DAndrea_AGU_2013_poster_2.0

A41H-0166
Understanding global secondary organic aerosol amount and
size-resolved condensational behavior
Stephen D'Andrea1,2, Silja Häkkinen3,4, Daniel Westervelt5, Chongai Kuang6, Ezra Levin2, Vijay Kanawade7
Richard Leaitch8, Dominick Spracklen9, Ilona Riipinen10, Jeff Pierce2,1
1Dalhousie
University, 2Colorado State University, 3University of Helsinki, 4Columbia University, 5Carnegie Mellon University, 6Brookhaven National Laboratory,
7Indian
Institute of Technology, 8Environment Canada, 9University of Leeds, 10Stockholm University
Summary
1. Amount of SOA is very uncertain
2. SOA condensation to surface area or mass
3. Model Description
Secondary organic aerosols (SOA) are major contributors to
ultrafine particle growth to CCN sized particles.
•  Uncertainty #1: Recent studies show that an additional
source of SOA of 100 Tg (SOA) yr-1 is required to match
measurements (Spracklen et al., 2011).
•  Uncertainty #2: Many models treat SOA as semivolatile
with condensation onto the aerosol mass distribution; however,
recent closure studies show a significant fraction of SOA
condensing to aerosol surface area (Riipinen et al., 2011).
•  We use GEOS-Chem-TOMAS to simulate these findings.
•  We added an extra 100 Tg (SOA) yr-1 spatially correlated
with the emissions of anthropogenic carbon monoxide (CO).
•  We find that the additional source of SOA correlated with the
anthropogenic CO yields a 30% increase in N80 (number of
particles with Dp > 80 nm) and a 27% decrease in N10
(number of particles with Dp > 10 nm) globally over the base
emissions when using the surface area condensation scheme.
•  We find that the surface area condensation scheme yields a
5% increase in N10 and a 9% increase in N80 over the mass
distribution condensation scheme.
•  Comparing to global ground-based measurements confirmed
a decrease in the model-measurement bias when using the
surface area condensation scheme and the extra SOA was
included.
Top-down estimates suggest a global source ranging from 120-1820 Tg (SOA) yr-1,
meanwhile atmospheric models using bottom-up estimates suggest SOA formation of
12-70 Tg (SOA) yr-1. Therefore the uncertainty in the global SOA source is very
uncertain with ranges of 12-1820 Tg (SOA) yr-1 (Spracklen et al., 2011) (Figure 1).
Freshly nucleated (~1 nm) particles must grow to CCN sizes in order to affect climate (Figure 2).
To model the growth of these particles the condensation of SOA to the particles must be
assessed. Many models treat SOA solely as semi-volatile (C* >≈ 10-1 µg m-3), which leads to
condensation of SOA onto the aerosol mass distribution. However, recent closure studies with
field measurements show that a significant fraction of SOA condenses to the aerosol surface
area, which suggests a very low volatility (C* <≈ 10-3 µg m-3) (Figure 3).
•  3D atmospheric model: GOES-Chem (www.geos-chem.org)
•  GEOS-3 meteorological fields
•  4°x5° horizontal resolution, 30 vertical layers (from the surface to 0.01 hPa)
•  Aerosol microphysics: TwO-Moment Aerosol Sectional (TOMAS) (Adams and
Seinfeld, 2002, Pierce and Adams, 2009b)
•  Condensation, coagulation, nucleation
•  40 size bins (1 nm – 10 µm)
•  Nucleation rates predicted by ternary homogeneous nucleation of H2SO4, NH3,
and H2O based on the parameterization of Napari et al. (2002) scaled down
globally by a constant factor of 10-5 (Westervelt et al., 2013).
•  Simulations were run for the model year 2001 with one month spin-up.
l 
Figure 2 (left). Schematic showing the
connection between SOA, ultrafine particles
and CCN (taken from Riipinen et al., 2011).
Figure 3 (below). Measurements and model
simulations showing the growth of freshly
nucleated aerosols on 15 April 2007 in
Hyytiälä, Finland (taken from Riipinen et al.,
2011).
The two simulations show
condensation of organics proportional to 0%
and 100% of the surface area of the aerosols
(the remaining SOA in each case partitions to
the mass distribution of aerosols).
Figure 1. Comparison of
global sources of SOA from
various studies (taken from
Spracklen et al., 2011).
Sensitivity tests
Simulation Name
MASS-BASE
By comparing GLOMAP model simulations to AMS measurements, Spracklen et al.
