Rapid transformation of wildfire emissions to harmful background aerosol

The study addresses a major discrepancy between bottom-up estimates of wildfire contributions to PM2.5 in Europe (over 85% of summer emissions) and top-down source apportionment analyses that attribute less than 10% of organic aerosol (OA) to biomass burning in summer. Measurements at a remote site in the eastern Mediterranean (Pertouli, Greece) during July 2022 occurred amidst widespread European wildfires, yet Positive Matrix Factorization (PMF) did not detect a strong fresh biomass burning OA (bbOA) factor, instead identifying highly oxidized secondary OA components. The authors hypothesize that rapid atmospheric aging of biomass burning emissions (evaporation, gas-phase oxidation and recondensation, heterogeneous and aqueous-phase reactions) transforms bbOA into secondary, highly oxidized OA (MO-OOA), erasing conventional chemical fingerprints (e.g., levoglucosan). This process would cause significant underestimation of wildfire contributions to summertime OA and suggests that aged biomass-burning secondary OA (bbSOA) constitutes a large fraction of regional background OA and associated health burdens across Europe.

Prior studies indicate: (a) European source apportionment analyses often report <10% biomass burning contribution to summertime OA, contrasting with emission inventory estimates implying dominant summer PM2.5 contributions from fires; (b) potassium (K+) and AMS m/z 60 and 73 (levoglucosan-related) are common bbOA tracers, but can be misleading after atmospheric aging due to volatility and rapid oxidation of levoglucosan, particularly in summer; (c) previous findings show MO-OOA correlates with K+ in aged fire plumes and that brown carbon undergoes photobleaching with processing; (d) oxidative potential (OP) of OA increases upon atmospheric processing, with oxidized OA exhibiting higher DTT activity than fresh OA in Mediterranean summer conditions. Emission inventories (e.g., IS4FIRES) and modeling frameworks (PMCAMx-SR with VBS) have been developed to simulate bbOA emissions and aging, though substantial inter-inventory uncertainties (factor ≥2) remain.

The study integrated field observations from a remote continental site (Pertouli, Greece) during July 2022 with chemical transport modeling.

• Field measurements: Submicron aerosol composition with an Aerodyne Aerosol Mass Spectrometer (AMS) at 3-min resolution; black carbon (BC) with an Aethalometer; ionic species including K+, Na+, Mg2+, Ca2+ via filter-based analyses; water-soluble brown carbon; oxidative potential (OP) of PM2.5 water-soluble fraction using a semi-automated dithiothreitol (DTT) assay on 16 daily quartz filter samples (reported as DTTa in nmol min−1 m−3 and DTTm in pmol min−1 µg−1). Auxiliary data included HYSPLIT back-trajectories.
• Source apportionment: PMF applied to AMS spectra (2–5 factor solutions explored, 3-factor solution retained: biogenic-OOA, LO-OOA, MO-OOA). Variables with low signal-to-noise were down-weighted; CO2-related variables down-weighted. ME-2 used for sensitivity, constraining a fresh bbOA spectrum (a = 0 for fresh bbPOA estimate; a = 0.5 yielded aged bbOA). Potassium was included in a sensitivity PMF run; 76% of K+ associated with MO-OOA.
• Biomass burning tracers: K+ apportionment to biomass burning by subtracting sea-salt and dust contributions using Na+, Mg2+, and Ca2+ ratios; bbOA estimated via bbOA:K+ ratios from literature (5–100, nominal 20).
• Fire case studies: Two nearby fires examined (Dadia, Greece; Albania), using HYSPLIT to estimate 1–2 day transport, and time-resolved AMS factors, K+, BC, f60, and nitrate to evaluate transformation.
• Chemical transport modeling: PMCAMx-SR over Europe (36×36 km, 14 vertical layers to ~6 km), June 28–July 31, 2022 (first 3 days spin-up). Meteorology from WRF v4.1.5. Fire emissions from IS4FIRES; biogenic emissions from MEGAN v3; marine emissions via O’Dowd/Monahan parameterizations; anthropogenic emissions from TNO inventory. OA represented with the Volatility Basis Set including wildfire-specific volatility distributions. Source-resolved outputs for bbPOA and bbSOA, and long-range transport (LRT) OA.
• Health impact estimation: Linear mortality-response assumption using European estimates (≈300,000 annual deaths at 14 µg m−3 annual PM2.5), yielding 1,800 deaths per µg m−3 per month. Applied to modeled bbSOA (2–3 µg m−3) to estimate premature deaths attributable to wildfire-related OA in summer.

