The FireWork air quality forecast system with near-real-time biomass burning emissions: Recent developments and evaluation of performance for the 2015 North American wildfire season

ABSTRACT

Environment and Climate Change Canada’s FireWork air quality (AQ) forecast system for North America with near-real-time biomass burning emissions has been running experimentally during the Canadian wildfire season since 2013. The system runs twice per day with model initializations at 00 UTC and 12 UTC, and produces numerical AQ forecast guidance with 48-hr lead time. In this work we describe the FireWork system, which incorporates near-real-time biomass burning emissions based on the Canadian Wildland Fire Information System (CWFIS) as an input to the operational Regional Air Quality Deterministic Prediction System (RAQDPS). To demonstrate the capability of the system we analyzed two forecast periods in 2015 (June 2–July 15, and August 15–31) when fire activity was high, and observed fire-smoke-impacted areas in western Canada and the western United States. Modeled PM_2.5 surface concentrations were compared with surface measurements and benchmarked with results from the operational RAQDPS, which did not consider near-real-time biomass burning emissions. Model performance statistics showed that FireWork outperformed RAQDPS with improvements in forecast hourly PM_2.5 across the region; the results were especially significant for stations near the path of fire plume trajectories. Although the hourly PM_2.5 concentrations predicted by FireWork still displayed bias for areas with active fires for these two periods (mean bias [MB] of –7.3 µg m⁻³ and 3.1 µg m⁻³), it showed better forecast skill than the RAQDPS (MB of –11.7 µg m⁻³ and –5.8 µg m⁻³) and demonstrated a greater ability to capture temporal variability of episodic PM_2.5 events (correlation coefficient values of 0.50 and 0.69 for FireWork compared to 0.03 and 0.11 for RAQDPS). A categorical forecast comparison based on an hourly PM_2.5 threshold of 30 µg m⁻³ also showed improved scores for probability of detection (POD), critical success index (CSI), and false alarm rate (FAR).

Implications: Smoke from wildfires can have a large impact on regional air quality (AQ) and can expose populations to elevated pollution levels. Environment and Climate Change Canada has been producing operational air quality forecasts for all of Canada since 2009 and is now working to include near-real-time wildfire emissions (NRTWE) in its operational AQ forecasting system. An experimental forecast system named FireWork, which includes NRTWE, has been undergoing testing and evaluation since 2013. A performance analysis of FireWork forecasts for the 2015 wildfire season shows that FireWork provides significant improvements to surface PM_2.5 forecasts and valuable guidance to regional forecasters and first responders.

Acknowledgment

The ECCC FireWork team is very grateful to the Natural Resources Canada team behind CWFIS, who provided crucial assistance in the development of the FireWork system. We are particularly grateful to Peter Englefield, who provided useful analyses and guidance about the CWFIS system. We also thank the U.S. Forest Service BlueSky team, who helped us in building the FireWork system, and particularly Drs. Susan O’Neill and Tara Strand, who provided guidance about the BlueSky system and with different analyses, advice, and wildfire emission estimates for the United States. We are indebted as well to other members of ECCC’s Air Quality Modelling Applications Section team, particularly Hugo Landry, who worked on FireWork system optimization, Mourad Sassi, who helped us with the treatment of wildfire emissions, Samuel Gilbert, who helped us to analyze and evaluate model performance, and Jacinthe Racine, who helped us to analyze and visualize model results. Lastly, special thanks are due to Dr. Sylvie Gravel of ECCC’s Air Quality Research Division, for supporting this project, for helping us to analyze FireWork performance, and for providing comments on the paper.

Funding

This work was supported, in part, by the Canadian Safety and Security Program (CSSP-2012-CP-1182), which is managed by Defence Research and Development Canada’s Centre for Security Science.

Supplemental Material

Supplemental data for this article cab be accessed on the publisher’s website.

Additional information

Funding

This work was supported, in part, by the Canadian Safety and Security Program (CSSP-2012-CP-1182), which is managed by Defence Research and Development Canada’s Centre for Security Science.

Notes on contributors

Radenko Pavlovic

Radenko Pavlovic is a meteorologist in environmental applications in the Air Quality Modelling and Applications Section (AQMAS), Canadian Meteorological Centre Operations (CMCO), Environment and Climate Change Canada (ECCC), Montreal, Quebec, Canada.

Jack Chen

Jack Chen is a modeling scientist with the Air Quality Research Division (AQRD) of ECCC in Ottawa, Ontario, Canada.

Kerry Anderson

Kerry Anderson is a fire research scientist with the Canadian Forest Service, Natural Resources Canada, Edmonton, Alberta, Canada.

Michael D. Moran

Michael D. Moran is a research scientist with the AQRD of ECCC in Toronto, Ontario, Canada.

Paul-André Beaulieu

Paul-André Beaulieu is a scientific programmer analyst at the AQMAS, CMCO, ECCC, Montreal, Quebec, Canada.

Didier Davignon

Didier Davignon is acting chief of Data, Performance and Standards, CMCO, ECCC, Montreal, Quebec, Canada.

Sophie Cousineau

Sophie Cousineau is acting chief of the AQMAS, CMCO, ECCC, Montreal, Quebec, Canada.