In Fairbanks, Winter Pollution Lives in a Layer Just 20 Meters Thick

On the coldest nights of an Alaskan winter, the city of Fairbanks disappears into its own exhaust. Emissions from idling cars and oil-fired home heaters pool beneath a lid of warm air, and for days at a time the entire downtown breathes a concentrated cocktail of sulfur dioxide, nitrogen oxides and particulate matter. A new study published in Atmospheric Chemistry and Physics has now mapped this invisible prison in unprecedented vertical detail, showing that Fairbanks’ winter smog is not spread through the lower atmosphere but squeezed into a layer often thinner than the height of a fifteen-story building.

The research, led by Jochen Stutz of the University of California, Los Angeles, draws on continuous measurements made during the 2022 Alaskan Layered Pollution and Chemical Analysis, or ALPACA, field campaign. Using a technique called long-path differential optical absorption spectroscopy, the team bounced ultraviolet light across downtown Fairbanks and back from reflector arrays mounted at 73, 115 and 191 meters on Birch Hill, as well as along a low path just 12 to 17 meters above the street. By scanning between these light paths every 30 to 60 minutes, the instrument recorded, hour by hour, how pollutant concentrations changed with altitude throughout the experiment, which ran from late January to late February 2022.

The measurements revealed a strikingly consistent structure. Whenever a surface temperature inversion formed under calm winds, emissions accumulated in a polluted surface layer just 20 to 40 meters deep, occasionally persisting for six days straight. Within that shallow blanket, sulfur dioxide reached mixing ratios of up to 35 parts per billion and nitrogen dioxide up to about 60 parts per billion, while nitrous acid climbed to 2.5 parts per billion and formaldehyde to 7 parts per billion. Above the layer’s sharp upper edge, the air was nearly pristine, with background ozone levels of 30 to 35 parts per billion prevailing aloft even as ozone near the ground collapsed almost to zero.

That ozone collapse is one of the study’s most vivid illustrations of how tightly chemistry and mixing are coupled in the shallow layer. Traffic emits nitric oxide, which reacts rapidly with ozone to form nitrogen dioxide. Because the layer barely exchanges air with the background atmosphere, the reaction consumes nearly all available ozone near the surface, and continued nitric oxide emissions can no longer be converted to nitrogen dioxide once the ozone is gone. The result is a mirror-image pair of profiles: nitrogen dioxide highest at the ground and falling steeply with height, ozone lowest at the ground and rising abruptly across the layer’s top. The team’s model shows the nitric oxide-ozone reaction rate peaking near the top of the polluted layer, precisely where the two gases first meet.

The height of the layer proved to be governed by two familiar meteorological variables: the strength of the temperature inversion and the wind speed. A statistical analysis showed that the surface layer height fell steadily from around 56 meters under weak temperature gradients to roughly 25 meters under the strongest inversions, with the variability shrinking as inversions intensified. Wind speed imposed an even clearer threshold. Below about 3 meters per second, inversions and shallow polluted layers formed reliably; above that speed, shear-induced turbulence destroyed the inversion and horizontal advection swept pollution out of the urban core, driving vertically integrated pollutant columns to nearly zero.

To move beyond description, the researchers built a conceptual model of the system and encoded it in a one-dimensional chemistry and transport model called PACT-1D. The model divides the lowest 200 meters of the atmosphere into 39 grid cells, with 11 of them packed into the lowest 20 meters, allowing it to resolve the fine structure of the layer. Vertical mixing was parameterized using an empirical formulation in which the eddy diffusivity profile depends on the observed layer height, itself derived from a lookup table linking layer height to the measured temperature gradient between 3 and 23 meters. Horizontal exchange with cleaner background air was calculated from wind speed profiles and the roughly 4-kilometer length scale of Fairbanks’ urban core.

The agreement between model and measurement was remarkable. PACT-1D reproduced the abrupt transitions between polluted and clean periods, the near step-like vertical profiles of sulfur dioxide and nitrogen dioxide, and the inverted ozone profiles, with the modeled layer height only about 10 meters higher than observed. The exercise also exposed a practical problem for regulators: the official emissions inventory substantially underestimated nitrogen oxides at extreme cold, forcing the team to scale nitric oxide emissions by a factor of 2.3 and nitrogen dioxide by a factor of 3 to match observations, consistent with recent independent estimates for conditions around minus 20 degrees Celsius.

Perhaps the study’s most consequential finding concerns time. By tracking an inert tracer through the model’s transport scheme, the researchers calculated the effective residence time of pollutants in the surface layer: between 0.2 hours when no layer was present and up to 4 hours during well-developed inversion events, with typical values of 1 to 4 hours. The popular image of an inversion as a sealed trap is therefore misleading. Pollution leaks steadily out of the top of the layer and is carried away by winds aloft, even as fresh emissions pour in from below. The concentration plateaus observed during pollution events represent a dynamic balance between emissions, chemistry and slow export, not a static accumulation.

That residence time also dictates which chemical processes matter. Reactions faster than 1 to 4 hours, such as the ozone-nitric oxide reaction, whose ozone lifetime in the layer was under 15 minutes, run essentially to completion inside the layer. Slower processes, including oxidation of sulfur dioxide and nitrogen dioxide by hydroxyl radicals, with chemical lifetimes exceeding 100 hours, barely proceed at all before the gases are exported. The short residence time likewise limits the formation of secondary aerosol species through multiphase chemistry, because particles are flushed from the layer by the same sluggish transport that removes the gases.

For a city that ranks among the most polluted in the United States during winter, the practical implications are direct. Because surface emissions dominate the pollution burden, mitigation efforts should target them at the source, for example by cutting the sulfur content of heating oil to reduce sulfur dioxide and primary sulfate aerosol. The study also sets a demanding technical bar for air quality modeling: standard models with coarse vertical grids cannot resolve a 20-meter layer, so meaningful simulations of Fairbanks winter air require grid spacing of 1 to 3 meters near the surface and accurate parameterizations of mixing under stable conditions. As high-latitude cities grow and extreme cold persists, the ALPACA measurements offer both a warning and a template, demonstrating that continuous, vertically resolved monitoring can turn a hidden atmospheric structure into a quantifiable, and ultimately manageable, public health problem.

Subject of Research: Vertical distribution of trace gases trapped in wintertime polluted surface layers in Fairbanks, Alaska

Article Title: Vertical distribution of pollution trapped by wintertime surface inversions in Fairbanks, Alaska, during the ALPACA experiment

Article References: Stutz, J., Kuhn, J., Cooperdock, S., Johnson, S., Guo, F., Flynn, J. H., Cesler-Maloney, M., & Simpson, W. R. (2026). Vertical distribution of pollution trapped by wintertime surface inversions in Fairbanks, Alaska, during the ALPACA experiment. Atmospheric Chemistry and Physics, 26(19), 14345-14365. https://doi.org/10.5194/acp-26-14345-2026

Image Credits: AI Generated

DOI: 10.5194/acp-26-14345-2026

Keywords: air pollution, temperature inversion, Fairbanks Alaska, ALPACA experiment, LP-DOAS, boundary layer, nitrogen dioxide, sulfur dioxide, ozone, PACT-1D model, wintertime smog, vertical mixing