1. MODELING ORGANIC AEROSOL: Where are we going wrong? Colette L. Heald (heald@atmos.colostate.edu) Telluride Workshop on Organic Aerosol August 4, 2008

2. ORGANIC CARBON AEROSOL SOURCES Semi- Volatiles Oxidation by OH, O 3, NO 3 Direct Emission Fossil Fuel Biomass Burning Monoterpenes Sesquiterpenes Nucleation or ReversibleCondensation Aromatics ANTHROPOGENIC SOURCESBIOGENIC SOURCES Isoprene S econdary O rganic A erosol P rimary O rganic A erosol

3. GEOS-Chem OC AEROSOL SIMULATION P rimary O rganic A erosol Fossil Fuel: Cooke / Bond Biofuel: Logan & Yevich Biomass burning: Duncan / GFEDv2 50% hydrophobic, 50% hydrophillic S econdary O rganic A erosol Monoterpenes/Sesq: Chung and Seinfeld [2002] Isoprene: Henze and Seinfeld [2006] Aromatics: Henze and Seinfeld [2008] SOA treated as hydrophilic, semi-volatile gases are soluble Empirical parameterization: 2-product model fit to smog chamber studies VOC + Oxidant  α 1 P 1 + α 2 P 2 Equilibrium Partitioning Depends on T, [OC] k 1 k 2

4. GEOS-Chem OC AEROSOL SOURCES Semi- Volatiles Oxidation by OH, O 3, NO 3 Direct Emission Fossil Fuel Biomass Burning Monoterpenes Sesquiterpenes Nucleation or ReversibleCondensation Aromatics Isoprene POA: 69 Tg/yr FF: 11 BF: 8 BB: 50 SOA: ~30 Tg/yr Monot/Sesq: 12 Isoprene: 14 Aromatics: 4 SOA ~ 1/3 of global source [Park et al., 2003; Henze and Seinfeld, 2008]

5. WHY MODEL ESTIMATES OF SOA CAN DIFFER EVEN WITH THE “SAME” FRAMEWORK 1.BVOC emission inventories (EFs, drivers, etc) 2.Aromatic emission inventories 3.Oxidants (related to NOx, CO, VOC levels) 4.Pre-existing seed (which aerosols? With which emissions?) 5.Reversible formation? 6.Solubility of SOA and precursor semi-volatiles 7.Meteorology, esp. precipitation, convection  aerosol lifetime Model estimates of SOA can *easily* differ by a factor of 2. Current models, SOA production: 19-55 Tg/yr.

6. AN EXAMPLE OF MODEL RANGE Tsigaridis and Kanakidou [2003] showed a large range of simulated OC due to various assumptions about SOA scheme.

7. ORGANIC CARBON AEROSOL: AT THE SURFACE Organic carbon constitutes 10-70% of aerosol mass at surface. ISSUE: Measurements do not differentiate POA and SOA IMPORTANT: Always remember what measurements we are targeting (PM 1 – PM 2.5 ) 2004 NARSTO Assessment Global measurements (surface 0.5-32 μgm -3 ) [Zhang et al., 2007]

8. COMPARISONS AT SURFACE SITES Problematic: Can always adjust uncertain primary sources to account for discrepancies… Good agreement between GEOS-Chem and IMPROVE observations for OC aerosol concentrations in the US (once primary sources corrected) [Park et al., 2003] Factor of 2 underestimate by GISS model apparent (note used OM:OC=1.3) but suggested due to emissions and/or grid scale [Chung and Seinfeld, 2002] OBS IMPROVE (1998) MODEL

9. NEAQS 2002: ORGANIC AEROSOL GROWTH IN ANTHROPOGENIC PLUMES NE US: Urban BVOC: ~ 0.5 ppb NOx: ~ 10 ppb Obs OC ~ 6 (0-20) µg/m 3 f OC/aer : ~65% “Anthropogenic” air masses show more aerosol growth than can be explained by the oxidation of aromatics. [de Gouw et al., 2005]

