Caltech scientists have published the first detailed analysis of the atmospheric particles in the smoke plume from the Eaton Fire that devastated much of the town of Altadena in Southern California's San Gabriel Valley and the surrounding areas in January 2025. The results confirm the very different chemical nature of the smoke produced by an urban fire as compared to the plume from a typical wildfire. In the early stages of the fire, when many structures and vehicles were burning, the concentration of toxic metals in the atmosphere spiked dramatically.
This chemical signature, which was measured by an advanced air monitoring station located downwind of the fire, then evolved as the blaze moved from urban areas up into the mountains above Altadena, part of the Angeles National Forest.
The scientists found that compared to wildfire smoke, the urban plume was enriched more than 50-fold in lead, 7-fold in chlorine and bromine, and 3-fold in arsenic. In fact, based on the data, they estimate that at the peak of the Eaton Fire's structural burning, nearly 7 million people in the Los Angeles metropolitan area were exposed to 24-hour lead concentrations 80 times higher than typical values. These results are among the first to characterize the air that residents may be breathing during urban fires and could inform models used to determine emergency response to future urban fires.
A paper describing the work was published on October 28, in the journal Science Advances . The lead author of the study is Haroula Baliaka (PhD '27). (The Eaton Fire burned just a few miles north of the Caltech campus.)
"Urban fire plumes are chemically distinct from those of wildfires. Now that we have these observational data about how they differ, we can go back and adjust our models to more accurately estimate how smoke might affect people," Baliaka says. "Smoke travels; it's not just about the burn area. What does a plume mean for people who are miles away from a fire but receive the smoke? We evacuate for fire, but we don't evacuate for smoke exposure."
Baliaka worked with former Caltech graduate student and co-author Ryan Ward (PhD '25) to set up an air-monitoring station in the city of Pico Rivera, located roughly 11 miles south of Altadena, that gathered data on the smoke plume throughout the Eaton Fire. The station consists of a suite of four advanced analytical instruments that operate nearly 24 hours a day, 365 days a year, collecting data about the chemistry and physical properties of aerosols. Aerosols are the tiny particulate matter suspended in the air that we commonly hear about in air quality reports (though those typically only report on PM2.5—particles that are 2.5 microns or smaller in diameter). Co-located at an air quality monitoring station operated by the South Coast Air Quality Management District, the station is one of 12 around the country that make up the Atmospheric Science and Chemistry mEasurement NeTwork (ASCENT ), funded by the National Science Foundation. ASCENT is managed by Nga Lee (Sally) Ng (PhD '07), Caltech's Elizabeth W. Gilloon Professor of Environmental Science and Engineering and Chemical Engineering and a corresponding author of the new paper.
"ASCENT has the unique capability to measure the chemical components and the size of these different particles in real time," Ng says. "ASCENT is the first network in the US to utilize these modern technologies for continuous aerosol measurements." All ASCENT data are now available to the public through both a long-term archive and near-real-time tools on the network's database .
The standard method for sampling aerosols is to collect a sample on a particle filter over a 24-hour period, every few days, and then to analyze the filters weeks or months later. However, with such a routine, it is impossible to capture concentration spikes that come and go rapidly—within hours, for example. Because ASCENT stations run continuously, the scientists were able to observe the changing chemical nature of the Eaton Fire smoke plume as the fire itself moved up the San Gabriel Mountains, away from developed areas, and transitioned from an urban fire into a wildfire.
In the study, the scientists compared data obtained during two different fires that had been measured by ASCENT stations. The Pico Rivera site provided measurements from the Eaton Fire, and another ASCENT site at Rubidoux, CA, gathered data on the so-called Airport Fire, which burned 24,000 acres of vegetation in Southern California's Riverside County, located to the east of the San Gabriel Valley, in September 2024.
Filling a Critical Gap
"Essentially, this comparison between the two fires indicates how different and chemically distinct an urban fire is," Baliaka explains. That might seem intuitive, given that houses, buildings, and everything within them, as well as vehicles, burn in an urban fire, but there has been a lack of data about the chemicals in smoke from urban fires. In fact, a 2022 report by the National Academies of Sciences, Engineering, and Medicine, titled "The Chemistry of Fires at the Wildland–Urban Interface," underscored the need for data on urban fire smoke plumes. And as demonstrated by recent blazes in Los Angeles—as well as in Northern California (the Camp Fire), Colorado (the Marshall Fire), Hawaii (the Lahaina Fire), and, this August, in Spokane, Washington—fires increasingly involve elements of urban burning at this wildland–urban interface, the zone where development meets vegetation.
