Organic Aerosol: A riddle, wrapped in a mystery, inside an enigma * University of Leeds June 2, 2010 Colette L. Heald Acknowledgements: Dave Ridley, Kateryna Lapina (CSU), Jesse Kroll (MIT), Jose Jimenez, Ken Docherty, Delphine Farmer, Pete DeCarlo, Allison Aiken (CU- Boulder, currently or former), Qi Chen & Scot Martin (Harvard), Paulo Artaxo (University of Sao Paulo) * I’m sure Winston Churchill would have agreed with me

AEROSOL: CONNECTION TO BIG RESEARCH TOPICS IN ENVIRONMENTAL SCIENCE…A MOTIVATION TO GET IT RIGHT! AIR QUALITY / HEALTH Clear Day VISIBILITY BIOSPHERE-ATMOSPHERE CLIMATE

+ oxidants Terpenes (gas-phase) PBAP Hydrocarbons (gas-phase & particulate) ORGANIC AEROSOL IN PARTICULAR… P rimary O rganic A erosol: emitted S econdary O rganic A erosol: formed NATURAL ANTHROPOGENIC

MOTIVATION #1: ORGANIC AEROSOL MAKES UP AN IMPORTANT FRACTION OF OBSERVED AEROSOL Globally makes up 25-75% of total fine aerosol at the surface (ignoring soot here) [Zhang et al., 2007] Sulfate Organics

MOTIVATION #2: ORGANIC AEROSOL MAY BECOME EVEN MORE IMPORTANT IN THE FUTURE! Present-Day Burden: Tg 1 Projection:↓ by > 50% by 2100? SULFATE OA 1 [Koch et al., 1999; Barth et al., 2000; Takemura et al., 2000] Present-Day Burden: ? (models ~1.6 Tg) Projection:  ? [Heald et al., 2008] OA Burden Andreae et al. [2005] suggest ↓ sulfate will accelerate greenhouse gas warming, but organics could compensate… ?

MOTIVATION #3: DEFINING THE “NATURAL” BACKGROUND Estimating the “natural” background is CRITICAL for the indirect effect – how much have CCN values increased?? July 97 [Spracklen et al., 2005] Pre-industrial values? More broadly, (1)How much is anthropogenic (controllable) (2)Can we predict how the natural fraction will respond to climate change?

CHALLENGES IN MODELING THE RIGHT LEVELS OF OA SOA measured/modeled = 4-100! [Volkamer et al., 2006] Models do get it right sometimes (even more puzzling?) but is it for the right reason? ITCT-2K4 IMPEX AMAZE-08 AMMA Egbert

WHY DON’T MODELS GET IT RIGHT…. Terpenes (gas-phase) PBAP Hydrocarbons (gas-phase & particulate) Uncertain Formation (Missing sources? Poorly understood processes?) Continuing Oxidation/Partitioning in the Atmosphere 10,000’s of (unidentified?) compounds with variable properties

FOCUS ON TODAY: (1) Trying to simplify our description of organic aerosol composition (2) Look at whether measurements of AOD from space can shed any light on the budget of OA H:C O:C

HOW ATMOSPHERIC AGING CAN CHANGE OA COMPOSITION OXIDATION: FunctionalizationFragmentation (break C-C bonds) OH Volatility  Generally O:C  Volatility  Composition may change (depends on where bond breaks and what caps end) VOLATILIZATION/CONDENSATION: More oxygenated material pulled into aerosol. PHYSICAL MIXING: Mean composition reflecting constituent air masses Organic Aerosol Composition is DYNAMIC – what a headache for modeling!

