From forest to front door: Understanding how wildfire spreads through communities

From forest to front door: Understanding how wildfire spreads through communities Lisa Lock Scientific Editor Andrew Zinin Lead Editor As California's population boomed—from 10 million in 1950 to over 40 million today—the number of people living in fire-prone areas multiplied. Over the decades, millions of new homes and commercial buildings sprang up to accommodate the needs of the state's growing population, and many of those structures stand in areas prone to wildfires. As a result, more communities are now in harm's way. The 2025 Palisades and Eaton fires in Los Angeles show exactly how destructive urban wildfires can be. More than 16,000 buildings were destroyed, and 31 people lost their lives. By the time the recovery is complete, the cost of repairing the damage from those wildfires could approach a quarter of a trillion dollars. "We are front and center in the wildfire crisis," says Michael Gollner, associate professor of mechanical engineering at UC Berkeley. "California is beautiful, but everything aligns for fires here. We need to learn to adapt, to limit the damage that wildfires cause. A lot of people live in fire-prone areas, and most of the country's major urban wildfire disasters have happened in California." Despite the thousands of wildfires in California each year, we still don't know that much about them—especially when it comes to how they spread in urban areas. The wildland-urban interface is the zone in which buildings and infrastructure border natural areas. Homes in this zone are at higher risk of burning, but quantifying that risk is challenging. Until recently, the mathematical models used to predict wildfire spread largely ignored these areas. Where a simulated wildfire reached a developed community, the models treated the land as unburnable. Which, of course, it's not. "Understanding fire spread between built structures is a totally different equation than understanding fire spread in wildlands," says Gollner. "It's different physics, and until recently, it was not well enough understood to be integrated into the parameters of mathematical models. We still can't do everything at scale, but we've begun to be able to simulate wildfires in urban areas." In the wildland-urban interface, fires will move freely between natural landscapes and developed lands. And the 2025 Palisades fire did exactly that. What began in the brush of the Santa Monica Mountains swept through residential neighborhoods in Pacific Palisades, before it crossed Mandeville Canyon and burned other parts of Greater Los Angeles. But predicting, and ultimately slowing, this kind of fire behavior remains a complex problem. "To make good decisions about the risks communities face, we need more science," Gollner says. Evolving science for a changing landscape The increase in wildfires tearing through California communities isn't just due to population growth. Historically, much of California burned every 7–15 years, a natural process that thinned forests and removed accumulated ground fuel. "It's actually restorative," says Scott Stephens, a professor in the Rausser College of Natural Resources. These periodic fires created forests that were more fire resistant. Over time, California's forests came to be dominated by fire- and drought-resistant pine trees, which shade the forest floor. But a century of fire suppression policies and logging has changed the composition of our forests. As many of the largest pines were felled, forests grew thicker and less fire-resistant tree species flourished. Now, California's forests are much more flammable than they once were. "The effects have been profound," says Stephens. "The fires we have today would have been unimaginable 50 years ago. In California, we put out more than 98% of wildfires, but the 2% that we can't put out are responsible for burning 95% of the land." And the amount of land that has burned in California is mind-bending: more than 12 million acres since 2013. That's an area more than twice the size of New Jersey. The damage hasn't been confined to forests and chaparral, but has also torn through some of California's largest cities. As wildfires push into urban areas, they behave in ways scientists are only beginning to understand. But Gollner is figuring out how to predict what urban wildfires will do next—by turning fire modeling into a complex, evolving problem. "There are so many things that go into modeling a fire; you need to account for the weather, the fuel chains and the fire itself," Gollner says. "We are working on how to do those calculations for urban areas, but there are so many unknowns and uncertainties. Sometimes, we are just revealing what those uncertainties are. Other times, we are starting to put numbers to them." In research published in 2024, Gollner developed a landscape-scale model for predicting the spread of fire in the wildland-urban interface. Unlike other models that existed at the time, Gollner's addressed the spread of wildfire through both vegetation and buildings. "Other models treated them separately, but ours seamlessly couples the two," says Dwi Marhaendro Jati Purnomo, a post-doctoral scholar in Gollner's lab who led the development of the model. "It's driven by the key mechanisms of fire spread in the wildland-urban interface: direct flame contact, radiation and firebrands—airborne pieces of burning material that are responsible for igniting many of the buildings that burn in wildfires. Our modeling framework also incorporates heat flux—the amount of heat that travels through materials. That allows us to quantify the hazard to nearby houses." Gollner and Purnomo tested their model using data from two highly destructive events from California's recent past—the 2017 Tubbs and Thomas fires. And they were able to fine-tune the model to make predictions with a high degree of accuracy. Their model predicted the extent of the fires with more than 85% accuracy and the number of damaged houses with roughly 70% accuracy. Importantly, the researchers found that nearly a third of house ignitions were caused by embers, something ordinary Californians can take meaningful steps to guard against. "There are two basic types of mitigation. The