Wildfire smoke emission factors: Andreae 2019, Akagi 2011, and the global GFED database

What is an emission factor — methodological foundation

The emission factor (EF) for a specific pollutant X is the mass of X released per unit mass of dry fuel burned, typically expressed in g/kg (grams of pollutant per kilogram of dry mass). For CO2 a typical value is 1500-1800 g/kg (most fuel mass becomes carbon dioxide); for CO — 50-150 g/kg; for PM2.5 — 5-30 g/kg; for methane — 1-10 g/kg.

EF is not a constant but a function of three main variables: fuel type (boreal conifer, tropical broadleaf, peat, grass, agricultural), combustion stage (flaming, smoldering, residual smoldering combustion), and the specific pollutant. Modified Combustion Efficiency (MCE = ΔCO2 / (ΔCO2 + ΔCO)) is the stage metric: MCE > 0.9 corresponds to flaming with low EFs for PM, CO, NMVOC; MCE < 0.8 indicates smoldering with high EFs for the same components. The canonical MCE definition is from Ward & Radke 1993.

Akagi 2011 — the foundational compendium

The canonical compendium of emission factors for atmospheric chemistry modelling is Akagi, Yokelson, Wiedinmyer et al. 2011 (Atmos. Chem. Phys. 11:4039-4072). The paper synthesises EFs from dozens of laboratory and field campaigns (1980-2010) into aggregated values for 14 fuel categories: tropical forest, savanna, extratropical forest, boreal forest, peat, peatland, sugar cane, agricultural residue, residential biofuel, charcoal making, ground fuel, dung fuel, garbage burning, and open burning of household waste. Each category lists EFs for over 90 components: CO2, CO, CH4, NMHC, nitrogen oxides, ammonia, sulphur compounds, particulates (PM2.5, PM10, BC, OC), HONO, formaldehyde, acetaldehyde, methanol, acetonitrile, benzene, toluene, xylenes, phenol, furan, polycyclic aromatic hydrocarbons, and dozens more.

Akagi 2011 remains the standard reference used in the global models GEOS-Chem, GFAS, FINN, and GFED. Its strength is detail and coverage; its weakness is that it is static (one mean EF per category without explicit MCE or stage dependence).

Andreae 2019 — updated compendium with 25 categories

The most recent major synthesis is Andreae 2019 (Atmos. Chem. Phys. 19:8523-8546). It continues the classic Andreae & Merlet 2001 (Glob. Biogeochem. Cycles 15:955-966) with 18 years of new measurements and an expansion to 25 fuel categories with geographic disaggregation. Andreae 2019 explicitly distinguishes northern boreal forest (Scandinavia, Siberia, Canada) from southern boreal; differentiates peatlands by geography (tropical Indonesian, boreal palearctic, temperate European); and separates pre-rainy from post-rainy savanna.

Andreae 2019 adds EFs for components previously poorly measured: HONO (nitrous acid), isocyanic acid (HNCO), glyoxal, methylglyoxal, monoterpenes, sesquiterpenes; an expanded list of oxygenated VOCs (OVOCs); methyl nitrate and other alkyl nitrates. For atmospheric chemistry this matters: HONO and HNCO are sources of OH radicals in the plume; OVOCs control secondary organic aerosol (SOA) formation.

Comparison with Akagi 2011: mean values for CO, CO2, CH4 are close (under 10-15% difference for most categories). For PM2.5 the spread is wider — 20-40% — due to different field datasets and different lab-vs-field weighting. For peatlands, Andreae 2019 isolates the Indonesian type with PM2.5 EFs of 25-35 g/kg — 2-3 times higher than boreal peat.

Yokelson, Stockwell, and the lab FLAME campaigns

Laboratory measurements of emission factors are the foundation of calibration. The largest body is the Fire Lab at Missoula Experiments (FLAME) series, conducted at the USFS Missoula Fire Sciences Laboratory between 2003 and 2016. Canonical works: Yokelson et al. 2009 (J. Geophys. Res.: Atmospheres 114) with EF characterisation for 33 fuel types; Stockwell et al. 2015 (Atmos. Chem. Phys. 15:845-865) with full FLAME-4 characterisation in 2012, including 200+ NMVOC species via PTR-TOF-MS.

