Just before sunrise on Nov. 8, 2018, a power line fell from a wind-worn Pacific Gas & Electric transmission tower and whipped into the structure nestled in the Sierra foothills. An electric arc sent molten metal into the dry vegetation below. It ignited.
Five minutes later, a PG&E employee spotted the fire while driving on a nearby highway and reported it. Within two hours of the sighting, the town of Paradise, seven miles away, sent its first evacuation order, but it was already too late. Within two minutes, flames were reported at the town’s edge. Landing embers quickly ignited dozens of spot fires in town. With only four major roads out of town, the streets quickly gridlocked. Paradise burned.
Sixty-four people died in Paradise during the agonizing seven-hour evacuation. Six of them were found in or next to their cars as they tried to evacuate.
Marc Levine, a state legislator at the time, listened over radio to the horrific scenes of people, stuck in traffic, abandoning their cars to flee on foot.
“It made me think of the people falling from the World Trade Centers on 9/11,” he said. “They were going to be incinerated or they were going to jump. … They knew they would die either way.”
So, Levine wrote legislation requiring California cities and counties to analyze whether their roads could support a quick evacuation during emergencies such as fires, floods and tsunamis. Assembly Bill 747 passed in 2019. Yet, to date, the city of Los Angeles has failed to publicly report such an analysis, while fire safety advocates say L.A. County’s evacuation analysis fails to meet the law’s requirements.
The Times reached out to nearly a dozen city, county and state agencies involved with evacuation planning. All either did not respond to requests for comment, could not to point to an evacuation analysis in line with the state’s guidelines for AB 747 or indicated the responsibility for doing the work lie with other agencies.
“The fact that local government leaders would not do as much as they can to protect human life and safety is just shocking to me,” Levine said.
In January, the streets of Pacific Palisades mirrored the scene that distressed Levine in 2018. Traffic was at a standstill on Sunset Boulevard and Palisades Drive — two of the only routes out of the burning landscape. When a spot fire exploded next to the route, police ran down the street, shouting at evacuees to run for their lives.
Every year, dozens of evacuations are ordered in California, organized and completed without any casualties — or even a news story. In these cases, public safety officials have all the lead time that they need to organize a safe and orderly evacuation before a fire reaches a community.
But it’s the much more dire evacuation scenarios — when the lead time is shorter than the time it takes to evacuate, like in the Palisades — where emergency planning is both most important and often ignored.
“There’s no incentive to ever present an evacuation plan that isn’t very positive,” said Thomas Cova, a professor at the University of Utah who studies wildfire evacuation analysis. “Why would an emergency planner — say some young upstart in an emergency operation center — ever want to present a plan to their colleagues that involves some people burning?”
The chaos of these worst-case scenario evacuations often look nothing like the orderly phased evacuations cities often focus on. Unlike in “blue sky” evacuations, smoke can hinder visibility and cause crashes. Often whole towns must leave at once. Power outages can prevent public safety officials from communicating with residents.
It’s why Marylee Guinon — president of the State Alliance for Firesafe Road Regulations, an advocacy group aimed at protecting and expanding the state’s community fire safety requirements — suspects AB 747 is facing pushback from local governments. “They don’t want data that would tell them that it’s going to be a nine-hour evacuation,” she said.
All the while, the risk of fast-moving fires is growing.
In a 2024 study, researchers from the University of Colorado in Boulder analyzed more than 60,000 fires documented by NASA satellites in the first two decades of the 21st century. By 2020, fires in California were growing, on average, four times faster than they were in 2001.
L.A.’s missing evacuation analyses
AB 747 requires local governments to include their evacuation analyses in the safety element of their general plan — the long-term blueprint for future development of a city or county.
The city of L.A.’s current safety element provides no such analysis. Instead, it simply lists evacuation planning as “ongoing.”
In a statement to The Times, the city’s Planning Department, responsible for writing and revising the general plan, said details of evacuation routes are not made publicly available since “large urban cities such as the City of Los Angeles are high profile targets for terrorist attacks.”
The city did not immediately clarify what legal authority it has to keep the analysis private as California law generally requires safety elements to be public.
“It is a mystery how hiding evacuation route capacity and viability can save lives when community members are fleeing a natural disaster,” Levine said in an email to The Times. “It is a dubious claim that terrorists could possibly be well positioned to take advantage of such a catastrophic situation.”
Meanwhile, the county said it complied through an analysis it included in its 2025 safety element. However, fire safety advocates criticized the county’s analysis as simplistic and failing to adequately determine whether quick and safe evacuations are feasible.
The Governor’s Office of Land Use and Climate Innovation, which provides guidance on state planning laws, recommended that local governments use traffic software to simulate different evacuations to estimate how long they might take. Instead, the county grabbed a list of all roads in unincorporated areas within its borders and listed them as “evacuation routes” so long as they were paved, public and not a dead end.
The intent of the law is “not ‘just list the roads you have,’” Levine said. “So I’m super disappointed that L.A. County is dismissive in this way. You would expect, particularly post this year’s fires, Palisades and Eaton, they would take this far more seriously.”
When pressed on their deviations from the state’s guidance, both the city and county planning departments passed the buck to the California Department of Forestry and Fire Protection, which indicated in assessments that the two departments’ safety elements were compliant.
Cal Fire, however, said that its assessments were nonbinding and that complying with the law falls on the city and county.
Yet, none of the local or state agencies directly responded to an inquiry from The Times asking them to explain the discrepancy between the guidance and the safety elements.
The city and county both have detailed evacuation plans that coordinate how public safety officials in the emergency operations, police and fire departments would orchestrate a mass exodus. However, the analysis of roadway networks to estimate how long those evacuations — even if perfectly orchestrated — may take, is different.