10. FIRST SUGGESTIONS OF HIGH ORGANIC CARBON AEROSOL CONCENTRATIONS IN THE FREE TROPOSPHERE Single particles over NA [Murphy et al., 1998] High organic loading in the UT TARFOX (E US) [Novakov et al.,1998] High organic loading in the FT

11. ACE-ASIA: FIRST OC AEROSOL MEASUREMENTS IN THE FREE TROPOSPHERE (Spring 2001) Mean Observations Mean Simulation Observations + Concentrations of OC in the FT were under-predicted by a factor of 10-100 We conclude this is missing SOA (only terpene SOA) [Heald et al., 2005] [Mader et al., 2002] [Huebert et al., 2003] [Maria et al., 2003] Off of Japan: Marine/polluted BVOC: low? NOx: high? Obs OC ~ 4 (1-12) µg/m 3 f OC/aer : ~50% (surface), ~80% (aloft)

12. ACE-ASIA: WHY WE THOUGHT IT WOULD BE SOA Scavenging Obs Model Typical primary aerosol profile (surface source only) BIOMASS BURNING? No fires in Siberia Agricultural fires in SE Asia will not affect FT off of Japan No apparent underestimate of primary sources and/or mechanism to loft into FT Production? ANTHROPOGENIC? Influence FT?? No correlation with a “pollution” tracer

13. TORCH 2003: BOX MODEL SIMULATIONS REQUIRE LARGE INCREASES IN PARTITIONING TO MATCH OBS These authors previously found that they needed to increases partitioning by a factor of 5-80 with the MCM to match aromatic SOA formation at the EUPHORE chamber [Johnson et al., 2004; 2005]. [Johnson et al., 2006] Southern UK: rural/occasional London plumes (high T) BVOC: medium? NOx: medium? Obs OC ~ 4 (0-10) µg/m 3 f OC/aer : ~46%  To get this agreement: 1.Add 0.7 µg/m 3 bkgd 2.Increase partitioning coefficients by factor of 500

14. MCMA 2003: UNDERESTIMATED ASOA Mexico City: highly polluted BVOC: low NOx: high Obs OC ~ 20 (0-40) µg/m 3 f OC/aer : ~70% [Volkamer et al., 2006] Excess SOA from first-generation AVOC oxidation

15. SEVERAL STUDIES SUGGESTING UNDERESTIMATE OF SOA [Volkamer et al., 2006] Global underestimate in SOA?

16. ITCT-2K4: MODEST MODEL UNDERESTIMATE Note: biomass burning plumes were removed While model underestimated only by ~25% we cannot simulate variability in observations (R=0.21)  incomplete understanding of formation. [Heald et al., 2006] Sulfur Oxides (SOx) Simulated source attribution (includes isoprene SOA) Observed Simulated NE US: urban/rural mix BVOC: < 1 ppb NOx: medium Obs OC ~ 1.5 (0-10) µg/m 3 f OC/aer : ~43% Water soluble OC Aerosol (WSOC)

17. MILAGRO 2006: EVOLUTION OF URBAN PLUMES [Kleinman et al., 2008] Jean-Francois Lamarque, in prep. NW US: Mexico (aircraft) BVOC: low NOx: high Obs OC ~ 5 (0-30) µg/m 3 f OC/aer : ~60% CAM-Chem simulation including aromatic SOA [Heald et al., 2008] for MILAGRO SOA/OC Model Obs Simulating increasing SOA fraction!

18. IMPEX: ASIAN PLUME TRANSPORT Asian Pollution Layers First OC overestimate by model! How does this view match with the outflow from Asia (ACE-Asia)? [Dunlea et al., submitted] NW US: Asian plume BVOC: low NOx: low Obs OC ~ 0.5 (0-5) µg/m 3 f OC/aer : ~20%

19. AMAZE-08: OC AEROSOL IN “PRISTINE” CONDITIONS Early Feb: observe significantly more organic aerosol than simulated (rain ends this period)  Likely fire influence (either local of African) Model does not significantly underestimate observed concentrations. Observed OC is pretty low! Near Manaus, Brazil: clean tropical BVOC: ~5 ppb NOx: 1-2 ppb Obs OC: ~ 1 (0-4) µg/m 3 f OC/aer : ~80% Preliminary AMS obs: Scot Martin, Qi Zhang (Harvard), Jose Jimenez, Delphine Farmer (CU Boulder)