• Rapid aging of biomass burning emissions transforms fresh bbOA to highly oxidized secondary OA (MO-OOA), erasing conventional bbOA fingerprints.
• Source apportionment at the remote site identified three secondary factors: bOOA (23%), LO-OOA (37%), MO-OOA (40%); fresh bbOA was not significant (ME-2 upper bound ≈3% of OA, 0.2 µg m−3). Average OA O:C = 0.85.
• Potassium (PM2.5) averaged 0.2 µg m−3 (biomass burning K+ ≈0.15 µg m−3 after accounting for sea-salt and dust). K+ correlated with MO-OOA (R2 = 0.61, daily), and PMF associated 76% of K+ with MO-OOA.
• Levoglucosan-related fragments were low: f60 ≈0.4% on average (background level), far below fresh bbOA signatures (up to 6%).
• Black carbon averaged 0.42 µg m−3; wood burning contributed ~40% of BC overall and >50% on high-K+ days (R2 = 0.58 between K+ and biomass-burning BC on those days).
• Modeling (PMCAMx-SR) at the site: bbSOA ≈40% and bbPOA ≈4% of total OA. Across Europe, average bbSOA ≈2 µg m−3 in July 2022; model likely conservative (underpredicts OA at Pertouli).
• Nearby fire case studies (1–2 day transport): MO-OOA dominated OA increases (e.g., +4 µg m−3 during Dadia, +3.1 µg m−3 during Albania), with MO-OOA about 10× and 6× bbPOA during the two events, respectively, indicating >80% of bbOA was secondary within 1–2 days. MO-OOA correlated strongly with K+ (R2 = 0.74 hourly), m/z 60 (R2 = 0.9 during fire periods vs 0.33 monthly), and nitrate (R2 = 0.84 during fire periods vs 0.17 monthly).
• Origin of bbSOA at site included long-range transport from Iberian Peninsula fires (thousands of km), plus Ukraine, Italy, and Balkans.
• Using K+ and literature bbOA:K+ ratios (nominal 20; range 5–100), bbOA at the site estimated ~3 µg m−3 (0.75–15 µg m−3), consistent with PMCAMx-SR (≈2 µg m−3); best estimate 2–3 µg m−3.
• PMCAMx-SR over Europe indicated 55% of OA due to fires within the domain and an additional 21% due to long-range transport.
• Oxidative potential: DTTm averaged 84 ± 27 pmol min−1 µg−1, higher than typical oxidized OA values in Mediterranean summer; DTTa averaged 0.20 ± 0.03 nmol min−1 m−3. Multiple linear regression implicated MO-OOA (and K+, Cl−) as major contributors to OP, with little contribution from biogenic OOA.
• Health impacts: Estimated wildfire-related bbSOA exposure of 2–3 µg m−3 across Europe corresponds to ~3,600–5,400 premature deaths per summer month, or ~10,800–16,200 per summer, implying 15–22% of PM2.5-attributable summer deaths are associated with wildfires.
• Overall, wildfire contributions to summertime OA in Europe have been underestimated by a factor of ≈4–7; wildfires were likely responsible for about half of total OA in Europe during July 2022.

The findings support the hypothesis that rapid atmospheric processing converts fresh biomass burning OA into highly oxidized secondary OA, leading to a loss of traditional source fingerprints and underestimation in conventional source apportionment analyses. Integration of field observations with source-resolved modeling indicates that bbSOA forms a substantial fraction of regional background OA, even at locations far from fires, and persists for days, enabling long-range transport. The strong associations between MO-OOA and biomass burning tracers (K+, m/z 60 during fire periods), as well as elevated oxidative potential dominated by MO-OOA, underscore the relevance of aged biomass burning aerosol to toxicity and health risk. By quantifying European-wide bbSOA levels and linking them to existing exposure–mortality functions, the study demonstrates that wildfire emissions may account for 15–22% of PM2.5-attributable summer deaths in Europe. These results reconcile the discrepancy between bottom-up emissions and top-down apportionment and highlight the need to explicitly account for bbOA aging in assessments of air quality and health impacts.

This study demonstrates that wildfire emissions rapidly transform into highly oxidized secondary OA that blends into the regional background, causing traditional tracers to underestimate biomass burning contributions. During July 2022, wildfires likely contributed about half of Europe’s OA, with average bbSOA levels of ~2–3 µg m−3 across the continent, and were associated with an estimated 10,800–16,200 premature deaths each summer. The work underscores the importance of including rapid bbOA aging in source apportionment and policy-relevant modeling to accurately assess exposure and health risks far from fire sources. Future research should: (a) extend analyses to multiple summers and regions to quantify interannual variability; (b) compare and reconcile multiple fire emission inventories and improve emission estimates; (c) further develop and evaluate bbOA aging schemes in models; (d) expand observational networks (including direct levoglucosan and other tracer measurements) to better constrain aging processes and long-range transport; and (e) refine exposure–response relationships specific to aged biomass burning aerosol toxicity.

• Modeling underestimation: PMCAMx-SR tended to underpredict OA (and bbSOA) at the measurement site; European-average bbSOA (2 µg m−3) is likely conservative.
• Emission uncertainties: Wildfire emission inventories (e.g., IS4FIRES) can differ by at least a factor of 2 across products and regions, propagating to bbSOA estimates.
• Tracer constraints: No direct levoglucosan measurements during the campaign; reliance on AMS f60/f73 proxies. K+ as a tracer can overestimate fresh bbOA in summer if aging is not accounted for; sea-salt and dust contributions were estimated and subtracted but retain uncertainty.
• Spatial-temporal scope: Single remote site in Greece and one-month period limit generalizability; broader spatial coverage and multi-summer analyses are needed.
• Health impact assumptions: Linear exposure–mortality relationship applied; toxicity of aged bbOA assumed to be at least comparable to average ambient PM2.5; uncertainties in OP-health linkages remain.