Carbon emissions from the 2023 Canadian wildfires

This study addresses how much carbon was emitted by the unprecedented 2023 Canadian wildfires and identifies the climatic drivers and future implications of such events. The authors contextualize the event as historically extreme in burned area and investigate emissions using satellite-constrained inverse modeling. They further assess how hot–dry weather in 2023 contributed to fire activity and examine climate projections indicating that similar conditions may become typical by the 2050s, raising concerns about Canada’s forest carbon sink and national greenhouse gas accounting.

Fire emissions can be monitored using bottom-up approaches (e.g., burned area or fire radiative power) combined with fuel loads and emission factors, and top-down approaches using atmospheric observations. Bottom-up inventories (e.g., GFED, GFAS, QFED) can diverge substantially in estimated emissions for trace gases and aerosols. The study builds on prior work by combining bottom-up inventories with top-down constraints from satellite CO, aiming to capture both flaming and smouldering combustion and reduce uncertainty. Previous studies have linked boreal fire activity to hot–dry conditions and increasing vapor pressure deficit, and have projected increased fire emissions with warming.

- Top-down inversion of CO emissions for May–September 2023 using satellite CO retrievals from TROPOMI. - Three bottom-up fire emission inventories used as priors: GFED4.1, GFAS v1.2, and QFED v2.1. - Combined carbon emissions (CO + CO₂) derived by applying inventory-specific emission factors/ratios to CO, acknowledging variability in CO₂-to-CO emission ratios across databases (7.7–10.8 g CO₂ per gC CO). - Sensitivity analyses indicate inversion results are relatively insensitive to inversion configuration but sensitive to assumed hydroxyl radical (OH) abundances, which control CO lifetime. - Historical top-down emissions (2010–2022) estimated using MOPITT (2010–2021) and TROPOMI (2019–2022) core trials for context and evaluation. - Supplementary sections provide inversion configuration details, performance evaluation, and comparisons with priors.

- Total Canadian fire carbon emissions (May–September 2023): 647 TgC with 5–95% range 570–727 TgC. - Bottom-up inventories for 2023 ranged from 234 to 735 TgC (mean 469 TgC); top-down constraints reduced the range by 69% and increased the mean to 647 TgC. - 2023 fire emissions far exceeded typical Canadian emissions: 2010–2022 averages were 29 TgC (bottom-up) and 121 TgC (top-down). - The 5-month 2023 emissions were >4× Canada’s annual fossil carbon emissions (149 TgC yr⁻¹) and comparable to India’s annual fossil emissions (740 TgC yr⁻¹), with only China, India, and the USA emitting more carbon annually. - Climate anomalies: 2023 was the driest January–September since at least 1980; May–September 2023 was the warmest since at least 1980 with ~100% of forest area above average and ~90% >1.5 °C above the 2003–2022 average. VPD was the third highest since 1980; ~85% of forest area above average and ~54% >1.1 above the 2003–2022 average. - Regional contributions: western Quebec accounted for ~15% of national emissions; the northwestern/Great Slave Lake region contributed ~61%. - Relationship to climate: fire emissions are largest under combined warm–dry conditions; 2023 saw a marked increase in such hot–dry conditions compared to prior years. - Managed lands: 2023 fire CO₂+CO emissions over managed forests estimated at 421 TgC (388–461 TgC), equivalent to ~2.5–3 years of projected 2023 national fossil emissions.

Top-down inversions constrained by satellite CO indicate that 2023 Canadian fires released an exceptional amount of carbon, resolving discrepancies among bottom-up inventories and providing tighter uncertainty bounds. The results link the extreme emissions to widespread hot–dry conditions across Canadian forests, consistent with established relationships between elevated temperature, moisture deficits, VPD, and fire activity. Climate projections (CMIP6, SSP2–4.5) suggest 2050s average temperatures similar to 2023, with only modest precipitation increases, implying continued or increased likelihood of warm–dry fire-favorable conditions. Consequently, increased fire activity may reduce net carbon uptake by Canadian forests, potentially undermining their role as a long-term carbon sink. In Canada’s greenhouse gas accounting, differences between IPCC guidelines and the NGHGI treatment of natural disturbances imply large discrepancies in reported emissions; the 2023 managed-land fire emissions (421 TgC) are substantial relative to national totals and highlight the importance of how disturbances are accounted for in policy and carbon budgets.

The study provides a satellite-constrained top-down estimate that 2023 Canadian wildfires emitted 647 TgC (570–727 TgC) during May–September, exceeding historical norms by a wide margin and rivaling annual fossil emissions of major nations. The extreme emissions were driven by unprecedented hot–dry conditions, which climate projections suggest may become typical by the 2050s under SSP2–4.5. These conditions will likely increase fire activity and diminish the Canadian forest carbon sink, posing challenges for national carbon accounting and climate mitigation. Future research should refine emission factor uncertainties (CO₂/CO ratios), improve representation of OH-driven CO lifetimes, integrate multispecies constraints, and assess ecosystem feedbacks (fuel loads, species composition) affecting long-term fire regimes and carbon dynamics.

- Uncertainty in CO₂-to-CO emission ratios across bottom-up inventories contributes to the propagated uncertainty in total carbon emissions. - Top-down estimates are sensitive to prescribed OH abundances, which determine CO atmospheric lifetime. - Inversion configuration details and some evaluations are provided at coarser resolution and shorter time periods than some bottom-up datasets. - Projecting future fire activity is complicated by ecosystem changes (fuel loads, species composition), adding uncertainty to long-term extrapolations.