NOAA's Next-Generation Air Quality Predictions with the Unified Forecast System

NOAA is developing the next generation air quality (AQ) prediction system for the United States (U.S.) and global aerosol predictions within the Unified Forecast System (UFS) framework to better represent and forecast impacts of wildfires on AQ and impacts of aerosols globally on weather out to subseasonal scales. A new regional UFS weather model is online coupled with chemistry represented by the EPA’s Community Multiscale AQ (CMAQ) modeling system with Carbon Bond VI and AERO6 mechanisms to form this new UFS-AQM system. Wildfire emissions are specified by the NESDIS Regional Hourly Advanced Baseline Imager (ABI) and Visible Infrared Imaging Radiometer Suite (VIIRS) Emissions (RAVE). Anthropogenic emissions are based on EPA’s National Emissions Inventories over CONUS and global inventories elsewhere. Lateral boundary conditions for aerosols are provided by the Global Ensemble Forecast System with the Goddard Chemistry Aerosol Radiation and Transport (GOCART) module (GEFS-Aerosols). Testing is performed over a large regional domain covering CONUS, Alaska and Hawaii. The evaluation includes retrospective periods and near-real-time testing and evaluation. Recent examples indicate much improved representation of wildfire impacts on AQ predictions. Some of the planned refinements for UFS-AQM include better representation of plume rise for wildfire smoke and for point source emissions. A bias correction post-processing procedure is included in UFS-AQM to improve prediction accuracy.

Longer term plans for AQ prediction include increased resolution consistent with the Rapid Refresh Forecast System (RRFS) under development and using aerosol lateral boundary conditions from a 6-way coupled atmosphere - ocean - land - sea-ice - waves - aerosols global UFS system, also under development. Data assimilation of fine particulate matter (PM2.5) observations from AirNow, VIIRS Aerosol Optical Depth (AOD) retrievals and TROPOspheric Monitoring Instrument (TROPOMI) NO2 retrievals into UFS-AQM is under development to constrain pollutant concentrations. Due to challenging computational requirements for UFS-AQM at high resolution, a machine learning emulator is being developed to improve computational efficiency for prediction of chemical transformations and tracer transport.