(2011) were able to close the measurement-model gap by adding 100 Tg (SOA) yr-1 of
SOA correlated with anthropogenic CO emissions to the model simulations. In this
work, we will use this additional 100 Tg (SOA) yr-1 of additional SOA (correlated with
anthropogenic CO) in the model simulations.
4. Additional SOA increases CCN number
5. Surface area increases CCN number
Figure 4. Simulations for model year 2001 of the percent change of N10 in the BL
between the additional SOA correlated to the anthropogenic CO emissions and the base
case emissions both using the surface area condensation scheme (SURF-XSOA –
SURF-BASE). The additional SOA yields a 27% global BL decrease in N10.
Figure 6. Simulations for model year 2001 of the percent change of N10 in the
continental boundary layer (BL) between condensation of SOA proportional to the
surface area compared to the mass distribution (SURF-BASE – MASS-BASE).
The surface area condensation scheme yields a 5% global BL increase in N10 over
the mass distribution scheme.
SURF-BASE
Base biogenic SOA emissions with SOA condensing
proportional to the aerosol surface area
MASS-XSOA
100 Tg yr-1 of anthropogenic SOA added correlating to the
emissions of anthropogenic CO using the mass distribution
condensation scheme
SURF-XSOA
100 Tg yr-1 of anthropogenic SOA added correlating to the
emissions of anthropogenic CO using the surface area
condensation scheme
6. Surface area condensation with the additional SOA performs best
Figure 8 (above left). Observed and simulated annual- and campaign-mean
particle number size distributions for 20 global ground-based sites.
Figure 9 (above right). 1:1 plots for measured and simulated annual-mean
N10, N40, N80 and N150, calculated log-mean bias (LMB), slope of the linear
regression (m) and correlation (R2). The dashed lines indicate 5:1 and 1:5 lines.
Simulation
LMB
N10
Figure 5. Simulations for model year 2001 of the percent change of N80 in the BL
between the additional SOA correlated with anthropogenic CO emissions and the base
case emissions both using the surface area condensation scheme (SURF-XSOA –
SURF-BASE). The additional SOA yields a 30% global BL increase in N80, however in
some regions with high anthropogenic CO emissions the increase in N80 exceeds 100%.
Figure 7. Simulations for model year 2001 of the percent change of N80 in the BL
between condensation of SOA to the surface area compared to the mass distribution
(SURF-BASE – MASS-BASE). The surface area condensation scheme yields a
9% global boundary-layer increase in N80 over the mass distribution scheme with
some biogenically active regions exceeding 100%.
Description
Base biogenic SOA emissions with SOA condensing
proportional to the aerosol mass distribution
N40
N80
N150
N10
N40
N80
N150 N10 N40 N80 N150
MASS-BASE
0.203 -0.047 -0.099 -0.199 1.031 0.825 0.729 0.628 0.91 0.90 0.86
0.80
SURF-BASE
0.203 -0.035 -0.083 -0.181 1.025 0.827 0.732 0.633 0.91 0.90 0.86
0.80
MASS-XSOA 0.100 -0.067 -0.084 -0.107 1.003 0.857 0.783 0.722 0.92 0.91 0.86
0.80
SURF-XSOA -0.030 -0.052 0.005
0.82
0.071 0.888 0.876 0.859 0.852 0.89 0.91 0.87
1. Riipinen, I., et al., Atmos. Chem. Phys., 11, 3865-3878, 2011.
2. Spracklen, D. V., et al., Atmos. Chem. Phys., 11, 12109-12136, 2011.
3. Napari, I., et al., J. Chem. Phys., 116, 4221-4227, 2002.
R2
m
7. References
Table I. Summary of the log-mean bias (LMB), slope of the linear regression
(m) and correlation (R2) for the different simulations. Red numbers represent
the best statistical result between all simulations
4. Adams, P. J., and J. H. Seinfeld, J. Geophys. Res., 10.1029/2001JD001010, 2002.
5. Pierce, J. R. and Adams, P. J., Geophys. Res. Lett., 36, L09820, 2009a.
6. Westervelt, D. M., et al., Atmos. Chem. Phys. Disc., 13, 8333-8386, 2013.
8. Acknowledgements
The authors would like to acknowledge the Natural Science and Engineering
Research Council (NSERC) of Canada for funding this project as well as the
Atlantic Computational Excellence Network (ACE-Net) for computational
resources.