20. ADDING TO OUR PICTURE… ITCT-2K4 IMPEX AMAZE-08 MILAGRO Model underestimates not tracking photochemical age in all environments. What can we learn about SOA? This does NOT mean that models are getting better!  Must break down at point where signatures diluted/removed

21. LIKELY(?) MODEL WEAKNESSES 1. Current Emissions 2. Missing Precursors? Including semi-volatiles 3. Application of SOA yields  not relevant to ambient? Not including important drivers? 4. Partitioning: T-dependence, amount of pre-existing aerosol mass 5. Fate of gas-phase intermediates? 6. OC aging? 7. Additional formation pathways? (aqueous) 8. The effects of mixing state 9. Improperly characterized solubility, thus, loss

22. SENSITIVITY OF SOA CONCENTRATIONS TO PRE- EXISTING POA AVAILABILITY Is SOA formation limited by availability of POA? YESNO [Park et al., 2006] Important effect on spatial distribution and amount of SOA formed

23. IMPLICATIONS OF NEW LAB YIELDS FOR α-PINENE+O 3 GEOS-Chem Simulation: Surface JJA SOA from α-pinene+O 3 Yield: Griffin et al., 1999Yield: Shilling et al., 2008Shilling - Griffin Global annual mean burden of SOA from monoterpenes (when this yield applied for all) almost doubles from 0.28 TgC to 0.48 TgC

24. MISSING SOURCE: PRIMARY BIOLOGICAL AEROSOL PARTICLES (PBAP) POLLEN BACTERIA VIRUSES FUNGUS ALGAE PLANT DEBRIS How much does this source contribute to sub-micron OC? Jaenicke [2005] suggests may be as large a source as dust/sea salt (1000s Tg/yr) Elbert et al. [2007] suggest emission of fungal spores ~ 50 Tg/yr

25. MISSING MODEL SOURCE: TERRESTRIAL PRIMARY BIOLOGICAL AEROSOL PARTICLES (PBAP) POLLEN BACTERIA VIRUSES FUNGUS ALGAE PLANT DEBRIS If we size-segregate the Elbert et al. [2007] estimate of emissions of PBAP, only 2 Tg/yr source of PM1 (total was 50 Tg/yr) Also note, no initial suggestion from AMAZE-08 data that there is a large sub- micron PBAP source in the Amazon ([OC] ~ 1 µg/m 3 ) LARGE particles (> 10 µm) From Andi Andreae (unpublished data) BUT some data suggests up to 40% of sub-micron OC is cellular…

26. MARINE PBAP [O’Dowd et al., 2008] Empirical marine PBAP source (GEOS-Chem) [Spracklen et al., 2008] A source of 8 Tg/yr marine PBAP improves agreement with marine OC obs. Marine PBAP has limited influence on continental OC concentrations. Other parameterizations for fraction of sea-spray that is organic. More from Maria Kanakidou… Obs Chl-a Mod Mace Head

27. AQUEOUS SOA FORMATION Glyoxal (45 Tg/yr) Methylglyoxal (145 Tg/yr) SOA source: 2.6 TgC/yr 8.0 TgC/yr VOCs (esp isoprene) oxidation Irreversible (?) Uptake  ~10 -3 [Fu et al., in press JGR] Comprising: 1.Aqueous oxidation  organic acids 2. Oligomerization An in situ SOA source with both biogenic and anthropogenic sources

28. Cloud Processing ONCE WE INCLUDE ALL SOURCES, WILL SIMPLIFIED MODELS BE ABLE TO ACCURATELY SIMULATE OC? Semi- Volatiles Oxidation by OH, O 3, NO 3 Direct Emission Fossil Fuel Biomass Burning Monoterpenes Sesquiterpenes Nucleation or ReversibleCondensation AromaticsIsoprene S econdary O rganic A erosol P rimary O rganic A erosol