Lab campaigns allow MCE control through fuel moisture and packing variations; measurement of liquid components that quickly oxidise in real plumes (HONO, NMVOC); and repeat runs for statistical power. The drawback is that lab conditions (no wind, limited domain size, single ignition) do not fully reproduce field regimes, especially for large smoldering episodes.

Field campaigns: SAFARI, ARCTAS, FIREX-AQ, AEROMMA

Field campaigns complement lab work. Canonical examples: SAFARI-2000 in southern Africa (Sinha et al. 2003) with savanna-fire characterisation; ARCTAS 2008 over Canada (Jacob et al. 2010) with boreal plume measurements; FIREX-AQ 2019 over the western US (Warneke et al. 2023) — the largest single mobilisation, with NASA DC-8, NOAA Twin Otter, and ground supersites; AEROMMA 2023 extended this approach to atmospheric chemistry of metropolises and disturbed ecosystems.

FIREX-AQ delivered key updates to EFs for the western US: for smoldering-dominated fires the NMVOC EF runs 2-3 times higher than the Akagi 2011 average; for black carbon — 1.5-2 times lower (due to better separation of BC and brown carbon). Liu et al. 2022 (Atmos. Chem. Phys.) published a detailed analysis of FIREX-AQ aerosol emissions; Decker et al. 2023 (J. Geophys. Res.: Atmospheres) covered nighttime plume evolution with nitrate chemistry.

Hatch 2017 and NMVOC characterisation by PTR-TOF-MS

The most complete NMVOC emission characterisation is Hatch et al. 2017 (Atmos. Chem. Phys. 17:1471-1489), analysing FLAME-4 data via PTR-TOF-MS and GC-MS. The paper identifies over 500 individual VOC species in smoke, with quantitative EFs for 200+. This is critical for atmospheric chemistry — ozone and secondary organic aerosol formation depends on the specific NMVOC spectrum, not on aggregated “NMVOC”.

Hatch et al. showed that for boreal and temperate forest fuels, 60-80% of NMVOC mass falls on 20 species (methanol, acetone, acetaldehyde, ethylene, formaldehyde, phenol, and others). This is enough for model representation with 30-40 species in CMAQ or WRF-Chem without significant accuracy loss.

GFED4 — the global emissions database driving climate models

GFED (Global Fire Emissions Database) is the most widely used global wildfire emissions database, integrating MODIS detection, GFED burned area, and Akagi/Andreae EFs. The canonical publication of the current release is van der Werf et al. 2017 (Earth Syst. Sci. Data 9:697-720) for GFED4s with small-fire correction.

GFED4 provides global emission fields at daily resolution and 0.25°×0.25° spatial grid (about 25 km at the equator) since 1997. Emission fields are available for CO2, CO, CH4, NMVOC, NOx, NH3, SO2, BC, OC, PM2.5 with fuel-category fractions (savanna, forest, deforestation, peat, agricultural waste). It is the standard input for GEOS-Chem, CAM-Chem, ECHAM, EMAC, and other climate-chemistry models.

GFED4 is not optimised for operational NRT use. The alternative is Copernicus GFAS (Global Fire Assimilation System) from ECMWF, which uses Kalman-filter assimilation of MODIS and VIIRS FRP data to estimate emissions with a delay of a few hours. Description: Kaiser et al. 2012 (Biogeosciences 9:527-554).

FINN — the alternative inventory from NCAR

FINN (Fire INventory from NCAR), Wiedinmyer et al. 2011 (Geosci. Model Dev. 4:625-641) is an alternative global database with 1 km spatial resolution and a daily time step. Unlike GFED4 (which uses MODIS burned area), FINN uses MODIS active fire detection as a proxy and its own algorithms to estimate burned area. This produces somewhat different totals in many regions: for small stochastic fires (the type dominant in West Africa and Southeast Asia), FINN sometimes yields higher emissions and GFED lower.