A SIMPLIFIED DESCRIPTION OF ORGANIC AEROSOL COMPOSITION [Goldstein et al., 2008] Typically < 20% of OA mass can be identified [Williams et al., 2007]. Even if we could identify these species, global models couldn’t handle this complexity! 2D chromatogram of OA Hydrogen Carbon Oxygen Other (N, S, etc) Alternate: Look at bulk elemental composition of aerosol Need a framework to compare composition & track changes…

THE VAN KREVELEN DIAGRAM Developed by Van Krevelen in 1950’s to describe oil formation Simple way to visualize changing composition

HOW DOES FUNCTIONALIZATION CHANGE AEROSOL COMPOSITION? If replace aliphatic carbon (-CH 2 -) with functional group, composition changes as follows: Example: Replace -CH 2 - with a carbonyl group -C(=O)-  Add 1O, lose 2H, slope = -2

LAB & FIELD ORGANIC AEROSOL LINE UP IN A VAN KREVELEN DIAGRAM! Surprisingly, despite complexity, aerosol composition changes during aging looks like carboxylation! [Heald et al., 2010] All measurements taken with the high resolution Aerosol Mass Spectrometer (HR-AMS)

EXAMPLES FROM THREE FIELD CAMPAIGNS… Riverside, California: dominated by urban sources Amazon basin: clean, low loadings, more oxygenated Mexico city (aircraft): regional sampling (clean & polluted) Photochemical clock shows moves “down” the line with aging. Some leveling off with long ages?

WHAT DOES THIS TELL US ABOUT PROCESSES AND/OR COMPOSITION? Lipids Monocarboxylic Acids Dicarboxylic Acids Proteins Lignin SOA Cellulose Tannins Condensed Hydrocarbons Observed OA Isoprene Tetrols → Glyoxal Oligomers Suggests that oligomerization is not the dominant aging process. Suggests that glyoxal oligomers and isoprene tetrols do not make up major fraction of mass. *regions overlaid from Wozniak et al., 2008

IMPLICATIONS 2.From a lab perspective: why does bulk OA “collapse” to this composition? What are the details of fragmentation & functionalization reactions in the atmosphere that result in net carboxylation? 1. From a modeling perspective: hope for a simple parameterization!  Need to understand aging timescale better (how fast do we move down the -1 slope?) … H:C O:C Acknowledgements: Jesse Kroll (MIT), Jose Jimenez, Ken Docherty, Delphine Farmer, Pete DeCarlo, Allison Aiken (CU-Boulder, currently or former), Qi Chen & Scot Martin (Harvard), Paulo Artaxo (University of Sao Paulo)

FOCUS ON TODAY: (1) Trying to simplify our description of organic aerosol composition (2) Look at whether measurements of AOD from space can shed any light on the budget of OA H:C O:C

A LARGE MISSING SOURCE OF ORGANIC AEROSOL? Goldstein and Galbally [2007] suggest that SOA source may be anywhere from TgC/yr. Can total aerosol optical depth (AOD) measurements shed any light on the total budget of OA? For comparison, current model (GEOS-Chem) estimates total ~50 TgC/yr (~2/3 POA, 1/3 SOA), equivalent burden ~0.81 TgC TgC yr -1

SOME CHALLENGES… 1.AOD is an integrated measure of ALL aerosols. 2.Satellite AOD observations don’t always agree, AERONET coverage sparse MODIS MISR Strategy: focus on continental AOD, use MISR as a global constraint Example of MODIS/MISR/AERONET (MAM 2008)

DO OBSERVATIONS SUPPORT MISSING SOURCE(S) OF OA? DJFJJA Model does underestimate observed AOD in most regions/seasons.

