Caltech Aerosol Monitoring Site Collects Data in Aftermath of Urban Firestorm

On January 9, approximately a full day after the Eaton fire devastated the foothill area of Altadena, north of the Caltech campus in Pasadena, California, several specialized scientific instruments in a shipping container in Pico Rivera, about 11 miles to the south, detected a spike in atmospheric concentrations, particularly of chlorine and lead—both known to be toxic at low levels. When at their maximum that day, chlorine levels reached about 40 times the normal amount, and lead peaked at more than 100 times the usual level.

It was not a coincidence that the air-analyzing instruments had been operating during and immediately after the fire. In fact, they operate nearly 24/7, 365 days a year.

Since July 2023, Caltech has overseen the Pico Rivera location of what is known as the Atmospheric Science and Chemistry mEasurement NeTwork (ASCENT), a network of 12 such sites. ASCENT, which is funded by the National Science Foundation and managed by Nga Lee (Sally) Ng (PhD '07) of the Georgia Institute of Technology, is designed to monitor particulate matter in the air continuously to get a more complete picture of the changing chemical composition and physical properties of these aerosols. Live data from the site are now available online.

"This network exists because we want to have good real-time, high-resolution characterization of particulate matter," explains Haroula Baliaka (MS '23), a graduate student at Caltech who, along with alumnus Ryan Ward (PhD '25), helped set up the ASCENT instruments in the shipping container co-located at an air quality monitoring station of the South Coast Air Quality Management District (SCAQMD). "When the Canadian fires happened last year, the New York ASCENT site was really important because it received the aged plume a few days later. And right now, we're getting data about the Eaton fire as well," Baliaka says.

During and immediately following the Los Angeles fires, the wind was largely blowing southward, pushing the plumes from the Eaton fire directly in the direction of the Pico Rivera ASCENT station. Therefore, the scientists believe that the data collected there mostly pertain to the Eaton fire.

Although the spikes that the instruments detected in the days after the fires are orders of magnitude higher than normal, Baliaka says it is important to put the data into some context. Lead and chlorine are known as trace elements in the air because they are supposed to stay in the nanograms-per-cubic-meter range. In the days after the fire, they reached levels of micrograms per cubic meter. But Baliaka says, "Even as recently as 2010, the lead concentrations that were regularly measured were higher than what we were seeing after the fires because of leaded gas, for example. The further back you go, the more these higher concentrations were what we considered 'normal.' Thankfully, we are now used to having cleaner air."

Data from the Environmental Protection Agency (EPA) indicate that due to regulations, including the removal of lead from gasoline, atmospheric lead levels decreased by 98 percent between 1980 and 2014.

The Caltech scientists note that the elemental measurements detected at the Pico Rivera site have now returned to normal levels.

A striking feature of the data is that it does not have the typical signature of a wildfire in terms of the elements present, says Richard Flagan, the Irma and Ross McCollum-William H. Corcoran Professor of Chemical Engineering and Environmental Science and Engineering at Caltech and co-principle investigator of the Pico Rivera ASCENT site along with Caltech's John Seinfeld, the Louis E. Nohl Professor of Chemical Engineering, Emeritus.

"Decades of studies have come up with a very clear signature of what a wildfire looks like," Flagan says. For example, potassium, which is critical to the life of plants, is typically used as a tracer of biomass burning. But following the Eaton fire, potassium levels were overshadowed by unexpected elements. "Our data provide a clear signature that this was not a wildfire event. This was a wind-driven urban firestorm, which is very different from the point of view of air pollution," he says.

A key feature of the ASCENT network is that its 12 locations all use the same suite of four instruments under the same set of operating protocols so the data can be compared across the country.

One of those instruments, called a scanning mobility particle sizer (SMPS), was invented by Flagan. It gives a full distribution of the size of detected particles over time. Initially, when the fires started, the SMPS data showed that the concentrations of submicron particles decreased because the high winds were clearing the air. But once the wind settled a bit and air from the fires reached Pico Rivera, the SMPS saw large increases in the concentration and size of particulates.

"We can see that a couple of days after the fires, when the smoke reaches Pico Rivera, we have a huge spike in the total mass of aerosols," says Vine Blankenship, a graduate student at Caltech who has focused on the SMPS data.

"There are clear statistical associations between aerosols—particulate matter smaller than 2.5 microns in diameter, known as PM2.5—and health," says Flagan. He notes that it took 20 years of data collection to go from just the definition of PM2.5 in the late 1970s to the establishment of statistical health associations that were required for the EPA to build PM2.5 into the air-quality standards. "Generally, air-quality measurements don't have the level of detail necessary to make those associations. It's all been data collected by networks built by research teams trying to understand the impacts of air pollution. That's why work like this continues to be important."

The ASCENT data have captured the interest of many air-quality researchers and agencies. Other institutions have already come to the Pico Rivera site to collect data with additional instruments. The SCAQMD is also currently setting up two additional aerosol measurement stations in order to capture additional data points.