Simulation of Organic Aerosol Concentrations in ThreeDimensional EmissionsBased

Simulation of Organic Aerosol Concentrations in Three-Dimensional Emissions-Based Atmospheric Models Robert Griffin Institute for the Study of Earth, Oceans, and Space University of New Hampshire EMEP Workshop on Particulate Matter Measurement & Modeling April 2004

Importance of Organic Aerosol Local: Slater and Dibb (2002) Global: Chung and Seinfeld (2002) Regional: Griffin et al. (2002) National: Schell et al. (2001)

Primary Organic Aerosols Schauer et al. , 2002

Secondary Organic Aerosols G 1, G 2, . . . , Gn VOC + ox P 1, P 2, …Pn A 1, A 2, . . . , An Kom, i Partitioning Theory Ai / Mo RT = ~ o Gi p L, i MWomgi AQ 1, AQ 2, . . . , AQn Henry’s Law and Dissociation Hi = giaq AQi/Gi AQi 2 - (further reactions in condensed phase…)

Modeling SOA – Where we? Fractional aerosol coefficient: Fixed percentage of hydrocarbon parent present is converted to SOA (Grosjean, 1992) Fixed aerosol yield: Fixed percentage of hydrocarbon parent oxidized is converted to SOA (Pandis et al. , 1992) Saturation: All mass of a product in excess of a given mixing ratio is converted to SOA (Strader et al. , 1999) Henry’s Law: Aqueous-phase partitioning only (Aumont et al. , 2000)

Modeling SOA – Where have we been recently? Varying yield approaches: ai. Kom, i Y = Mo å 1+ Kom, i Mo i æ ç ç ç è ö ÷ ÷ ÷ ø Issues: Based on smog chambers No revolatilization Implicit ‘previous step’ assumption Lack et al. , 2004 Debate: Incremental vs. overall aerosol yield (d. DMo/d. DROG) (Jiang, 2003)

Modeling SOA – Where are we now? Partitioning approaches Use a and K and allow partitioning to change based on T and conditions Predict chemical products via observed stoichiometry and use observed K with kon (mass transfer) and koff (equilibrium) (Kamens and Jaoui, 2001) Pun et al. , 2003* *Presents an in-depth discussion of SOA modeling uncertainties Andersson-Sköld and Simpson, 2001

Modeling SOA – Where are we now? Predict chemical products theoretically and predict K based on molecular properties at equilibrium (Griffin et al. , 2002; Pun et al. , 2002) Pun et al. , 2003 Griffin et al. , 2002 Issues: Largest uncertainties, highest computational burden

Modeling OA – Where are we going? Comparison to both observed and experimental results Internal versus external (Kleeman et al. , 1997) Equilibrium versus dynamic approach (Wexler et al. , 1994) Inorganic-organic interactions in the aqueous phase (Clegg et al. , 2001, 2003) What about other species? n-Alkanes, PAHs (UC-Riverside) Is POA really a non-volatile, inert component? Primary SVOC partitioning? Location, location – Houston (UT/UH), Atlanta (GIT), New England (UNH), Asia (Iowa? ), … New phenomena (Jang et al. , 2002; Kalberer et al. , 2004)

Modeling SOA – How can we use it? Pun et al. , 2003 Carreras et al. , 2004 (submitted) Molecular types of species in SOA Effects of emissions controls (? ) Health, cloud, and climate simulations? Griffin et al. , 2002

New Phenomenon: Acid-Catalyzed Chemistry For what parent species is it relevant? Only biogenics and aromatics? For what seeds and relative acidities is it relevant? CIT/UNC-CH/ETHPSI Adapted from Jang et al. (2002) Czoschke and Jang, 2003

New Phenomenon: Acid-Catalyzed Chemistry What are the kinetic and thermodynamic parameters associated with the chemistry? Barsanti and Pankow, 2004 (in press) ***How do we represent this in a three-dimensional model? ***

Other Needs Continued/improved characterization of gas- and particle-phase products from VOC oxidation, including identification and molecular properties Improved emissions inventories and speciation of POA (no more MUC) Investigation of isoprene chemistry – can it contribute to SOA in meaningful amounts after all? (Claeys et al. , 2004)

Conclusions/Acknowledgments OA most difficult because of the large number of species and incomplete understanding of relevant processes General underpredictions of OA: a combination of lack of appropriate molecular properties and ignorance of heterogeneous chemistry? SVOCs? Incremental versus overall yield? We’ve come a long way but still have a long way to go: integration of most rigorous modeling approaches and improved scientific understanding while considering computational burden Thanks: John Seinfeld, Donald Dabdub, Betty Pun, and Christian Seigneur (helpful discussions) and Myoseon Jang and Jim Pankow (provision of graphics)