first is defensible space—basically clearing flammable things away from a house and reducing the amount of fuel available to a fire," says Gollner. "The other is to harden a house, to make it more difficult for the building to ignite by changing the windows and siding or by installing ember-resistant vents. These are simple-ish things, but hardening a house can be expensive." Fire-resistant siding is made of non-combustible materials like stucco or metal, and installing it in a new build doesn't cost that much more than traditional wooden siding does. But the costs of replacing the siding on existing houses can be high, especially when you multiply those costs across millions of houses. Gollner's modeling is helping identify in which areas this type of mitigation will have the greatest impact. "That's the billion-dollar question," he says. "We don't have all the tools yet, but we do know that if you did all these things and have good spacing between your house and your neighbor's house, you will be way better protected. But it takes money and time, so there is a question of exactly how much needs to be done. A house right next to the forest? Sure, we need to do that. But how far into a community do these mitigations need to go? We still don't know exactly in which areas these investments will have the biggest payout." Developing a playbook to fight fire in urban areas Gollner's ultimate goal is to craft a playbook for wildfires that communities can use to protect themselves. Right now, there isn't one. But Gollner's work to quantify wildfire risk is giving communities the data they need to set priorities. And one of the communities doing that is Berkeley. "Berkeley is in a fire-prone landscape," he says. "We had major fires in 1991 and 1923, and we have mostly forgotten about them. But the weather conditions that happened during those fires will happen again; we just don't know when those conditions will occur." Even the tiniest spark can generate an inferno, he adds. And whether one does is largely a matter of luck. In the right conditions, a fire can spread rapidly. Right now, there's little to stop a fire from spreading through the tightly packed wood-framed buildings that line Berkeley's narrow streets. "How destructive a fire is depends a lot on the direction and the timing of the winds," Gollner says. "If the winds only last a few hours, the fire might stop, but when a wildfire is raging at peak intensity, there is just not a lot that humans can do about it." When a wildfire enters a city, it sends embers flying everywhere, and they can ignite buildings by the thousands. No fire department in the world can handle a hundred burning buildings, much less a thousand, and no amount of investment in firefighting will allow us to extinguish urban fires on this scale. "There is just no way to tackle this problem with an active approach," says Gollner. "We need to limit the ways fire can spread. If we can do those simple things, we can make a difference." But you need to get people on board, and that isn't always easy. In January 2026, the City of Berkeley implemented an ordinance called Effective Mitigations for Berkeley Ember Resilience (EMBER). It mandates a five-foot vegetation-free buffer zone around houses and also identifies areas at the highest risk from wildfires as priorities for enforcement. Under EMBER, fines for non-compliance can reach up to $500 per day, and some local opponents have argued that EMBER is overly intrusive. But Gollner's research highlights the value of defensible space. In a paper published in 2025, he found that houses with a five-foot, vegetation-free buffer were much less likely to be destroyed in a wildfire. Gollner conducted a machine learning analysis of data collected on the ground after five major historical fires in California's wildland-urban interface—the 2017 Tubbs, 2017 Thomas, 2018 Camp, 2019 Kincade and 2020 Glass fires—and found that houses with a vegetation-free buffer area were much more likely to survive a wildfire. The difference is stark. Just 20% of houses without a buffer survived these wildfires, while 37% of houses with a buffer did. In a major urban wildfire, that 17% difference could add up to hundreds or even thousands of houses saved. "Clearing the vegetation that is closest to the house isn't perfect, but it is one of the cheapest and easiest ways to reduce risk," Gollner says. "The vegetation that's right there has the highest likelihood of igniting a house. But people can be really protective of their plants, and some people are just opposed to doing it." But retired fire chief Dave Winnacker is using the data to convince people of the value in compliance. Winnacker spent about 20 years in the fire service in Fresno, Alameda County and the Moraga-Orinda Fire District. And now, he's returned to his hometown of Berkeley to help the city implement EMBER. "Dr. Gollner's research has really given Berkeley a theoretical grounding to operationalize this initiative," says Winnacker. In the same 2025 paper, Gollner calculated that implementing both house hardening and defensible space could prevent the loss of 52% of all the structures lost in the major urban wildfires he studied. "Fire is opportunistic," Winnacker says. "And it will find a way to spread. But when you take mitigation measures, communities won't burn. In the last 20 years, 80,000 homes in California burned, give or take. Halving that number to 40,000 homes still would not be great. But it sure would be a lot better than 80,000." When more people adopt defensible space and harden their houses, the effect is amplified. Each house that doesn't burn breaks a chain of ignitions—it means fewer houses burn overall. That's a message that Winnacker has taken to the public. "Urban fire is a big macro-phenomenon, and it is hard for people to connect the dots between these micro-mitigations, which can seem inconsequential, and this huge destructive thing," Winnacker says. "But Dr. Gollner's work allows us to do that. It helps us understand the drivers of loss in the urban environment. And from there, we can prioritize mitigations, so that we get the greatest value from a risk reduction standpoint. Otherwise, we would just be guessing." Provided by University of California - Berkeley