People assume a fire investigation begins with a suspicion and looks for proof. It begins with the opposite: a burnt landscape and no idea which end of it came first. The Eaton Fire origin was found by reading the ground backwards, and the report is unusually candid about how hard that was here.
Night one
The investigation started before anyone knew there would be one. On January 6, 2025, the LACoFD Arson Fire Investigation Unit augmented itself with an additional investigator for the following day, because a strong high pressure system over the Great Basin was creating Santa Ana conditions that would likely increase call volume. That extra investigator became the lead investigator on the Eaton Fire.
By the evening of January 7 the Palisades Fire was already burning and the Governor had proclaimed a state of emergency. The alert for a brush fire in Altadena went out over Tablet Command at approximately 1820 hours. The investigator drove from Stevenson Ranch, monitored the command frequency, and reached the incident command post at Farnsworth Park at approximately 1910 hours, assuming the role of lead investigator.
What he found was not investigable. Driving through Altadena he observed fallen trees and electrical wires blocking many streets and multiple blocks of residential and commercial buildings burning. At the first vantage point over Eaton Wash it was dark and smoky with minimal visibility and extreme fire behaviour still running in the wash. He noted the location, planned to return at first light, and then made a choice the report records without comment: considering that the primary focus of the incident was the preservation of life and that resources were limited, he stayed in Altadena to help locate and remove residents trapped in homes and vehicles.
Reading the ground
Origin determination under FI-210 works from macro to micro. Investigators start wide, identify a general origin area, then narrow. The indicators are physical marks that fire leaves on objects and record which way it was travelling: angle of char on trees, sooting and staining on rocks, protection, grass stem fall, freezing, spalling, curling, cupping, and U or V patterns.
The first reads came on January 8 from the back of a resident's house and from the Eaton Canyon trail parking lot: predominantly advancing fire indicators showing the fire moved from the north to the southwest in the Eaton Wash below the bluff, plus macro indicators and lateral indicators on the west facing canyon slopes on the north side of the wash.
Walking north from the trailhead, the team read angle of char on trees indicating advancing fire that progressed from further west on the wash. At the Coyote Canyon trail sign, roughly a third to half a mile in, they picked up the lateral indicators seen from the house. Continuing west, the advancing indicators shifted to a western progression, indicating the fire had descended what they would later learn was Coyote Canyon. Following that back up the spur ridge, the lead investigator observed angle of char on trees, sooting and staining on rocks, and a fire advancing westerly. Every step of that traverse pushed the origin further up the ridge toward the towers.
Two teams
On January 9 the team met at Fire Station 181 and split. The Origin and Cause team, working the ground, was made up of CAL FIRE Battalion Chiefs Gilbert, Green and Wallace, CAL FIRE Captain Milloy, and the lead investigator. The Interview and Intelligence team was CAL FIRE Battalion Chiefs Howerton and Kirkhart, Office of the State Fire Marshal investigators Schroeder and Elder, and LACoFD Investigator Ramirez. A LiDAR team was ordered the same day.
The division matters because it produced two independent lines of evidence. The ground team was reading char patterns up a ridge. The interview team was knocking on doors in Kinneloa Mesa and Eaton Canyon. Neither knew what the other would find, and both converged on the same three towers. Several residents described seeing the fire at the base of the middle towers. Others described seeing an arc or a blueish white light followed by a small fire at the base of the towers. One reported his lights surged at approximately 1600 hours. Electrical outages were reported roughly forty five minutes to an hour before the fire by multiple households.
Flags on a razorback
On January 9 at approximately 1430 hours the team ascended the ridge to begin marking indicators. The colour code is standard: red for advancing, yellow for lateral, blue for backing, green for items of interest.
The terrain refused the standard method. Investigators normally work a Z pattern across a slope. Here, the report states, due to the topography they were forced not to follow the normal process of moving in a Z pattern. Milloy adapted by working along the lateral edge of the burn scar to the east until it was no longer safe to place flags. The spur ridge is narrow, the canyon walls extremely steep, and the right side has eroded into a cliff face near the southeast leg of Tower #1. On the east facing slope the loose soil slid downhill with each step, sending rocks and boulders into the canyon below.
The pattern that emerged is stated cleanly: the overall pattern of flagged indicators encompassed the entirety of the bases of Tower #1 and Tower #2, a large area where the lateral edges narrowed as they moved east. There were indicators of lateral or backing fire on the slope below the two southern legs of Tower #1, with advancing indicators running beneath the tower. The span of Tower #1's legs encompassed the entirety of the top of the spur ridge.
The heel that was not there
The most credible passage in the report is the one that reports a negative.
The team noted no discernible heel where the two narrowing lateral edges meet. In a textbook fire, that convergence is where you expect the origin. Finding nothing there is the kind of gap that a weak report skips.
Instead the report reasons about it in the open. Based on the indicators present, the topography, wind, and the fine fuels, backing indicators may have been blown from the razorback spur ridge by the extremely high winds, or the indicators had been denuded and slid off into the canyon below after the fire moved through. It also explains the multiple advancing runs seen down the western slope: they were caused by rollout into the canyon, followed by slope and wind aligned uphill runs. Rollout is burning material rolling downhill and starting new runs, which is exactly what produces several apparent origins on a steep slope when there is only one.
The team also documented erosion between the fire and their examination: rocks with sooting and staining showing a clean line of delineation, indicating that significant quantities of loose soil had been eroded away in the interval.
Instruments
Beyond flags, the scene was worked with four other tools. A CAL FIRE LiDAR team, named in the report as Nick Labedzki, David Karoly and Nicholas Hancock, documented the scene on January 11. Drone missions on January 13 and January 15 captured video and photographs of each tower, flown by CAL FIRE pilots with the ground team directing the shots. On January 13 the origin area was searched systematically with magnets first and then metal detectors, covering the flat portion of the ridge along and underneath the towers from southwest to northeast, then the slope to the southeast as far as was safe. An aluminium coloured item not attracted to the magnet was flagged green on January 13, and an additional small metallic item was located and flagged on January 14. Both were collected and transferred into the evidence locker. On April 9, 2025 a further surveillance flight captured video of the SCE conductors over the Eaton Canyon Falls area, looking specifically for evidence of arcing, and that footage went to the electrical engineer for analysis.
Holding the scene
Physical evidence in an open canyon degrades and walks away. Crime scene tape went across the trailhead on January 8. Private security was ordered, arrived at 2100 hours that night, and held the trailhead in two shifts until January 16 at approximately 0900 hours.
Two intrusions are recorded. On January 14 two people were seen climbing the hill toward the towers where the team was working, contacted, and identified as reporters from The New York Times. They left without incident. The same afternoon, at approximately 1500 hours, a helicopter known to be operated by SCE hovered at low altitude over the scene while the team examined the ground below Tower #1, flying low enough that the two investigators had to stand beside each other and speak loudly to communicate. They photographed the tail number, pulled up FlightRadar24, and filed a safety communication report with the Federal Aviation Administration.
A related entry sits the day before. On January 13 an investigator arrived at a green water tank on Kinneloa Mesa, a high point with a direct view of the origin area, and found two SCE marked vehicles parked on the service road with two men in yellow safety shirts standing alongside. He remained in the area for several hours, until both vehicles left at approximately 1715 hours, about the same time the Origin and Cause team finished work in the origin area for the day. He never observed them performing any work on SCE equipment. Most of the time both stayed on the roadside overlooking the origin area, occasionally using binoculars to look toward it.
The report records these observations without drawing a conclusion from them, and so do we.