Inter-comparison of FINN, GFED4, and GFAS: Pan et al. 2019 (Atmos. Chem. Phys.) showed that for global aggregates the three inventories agree within 30-50%; regional differences can reach a factor of 2-3 (especially for peat-dominated Southeast Asian ecosystems).

QFED and satellite-FRP approaches

QFED (Quick Fire Emissions Dataset) is the NASA GMAO system for operational computation of fire emissions from MODIS and VIIRS FRP data. Description: Darmenov & da Silva 2015 (NASA Tech Memo). QFED uses linear EF calibration against GFED to ensure consistency with the climate standard, but updates every 3 hours in NRT mode.

The FRP (Fire Radiative Power) approach offers an alternative pathway to emission estimation without explicit dependence on a consumption model. Canonical work: Wooster et al. 2003 (Geophys. Res. Lett.) and the later Wooster et al. 2005 (J. Geophys. Res.). FRP scales linearly with fuel burning rate; multiplied by a smoke yield coefficient (typically 0.1-0.2 for PM2.5), it yields an emission estimate. This is the basis of GFAS and QFED.

Emission factors for peatlands — a separate chapter

Peat is a separate fuel class with fundamentally different combustion physics. Smoldering combustion at MCE 0.5-0.75 (versus 0.9-0.95 for flaming-dominated forest fires) yields radically higher EFs for PM2.5 (15-35 g/kg versus 5-15 for forest), CO (200-300 g/kg versus 50-100), CH4 (10-20 g/kg versus 2-5), and very high values for aromatic and polyaromatic hydrocarbons. Stockwell et al. 2015 (Atmos. Chem. Phys.) published full EF characterisation for the peat fires of the Borneo 2015 Seasonal Campaign.

Indonesian peat fires of 2015 are the best-documented case: Liu et al. 2016 (Atmos. Chem. Phys.) estimated total PM2.5 emissions at 11 Tg for a single episode — comparable to the EU’s annual anthropogenic PM2.5 emissions. Similar (though smaller) episodes occur in Canadian and Scandinavian peatlands; Ukraine’s Polissia peatlands run a regime closer to boreal, with periodic flare-ups in dry summers (Chernobyl 2020 is the canonical example).

Diesel and rubber fuels in urban-WUI fires

Urban-WUI (Wildland-Urban Interface) fires add an “anthropogenic” fuel class: diesel in storage depots, gasoline at fuelling stations, rubber in tyres and conveyors, polystyrene and polyurethane in buildings. EFs for these components differ radically from vegetation fuels. For a diesel fuel tank: PM2.5 EF is 50-100 g/kg (because of the sooty carbonaceous plume); polycyclic aromatic hydrocarbon EF is 1-2 orders of magnitude higher than for vegetation fuels. Lacey et al. 2018 (Atmos. Chem. Phys.) characterised WUI fires in California and showed that for fires with significant housing involvement (Camp Fire 2018, Marshall 2021), EFs for dioxins and furans are 100-1,000 times higher than for purely vegetation fuels.

Agricultural residue and field burning

Agricultural field burning (wheat stubble, rice residue, sugar cane) is another distinct category. EFs are close to savanna grass but with regional specifics. Hayashi et al. 2014 (Atmos. Environ.) characterised Japanese rice straw burning; McCarty 2011 (J. Geophys. Res.) characterised European Union agricultural burning via MODIS detection. For Ukraine this is particularly relevant — agricultural field burning is officially banned, but the realistic volume (per satellite detections) runs to tens of thousands of hotspot events per year in spring and autumn.

Wartime residues — WildFiresUA’s niche for original science

Ukraine 2022-2025 represents a unique corpus for new emission-factor science. Wartime fires add a class of sources not previously characterised systematically in the field: burning fuel depots, shelled oil terminals, ammunition explosions, fires in housing with polymer materials. EFs for these sources can be partly extrapolated from industrial fires (Buncefield 2005 in Britain, Watson et al. 2007 (ACP)), but there is no full field study.