“Historically, fire agencies put forth evacuation plans that are operationally driven,” said retired fire Battalion Chief Doug Flaherty. “They talk about communications. They talk about unified command. … What is missing is an actual detailed, road-by-roadway capacity analysis of the time that it’s going to take for people to safely evacuate the area.”
For Guinon, the lack of follow-through from cities and counties across the state is indicative of a common trend in wildfire legislation.
“Virtually every piece of legislation that I dig into, I find out it was the result of a tragic catastrophe,” she said. “This legislation comes out with really, really clear intent over and over, and then it gets forgotten.”
Understanding the problem is half the solution
Despite the complexity of simulating cars on a computerized network of roads to understand evacuation times, the scientific prowess exists — and the software to do it is widely accessible.
When Flaherty, a Tahoe Basin resident, became frustrated with his area’s lack of movement on the issue, he commissioned an evacuation study through his nonprofit, TahoeCleanAir.org.
The retired fire battalion chief, with 50 years of emergency response planning experience under his belt, partnered with Leo Zlimen, fresh out of UC Berkeley and co-founder of the emergency management software startup Ladris.
“We fell into this wildfire space because everywhere we looked, people were [asking] ‘draw a circle on the map and tell me how long it takes to get those people out,’” Zlimen said. “And it turns out, that’s actually a really complicated problem.”
Ladris’ software simulates realistic fire evacuations. It starts by taking a map of roads in a community and plopping little purple dots at virtually every home. Each one represents a vehicle.
A fire starts on the map. It spreads. The purple dots get orders to flee, and the evacuation starts.
The computer can play out a multihour evacuation in mere seconds, and it can account for an excruciating amount of detail. A roadblock, representing a falling tree or car crash, can stop purple dots from using a portion of the road. Some purple dots, not realizing how dire the situation is, wait an extra few minutes — or hours — to evacuate. The dots even wait their turn at stop signs, crosswalks and traffic lights.
Ladris’ software simulates realistic fire evacuations. It starts by taking a map of roads in a community and plopping little purple dots at virtually every home. Each one represents a vehicle.
(Ladris)
Ladris’ program almost looks like a video game. Officials can test evacuation scenarios far in advance or in real time during an emergency. The company is also working to use artificial intelligence to help quickly configure scenarios so users can almost literally “draw a circle on the map” and get an evacuation time.
Flaherty said his detailed Tahoe Basin study, a comprehensive analysis based on Ladris’ simulations, had a price tag just shy of $100,000 — roughly equivalent to the cost of installing one traffic light in town.
“In the scheme of things, it’s very cost effective and reasonably priced,” he said.
Another piece of software from Old Dominion University — simpler than Ladris’ — is available to the public for free. It takes less than half an hour to set up a simulation in the program, called FLEET (for “Fast Local Emergency Evacuation Times”).
Consequently, it’s been used not only by local governments making fire evacuation plans, but also by Scouting America troops interested in flood hazards and event planners wondering how bad the postgame traffic may be.
Among those using FLEET simulations for evacuation planning: the town of Paradise.
How to shorten a five-hour evacuation
After the Camp fire, Paradise became an inadvertent experiment in how towns can better prepare for evacuations. After the disaster, it won a $199-million federal grant for infrastructure projects designed to rebuild Paradise into a more fire-safe town. Before the fire, the town’s entire yearly budget was around $12 million.
After the Camp fire, Paradise hired a traffic consulting firm that used FLEET. It found an evacuation would take over five hours under perfect conditions while utilizing all traffic lanes. It then used the modeling to understand what could be done to alter traffic flow to reduce that time.
For Paradise — as is the case for many towns — a big problem is traffic bottlenecks: To evacuate, virtually the entire town has to use one of four main roads.
The seemingly most straight-forward solution? Build more roads. However, these projects get complicated fast, said Marc Mattox, Paradise’s public works director and town engineer.
Often the roads that a municipality needs to improve evacuation would have to go through private property — a nonstarter for residents in the proposed path.
Or, it’s simply too costly. Although Paradise has received funds to widen two evacuation routes and connect three dead ends with the rest of town, a new evacuation route out of town would be prohibitively expensive. Mattox estimated such a route, navigating Paradise’s steep ridges and canyons, would cost in excess of half a billion dollars.
So, Paradise has also invested in a much cheaper, yet still effective tool to speed up evacuations: clear communication.
Paradise installed signs all over town that proclaim when residents enter and exit different evacuation zones. The town is also looking into using a different color sign for private or dead-end roads that warn drivers to avoid them, as well as digital signs above key roadways that can display real-time evacuation information.
In Southern California, Malibu — which completed an evacuation analysis after it suffered the Woolsey fire the same day as the Camp fire — has similar plans.
Malibu is adding reflective markers to roadways to reduce the chances of crashes amid thick smoke. For neighborhoods with few evacuation routes and individuals with limited mobility, the city encourages evacuating whenever the National Weather Service warns of dangerous fire weather — well before a possible ignition.
Los Angeles is much bigger than Malibu and Paradise — L.A. has a population of 3.9 million; Paradise’s is just over 9,100. But evacuation experts said it’s no excuse for letting California’s rural towns take the lead on evacuation planning.
Asked whether the sprawling labyrinth of L.A. roads would make doing these analyses more difficult, Zlimen smiled.
“Not really — no,” he said, noting Ladris has completed analyses in the San Francisco Bay Area. “It’s totally possible.”
Guinon hopes the results of evacuation analyses can also help — or force — cities to make more responsible residential development plans in the first place.
“It’s not rocket science,” she said. “Let’s just take on protection of our existing communities and let the chips fall where they may with new development: If it’s unsafe, don’t build it.”