As the researchers pored over the ASCENT data from the Eaton and Airport Fires, they noted an important relationship between the concentrations of two specific chemical elements, potassium and lead. Potassium is an essential nutrient that plants naturally take up from the soil. When vegetation and wood burn, some of that potassium is released into smoke, and so, unsurprisingly, the fire plumes from both the wildland and the urban fire showed elevated levels of potassium. But in the Eaton Fire, which began on the evening of January 7, lead concentrations spiked across eastern Los Angeles and the San Gabriel Valley on January 8 and 9. The highest lead values were measured directly downwind of Altadena. The ratio of potassium to lead was at its lowest level (around 5:1, indicating the presence of five times more potassium than lead in the fire plume) on those days. As the fire moved into the mountains, the ratio increased toward wildfire-like values (a ratio of about 50:1 by January 11, and then climbing into the hundreds by January 12).
"What was really amazing is that we were able to capture that movement toward a more traditional wildfire signature within a single fire," Baliaka says.
For context, the scientists note that since the early 1980s, when leaded gasoline was being phased out in Los Angeles, typical concentrations of lead in the Los Angeles basin have declined by more than two orders of magnitude, or by about 99 percent. So while the spike in lead values is dramatic compared to current baselines, it was in line with what was normal before the 1980s.
But they also point out that scientists do not yet know the health effects of the kind of short duration, high-concentration exposure to toxic chemicals that millions of Angelenos were exposed to during the Los Angeles Fires. "Unfortunately, Californians get exposed to fires all the time. What do all of these short-term but elevated air toxins, not just lead, mean for our health?" says Baliaka. "We need to start figuring that out."
Updating Models for Fire Response
A goal for the scientists moving forward is to use the data to update models that predict air quality and the amount of pollutants expected to be released into the atmosphere as a result of both urban fires and wildfires. Such models are critical in emergency fire response for predicting where toxic smoke plumes will travel and in helping determine what actions should be taken for public safety.
Richard Flagan , Caltech's Irma and Ross McCollum-William H. Corcoran Professor of Chemical Engineering and Environmental Science and Engineering, says that the response to the Eaton Fire and to contamination of the debris left in its wake were predicated on the firestorm being a wildfire event. "But the Eaton Fire was an extreme urban fire event," he says. "Perhaps it is time to reconsider fire codes to address the possibility of urban firestorms in densely populated urban areas, and consider the downwind health impacts and excess mortality as the important impacts of such events."
The researchers acknowledge that the strong signature of lead might be unique to the Eaton Fire, given the large number of older homes and structures that burned in Altadena, but they hope that the results provide a baseline for comparison with future urban fires.
Ng emphasizes that there is a need for additional real-time continuous aerosol-monitoring stations around the country to develop a comprehensive understanding of fires at the wildland–urban interface. "The existing monitoring stations provide a vital foundation, and ASCENT shows how we can build on it to modernize this country's air quality measurement infrastructure," she says. "With ASCENT, we are not only seeing how much particulate matter is in the air but what it is made of, and we can watch it as the event unfolds. This kind of information is very useful for tracking the atmospheric chemical evolution of fire smoke, and it helps emergency managers and public health officials better understand what communities are being exposed to in real time."
The paper is titled "Fire at the boundary: airborne toxic metals from the 2025 Los Angeles urban firestorm." Along with Baliaka, Ward, Ng, and Flagan, Caltech authors include graduate student Can Barut; Sina Hasheminassab, a visitor in environmental science and engineering and a research scientist at NASA's Jet Propulsion Laboratory, which Caltech manages for NASA; Francois L. H. Tissot , professor of geochemistry and a Heritage Medical Research Institute Investigator; John H. Seinfeld , Louis E. Nohl Professor of Chemical Engineering, Emeritus; and Paul O. Wennberg , the R. Stanton Avery Professor of Atmospheric Chemistry and Environmental Science and Engineering. Additional authors are Sean M. Raffuse of UC Davis; Bradley Ries, Yumeng Cui, and Roya Bahreini (PhD '05) of UC Riverside; former postdoctoral scholar Dien Wu, now of Colorado State University; and Ruizhe Liu and Armistead G. Russell (PhD '85) of the Georgia Institute of Technology. Baliaka received a scholarship from the Onassis Foundation, and both Baliaka and Ward received support from the Resnick Sustainability Institute at Caltech.