Organic Carbon Aerosol in the Free Troposphere:
Insights from ACE-Asia and ICARTT
Colette L. Heald, Daniel J. Jacob,
Rokjin J. Park, Solène Turquety,
Rynda C. Hudman
Rodney J. Weber, Rick Peltier, Amy Sullivan, Lynn M.
Russell, Barry J. Huebert, John H. Seinfeld, Hong Liao
Acknowledgements: NOAA-OGP, EPA-STAR, NSF-ATM
Fall AGU
December 8, 2005
ORGANIC CARBON AEROSOL
*Numbers from IPCC [2001]
Reactive
Organic
Gases
Secondary Organic
Aerosol (SOA): 8-40 TgC/yr
Nucleation or Condensation
OC
Global Model Representation of SOA:
1. “Effective primary”
yield
Oxidation
2. Two-product empirical
to3 smog chamber data
by OH, O3,fit
NO
Monoterpenes
FF: 45-80 TgC/yr
BB: 10-30 TgC/yr
Aromatics
Direct
Emission
Fossil Fuel
BIOGENIC SOURCES
Biomass
Burning
ANTHROPOGENIC SOURCES
FIRST SUGGESTIONS OF HIGH ORGANIC CARBON
AEROSOL CONCENTRATIONS IN THE FREE TROPOSPHERE
High organic loading
in the FT
High organic loading
in the UT
Single particles over NA
[Murphy et al., Science, 1998]
TARFOX (E US)
[Novakov et al., JGR, 1998]
ACE-ASIA: OC AEROSOL MEASUREMENTS IN THE
FREE TROPOSPHERE
(ACE-Asia aircraft campaign conducted off of Japan during April/May 2001)
Seinfeld group
Huebert group
+
Russell group
Mean Observations
Mean Simulation (GEOS-Chem [Park et al., 2003])
Observations
High Levels of OC were observed in the FT during ACE-Asia by 2 independent
measurement techniques. We cannot simulate this OC with current models.
ACE-ASIA: MODEL REPRODUCES OTHER AEROSOL
PROFILES
Secondary
production
Scavenging
Scavenging
Mean Observations
Mean Simulation (GEOS-Chem)
GEOS-Chem simulates both the magnitude and shape of sulfate and EC
concentrations throughout the troposphere  what is different about OC?
ACE-ASIA: SECONDARY ORGANIC AEROSOL
UNDERESTIMATED?
Secondary
Organic Aerosol
Condensation of low
vapour pressure ROGs
on pre-existing aerosol
Reactive
Organic Gases
Oxidation by
OH, O3, NO3
SOA is a good candidate:
condense more easily with colder temperature
AND be produced in the FT (escape scavenging)
GEOS-CHEM April Biogenic SOA
[Chung and Seinfeld, 2002]
mechanism
Biogenic VOCs
(eg. monoterpenes)
FT observations ~ 4mg/m3
Simulated biogenic SOA far too small!
ICARTT: COORDINATED ATMOSPHERIC CHEMISTRY
CAMPAIGN OVER EASTERN NORTH AMERICA AND NORTH
ATLANTIC IN SUMMER 2004
Multi-agency,
International Collaboration
MOPITT Observations of CO Transport
(July 17-19) [Turquety et al., in prep]
2004 fire season in North America:
• worst fire season on record
in Alaska
Emissions derived from MODIS
hot spots [Turquety et al., in prep]
OC: 1.4 TgC
OC emissions from biomass burning were 4 times climatological average!
UNDERESTIMATE OF OC AEROSOL DURING ICARTT
Observations
GEOS-Chem Simulation
OMC=organic molecular carbon (=1.4xOC)
WSOMC
WS=water soluble (40-80% of total OC, primarily SOA)
NOAA ITCT-2K4 flight tracks
SOA
OMC
(=POA+SOA)
(R. Weber’s PILS instrument aboard)
OC aerosol underestimate observed
over North America as well.