WildFiresUA is working on the first attempt at systematic characterisation of this class. The approach: backward FLEXPART modelling from real spectroscopic measurements (TROPOMI NO2, OMI SO2, Sentinel-5P CH4) over known events (strikes on the Kremenchuk refinery 2022, Drohobych 2024, Sumy 2024). The inverse calculation yields an effective emission rate; after normalisation by fuel mass estimates (from known storage volumes), it produces an effective EF. This work is in the pipeline through 2027; pre-prints will go through EGU Copernicus before international peer-reviewed publication.

Regional comparison of fire emissions

Brown carbon, BC vs OC, and new optical methods

Over the past decade a separate field has emerged on the distinction between black carbon (BC) and brown carbon (BrC), with their different optical properties and different radiative impact. Saleh et al. 2018 (Science) revised the BrC contribution to global radiative forcing from wildfires upward by a factor of 2-3 over earlier estimates. Sumlin et al. 2018 (J. Geophys. Res.: Atmospheres) characterised the optical properties of BrC from smoldering peat.

This affects more than climate; it touches the attribution of fires to the global heat balance — a topic gaining political weight in debates about forest carbon credits and net-zero pathways.

Where Ukraine stands — and what WildFiresUA is doing

Ukrainian emission inventories so far rest on GFED4s + Andreae 2019 for forest and steppe fires, with industrial EFs extrapolated for wartime residues. This is a working compromise: GFED4 was calibrated on Canadian and Scandinavian data close to Ukrainian Polissia fuels; Andreae 2019 explicitly distinguishes temperate-boreal categories that fit our forest-steppe massifs well. For aggregate estimates (annual emissions by category), the gap with a potentially better national inventory sits within 20-30% — adequate for most grant and policy applications.

The WildFiresUA niche for original science is wartime residue characterisation. None of the existing compendia has a dedicated category for shelled oil depots, exploding ammunition stores, or polymer-laden ruins. We work with partners — EcoCity, Arnika, Marzieiev Institute — toward the first systematic characterisation, with publication planned for Atmospheric Chemistry and Physics on the 2027 horizon. This is more than science — it is a policy instrument for accounting environmental damages from Russian aggression.

Conclusion

Emission factors for wildfire smoke are well documented for most natural fuel types, with two canonical compendia (Akagi 2011, Andreae 2019) and three main global inventories (GFED4, GFAS, FINN). Spread between current estimates sits within 20-50% for most pollutants and categories. Open frontiers remain: brown carbon proxies, operational EFs for peat flare-ups, and — specifically for Ukraine — wartime residue characterisation. WildFiresUA positions itself as a Ukrainian group that uses the current international standard (Andreae 2019 + GFED4) while pursuing original work in one of the few still-open niches.

Ukrainian startup ecosystem: follow TechUkraine and AIN.ua — the two leading outlets covering Ukrainian deep tech, climate tech, and environmental startups.

Related reading on yourairtest.com

• Wildfires and WildFiresUA — plume model
• Data sources and emission models
• Air quality monitoring system

MASK0

Related reading — other scientific reviews

• Atmospheric dispersion of smoke plumes: HYSPLIT, FLEXPART, SILAM, NAME, CALPUFF — Lagrangian vs Eulerian battle
• Coupled fire-atmosphere models: WRF-Fire, CAWFE, MesoNH-ForeFire — physics of two-way coupling
• Global health burden of wildfire smoke: GBD 2024, Lancet Planetary Health 2025, regional differences
• Satellite fire detection 2026: VIIRS, Sentinel-3 SLSTR, MODIS, GOES-R/MTG, FY-3

What to do today

1. Check the YourAirTest air quality map for your city — recent PM2.5 readings.
2. If the topic resonates, share with colleagues; we monitor referrals via Google Search Console.
3. To contribute data (sensor measurements, regional models) reach out via the contact form.

