Improve the skill of NOAA's air quality forecasts of ozone and particulate matter (PM2.5), especially high concentrations that result from forest fires

The National Oceanic and Atmospheric Administration awarded a $170,629 cooperative agreement to the University of Colorado Boulder under the Cooperative Institute program (CFDA 11.405) from March 1, 2023 to December 31, 2024. The University will improve NOAA's air quality forecasts of ozone and particulate matter, especially from forest fires, by reducing systematic errors in the new forecast model through analysis of past analogous fire environments. This error correction technique will be applied to both the full new forecast model and a version using machine learning chemistry emulation to greatly reduce computational requirements. Successful completion of this research is expected to provide more accurate predictions of smoke and ozone concentrations resulting from forest fires in terms of timing, location, and intensity.

NOAA is developing a new air quality forecast model, and although IT is expected to be more skillful than the current model, IT will nevertheless have significant systematic errors. The goal of this project is to reduce these systematic errors through an analysis of those errors in past analogous weather and fire environments. This error correction technique will be applied both to the full new forecast model, and to a version that will use a machine learning (artificial intelligence) chemistry emulator that is anticipated will greatly reduce the computational requirements of the air quality model. If successful, this project will provide more accurate predictions of the timing, location, and intensity of smoke and ozone concentrations that result from forest fires.