Note: biomass burning plumes were removed
EMISSIONS OF OC FROM BOREAL FIRES IN ALASKA/YUKON
Fires over boreal regions generate enough energy to inject emissions into FT.
Following Turquety et al. [in prep], we inject 60% of emissions directly into FT
(3-5km) making these emissions a dominant source of OC in the FT.
ITCT 2K4 OMC
Observations (WS only)
GEOS-Chem Simulation
(with injected emiss)
GEOS-Chem Simulation
(with emission in BL)
Injection of BB emissions into the FT increases the OC observed in the
FT down-wind. However lack of correlation with CO in observations suggests
that not all the OC can be attributed to the BB source.
INCLUDING ISOPRENE AS A SOURCE OF SOA
Recent study: yield of SOA from isoprene is 0.9-3.0%[Kroll et al., 2005].
Isoprene oxidation products have been observed in the particulate phase
[Claeys et al., 2004; Matsunaga et al., 2005]
GEIA Emissions July/August 2004
3% yield
= 0.4 Tg SOA
10% yield
= 0.8 Tg SOA
Applying smog chamber estimates [Kroll et al., 2005] to isoprene emissions
inventories suggests a 50% increase in the SOA source over NA.
ISOPRENE SOA SOURCE: COMPARISON WITH OBSERVATIONS
IMPROVE
(July-August 2004)
GEOS-Chem:
2-product SOA model
ITCT-2K4 OMC
POA: 0.73
SOA: 0.45
GEOS-Chem:
10%terp
POA: 0.73
SOA: 0.75
GEOS-Chem:
10%terp+3%isop
Observations (WS only)
GEOS-Chem (2-product SOA)
GEOS-Chem (10% terp SOA)
POA: 0.73 GEOS-Chem
SOA: 1.16
(10%terp + 3%isop SOA)
Isoprene SOA sources improves agreement with IMPROVE surface
observations (improves spatial correlation) particularly in the East
CLUES FROM CORRELATIONS WITH OTHER ICARTT SPECIES?
BL (< 2km)
FT (> 2km)
Weak correlation
with pollution
In the FT
Cloud-processing?
[Lim et al., 2005]
No correlation with
photochemicallyproduced O3
In the FT
Weak correlation
with biogenic
tracer in the FT
No correlation
with product of
isoprene oxidation
(Kroll et al. suggest
MACR forms SOA)
No correlation with
aromatic SOA
precursor
Note: BB plumes removed
IS SCAVENGING OF OC AEROSOLS OVERESTIMATED IN
MODELS?
Hydrophillic aerosols are wet scavenged assuming 100% solubility.
Recent analysis of cloud events at Puy de Dome suggest scavenging efficiency of
OC may be much lower [Sellegri et al., 2003].
ITCT 2K4 OMC
Observations
GEOS-Chem Simulation
GEOS-Chem Simulation
(with scavenging e=0.14)
A large decrease in scavenging efficiency increases OMC concentrations
throughout the troposphere. To what degree are OC aerosols internally mixed?
ORGANIC CARBON IN THE FT: AN ONGOING QUESTION
Chemistry leading to SOA production is not well understood!
(And not represented in global models)
BUT field observations can provide insights.
Conclusions:
1. The large underestimate in OC aerosol concentrations observed during
ACE-Asia cannot be explained by an underestimate in primary
emissions
2. High OC concentrations in the FT observed during ICARTT can be
partially explained by injection of aerosols from boreal fires in Alaska.
3. Including direct production of SOA from isoprene improves the
correlation with surface observations during ICARTT.
4. OC concentrations in the FT are sensitive to efficiency of wet loss
processes. How internally mixed are OC aerosols?
5. Many, many processes are not included in global models (SOA formation
in clouds, polymerization reactions, heterogeneous reactions, etc.). To
what degree can models represent OC aerosol concentrations (and the
important biosphere-atmosphere feedbacks) using simple
parameterizations?