD's first successful evacuation in the Hollywood Hills. Three hundred people were pulled from what Mark Saxelby calls "all those goofy streets that run all over the goddamn place in Los Feliz." After years of speaking to community groups about the danger of fire in the Hollywood Hills, Fry had begun to wonder if he was the boy crying wolf. On a good day, it was hard enough getting an ambulance into the hills, the roads were so narrow. Fry had once staged an evacuation drill, but residents walked out, began chatting with the firefighters, and passed on the exodus.

In wildfires people commonly die during evacuation. They drive off roads, or flip their cars, or drop dead from heart attacks. They make bad decisions and run back into homes to rescue pets, as a Malibu couple once did to save a parakeet that knew just one phrase: "Get me out of this damn firetrap." Most people who died in the Cedar Fire did so while trying to evacuate. "After Cedar," says Fry, "I thought, 'What would I say if I had to stand in front of a commission and explain why a hundred people had died?'"

The Hollywood Hills are the most serious hazard in the LAFD's purview—the twisting canyons, the housing density, the terrible road access. "If you had a strong Santa Ana wind blowing that night into the Hollywood Hills," says battalion chief Corey Rose, who was assigned by the LAFD to compile a history of the fire, "God only knows what would have happened. We did have some wind changes on May 8, but that truly wasn't a wind event. If you add a Santa Ana to that, that fire is going to move—it's going to move extremely fast, and you will have spotting way outside the line. The key moment of May 8 was when the fire began to spot outside Aberdeen, because that was the premise of everyone's worries for this year's fire season."

Quietly, commanders will tell you that in a conflagration such as the one Rose and Fry imagine—Griffith Park as the sulfur tip of a matchstick that is the Hollywood Hills—a thousand homes could be destroyed, and success would be not losing a life. On May 8, they say, the Hollywood Hills skirted disaster.