Evaluation of the Unified Forecast System Air Quality Model (UFS-AQM) Online Air Quality Prediction System During the 2020 US Wildfire Season

In 2020, a record-breaking 10.1 million acres burned in US wildfires. To investigate this prolific fire season's impacts on US air quality, the National Oceanic and Atmospheric Administration (NOAA)'s 72-hr Unified Forecast System Air Quality Model (UFS-AQM) is run once-daily from 15 August–30 September 2020. A meteorological and air quality-based forecast verification is carried out for this period. The 72-hr forecasts of near-surface temperature, moisture, and winds perform quite well against observations in terms of correlation and bias. The UFS-AQM surface ozone and fine particulate matter (PM_2.5) often concur with US Environmental Protection Agency (EPA) AirNow station data, though the model ozone displays a consistent positive bias ranging from 8.5 to 15.4 ppb. Anomalously poor PM_2.5 predictions are identified during extreme wildfire activity in the Pacific Northwest between 13 and 20 September. Here, in addition to periods of widespread cloudiness, smoke is quite thick. As a result, fire activity is obscured from satellite detection. The UFS-AQM thus underestimates wildfire emissions in the Pacific Northwest. The lack of dynamic aerosol–radiation interactions in UFS-AQM coupled with uncertainties in fire emissions leads to a large underestimation of surface PM_2.5 there, along with high temperature and low humidity biases. Finally, satellite-based measurements are also employed to evaluate the UFS-AQM's performance in the deeper troposphere. The Visible Infrared Imaging Radiometer Suite 550-nm aerosol optical depth provides additional insights into the UFS-AQM's handling of smoke transport, while the TROPOspheric Monitoring Instrument suggests UFS-AQM suffers from overpredictions of NO₂ and CO in active fire regions.