IF ONLY AEROSOL IN THE ATMOSPHERE WAS OA, WHAT LOADING IS IMPLIED BY SATELLITE AOD? Calculate the “hypothetical” AOD implied by a constant 1  g/m 3 profile over the land, and see how we need to scale this locally to make up ENTIRE AOD reported by MISR. OA loading is 3.7 TgC (~230 TgC/yr) over land  extrapolate to include outflow ~460 TgC/yr. (middle of Goldstein & Galbally range) Inverted total MISR AOD: Equivalent uniform OA profile

A MORE REALISTIC POSSIBILITY: ATTRIBUTE ALL OF MODEL UNDERESTIMATE TO OA (assuming all the negative bias is ONLY OA) Estimate that ~95 TgC/yr over land required to close the MISR-GEOS-Chem discrepancy.  extrapolate to global ~200 TgC/yr (approximately 4 times current model OA source) IF we remove N. Africa & the Middle East (dust) from this, total is reduced to ~160 TgC/yr Likely still too high… DJF JJA Possibly dust? Biofuels? Agriculture

UNCERTAINTIES ATTRIBUTED TO VERTICAL DISTRIBUTION 1212 Uniform vertical profile perhaps not very realistic… If the same mass is distributed with exponential drop off (atmospheric scale height assumed), the AOD increases by 15%. OA burden implied by AOD would be 15% lower (140 TgC/yr) if distributed exponentially. Surface concentrations would be ~ twice as high. OA [  g/m 3 ]

SEASONALITY OF “INVERTED” OA BURDEN Seasonality of “inverted” OA peaks in local spring/summer Coincident to peak in BVOC & oxidants…

COMPARISON AT AERONET SITES Adding “extra AOD” observed by MISR does improve model ability to capture both magnitude and variability of AERONET AOD DJFMAM JJASON GEOS-Chem simulation GEOS-Chem + “extra” AOD

DO THE SURFACE OBSERVATIONS OF OA SUPPORT A SIMILAR CONCLUSION? Monthly mean GEOS-Chem simulation (for 2003) sampled at sites from Zhang et al, [2007] (AMS observations from not exactly matched) Jungfraujoch Tokyo (winter) Duke Forest ACE-Asia Vancouver Fukue (May) Riverside Taunus Beijing (32  g/m 3 ) Mexico (22  g/m 3 ) Model does generally underestimate surface OA. “Inverted” surface OA will depend on vertical distribution (use here constant profile). Adding inverted OA marginally improves ability to capture AMS magnitude and variability. The data set is limited geographically. // Surface OA Surface OA (including inverted) (removing Beijing & Mexico) Jungfraujoch Storm Peak Duke Forest

SENSITIVITY OF ORGANIC AEROSOL OPTICS Assumed optical properties based on GADS database and log-normal size distribution recently evaluated by Drury et al. [2010] Uncertainty in simulated AOD due to optical properties a factor of

HAVE WE REDUCED THE UNCERTAINTY ON THE OA BUDGET? [Heald et al., in prep] 910 TgC/yr 460 TgC/yr 200 TgC/yr 50 TgC/yr 160 TgC/yr Goldstein and Galbally [2007] If all AOD inverted for OA Existing GEOS-Chem sources If attribute all MISR “excess” AOD to OA If remove Africa & Middle East Observations support a continental OA source of 140 TgC/yr (or less). Range estimated from uncertainty in optical properties 140 TgC/yrAccount for “likely” vertical distribution 140 TgC/yr

WHAT ABOUT MARINE SOURCES OF OA? Kateryna Lapina Oceanic AOD is underestimated in the model – patterns seem to match the marine OA estimated by Spracklen et al. [2008], but scaling up source by a factor of 10 required. Spracklen et al, annual mean marine OA emissions Spracklen et al., [2008] estimate source of 4 TgC/yr, Roelofs et al., [2008] estimate 75 TgC/yr, Gantt et al. [2009] estimate 22.3 TgC/yr, Vignati et al. [2010] estimate 4.1 TgC/yr What about in terms of AOD?

910 TgC/yr Goldstein and Galbally [2007] Suggests a MAXIMUM global source of ~180 TgC/yr of OA to the atmosphere. This is more than THREE TIMES what is currently included in global models…. BUT at the low end of Goldstein & Gallbally [2007] range. 140 TgC/yr 180 TgC/yr HAVE WE REDUCED THE UNCERTAINTY ON THE OA BUDGET? CONTINENTAL MARINE