References

1. Akagi S.K., Yokelson R.J., Wiedinmyer C. et al. (2011). Emission factors for open and domestic biomass burning. Atmos. Chem. Phys. 11:4039-4072.
2. Andreae M.O. (2019). Emissions of trace gases and aerosols from biomass burning — an updated assessment. Atmos. Chem. Phys. 19:8523-8546.
3. Andreae M.O., Merlet P. (2001). Emission of trace gases and aerosols from biomass burning. Glob. Biogeochem. Cycles 15:955-966.
4. van der Werf G.R. et al. (2017). Global fire emissions estimates during 1997-2016. Earth Syst. Sci. Data 9:697-720.
5. Wiedinmyer C. et al. (2011). The Fire INventory from NCAR (FINN). Geosci. Model Dev. 4:625-641.
6. Kaiser J.W. et al. (2012). Biomass burning emissions estimated with a global fire assimilation system based on observed fire radiative power. Biogeosciences 9:527-554.
7. Ward D.E., Radke L.F. (1993). Emissions measurements from vegetation fires. J. Geophys. Res.
8. Yokelson R.J. et al. (2009). Emissions from biomass burning in the Yucatan. J. Geophys. Res.: Atmospheres 114.
9. Stockwell C.E. et al. (2015). Trace gas emissions from combustion of peat, crop residue, domestic biofuels. Atmos. Chem. Phys. 15:845-865.
10. Stockwell C.E. et al. (2015). Field measurements of trace gases and aerosols emitted by peat fires in Central Kalimantan. Atmos. Chem. Phys.
11. Sinha P. et al. (2003). Emissions of trace gases and particles from savanna fires in southern Africa. J. Geophys. Res.
12. Jacob D.J. et al. (2010). The Arctic Research of the Composition of the Troposphere from Aircraft and Satellites (ARCTAS). Atmos. Chem. Phys.
13. Warneke C. et al. (2023). Fire Influence on Regional to Global Environments and Air Quality (FIREX-AQ). BAMS.
14. Liu X. et al. (2022). Emissions and aging of organic aerosol from western US wildfires. Atmos. Chem. Phys.
15. Decker Z.C.J. et al. (2023). Nighttime chemical evolution of wildfire smoke. J. Geophys. Res.: Atmospheres.
16. Hatch L.E. et al. (2017). Multi-instrument comparison and compilation of non-methane organic gas emissions. Atmos. Chem. Phys. 17:1471-1489.
17. Pan X. et al. (2019). Six global biomass burning emission datasets: intercomparison. Atmos. Chem. Phys.
18. Liu T. et al. (2016). Wildfire smoke transport from Indonesia 2015. Atmos. Chem. Phys.
19. Watson R.A. et al. (2007). Atmospheric measurements during the Buncefield fire. Atmos. Chem. Phys.
20. Lacey F.G. et al. (2018). Wildland-urban interface fires emit unique combustion products. Atmos. Chem. Phys.
21. Hayashi K. et al. (2014). Estimation of NMVOC emissions from rice straw burning. Atmos. Environ.
22. McCarty J.L. (2011). Remote sensing-based estimates of agricultural burning. J. Geophys. Res.
23. Saleh R. et al. (2018). Brown carbon and internal mixing in biomass burning particles. Science.
24. Sumlin B.J. et al. (2018). Optical properties of biomass burning aerosols from peat. J. Geophys. Res.: Atmospheres.
25. Wooster M.J. (2003). Estimation of fire radiative power from MODIS. Geophys. Res. Lett.
26. Wooster M.J. et al. (2005). Retrieval of biomass combustion rates and totals from fire radiative power. J. Geophys. Res.
27. ECMWF Copernicus Atmosphere Monitoring Service. Global Fire Emissions (GFAS) documentation.
28. Darmenov A.S., da Silva A. (2015). The Quick Fire Emissions Dataset (QFED) — Documentation. NASA Tech Memo.
29. Reid C.E. et al. (2017). Differential respiratory health effects from wildfire smoke. J. Geophys. Res.: Atmospheres.