1. Implementation of Sulfate and Sea-Salt Aerosol Microphysics in GEOS-Chem
Hi everyone. My name is Win Trivitayanurak… I’m a PhD student working with Peter Adams at Carnegie Mellon. I am going to talk about aerosol microphysics implementation in GEOSChem.
Win Trivitayanurak & Peter Adams
Carnegie Mellon University
3rd GEOS-Chem Users’ Meeting
April 11, 2007

2. Radiative Forcing Estimates
Global-average estimates for 2005, relative to 1750
IPCC (2007)
+2.6 W/m2
+1.6 W/m2
-1.2 W/m2
Level of scientific understanding
Our project is driven by the need to improve aerosol indirect effect estimate. This chart from IPCC forth assessment report shows low level of scientific understanding for aerosol indirect effect. Notice the error bar. And the cloud lifetime/distribution effect (or the 2nd effect) is not even in the chart yet. [click] So how do improve the estimate?
Must improve aerosol indirect effect estimate!

3. Improving Aerosol Microphysics
GISS GCM-II’
Produce its own meteorology
Cannot predict at a specific time
GEOSCHEM
Driven by assimilated meteorology
Good for predicting at specific times for comparison with observations
The goal is to improve the estimate of cloud condensation nuclei (CCN) in the model by improving the aerosol microphysics. Originally there is this aerosol microphyics [click] in GISS GCM and this package would be for climate change and forcing estimate [click]. But the problem with GISS is that it cannot predict meteorogy accurately at a specific time – so I can’t compare aerosol prediction from GISS GCM to field campaign data like ACE-Asia. So I had to put this microphysics [click] into GEOSCHEM so that I can compare it with field campaign data. [click] With GEOSCHEM hosting this microphysics, this package is for aerosol model evaluation purpose. [click] Now let’s get to know more about the microphysics
Microphysics
Microphysics
Climate change &
Forcing estimate purpose
Aerosol model
evaluation purpose

4. Size-resolved Microphysics
TOMAS (TwO-Moment Aerosol Sectional microphysics algorithm)
[Adams and Seinfeld, 2002]
Moments = 1) aerosol number and 2) aerosol mass
30 bins segregated by dry mass per particle
Size range is about 10 nm – 10 μm
Size
Distribution
The microphysical algorithm is called TOMAS, which stands for Two Moment Aerosol Sectional microphysics. The first moment is aerosol number and the second moment is aerosol mass. There are 30 size sections segregated by dry mass per particle covering size range of about 10nm to 10um dry diameter. This picture shows what TOMAS tracks, in each bin, we have mass of each component such as sulfate, sea-salt, and so on… and also aerosol number. So we tracks three critical information, composition, amount, and size. [click] N7 … N6 N4 N5
NaCl
N30
NaCl N1 N3 N2
SO42- …
SO42- Dp

5. TOMAS processes Coagulation Condensation/ evaporation Nucleation
In-cloud sulfur oxidation
Size-resolved dry deposition
Size-resolved wet deposition
These are the microphysical processes. Apart from adding them, I also had to modify existing processes to work properly with aerosol size distribution. [click]

6. TOMAS Status Species: Number, SO42-, Sea-salt Advection Convection
Problem: TPCORE/ PPM parabolic interpolation artificially grows/shrinks particles
Fixed (see me for detail)
Convection
Negative entrainment at some time steps
Now let’s talk about the status of TOMAS and the GEOS-CHEM processes that were modified accordingly. Right now the aerosol species we have number, sulfate, and sea-salt. For advection, there was a problem with the piecewise parabolic method inside TPCORE module doing parabolic interpolation and grow or shrink particles artificially. This problem is fixed and you can talk to me for more detail. Convection, I observed some negative entrainment but this is probably insignificant errors and still unresolved.

7. TOMAS Status Emission Nucleation : Binary nucleation (H2SO4-H2O)
Sulfate: Same total, apply size distribution
Sea-salt: Clarke et al (2006) parameterization
Number: calculated from emitted mass and assumed size distribution
Nucleation : Binary nucleation (H2SO4-H2O)
Aqueous oxidation of sulfate
Distribute over activated size range using condensation/evaporation algorithm
Emissions: I apply assumed size distribution for the same total of sulfate. Sea-salt is emitted using Clarke 2006 param. Number is emitted according to mass and assumed size dist. Another source of aerosol particles is nucleation, I used the binary nucleation of sulfuric acid and water. Next, another source of sulfate mass from aqueous oxidation; this portion of mass is distributed over activated size range using condensation algorithm.

8. TOMAS Status Wet deposition In-cloud:
Activated particles are removed same way as the original rainout (1st order loss).
Interstitial particles are assumed inert.
Below-cloud: Apply size-resolved washout rate
For wet deposition. The in-cloud deposition is treated with the assumption that only particles larger than the activation diameter gets to be removed using the original rainout rate and interstitial particles are not removed. As for below cloud deposition, I applied size-resolved washout rate.

9. Dry particle diameter (mm)
TOMAS Status
Dry deposition
Size-resolved scheme of Zhang et al (2001) as in later versions of G-C
Dry particle diameter (mm) 10 1
0.1
0.01
Deposition Velocity (cm s-1)
Original
New

10. Aerosol Prediction from TOMAS

11. Model Results
Total Number Concentration (cm-3)

12. Model Results CCN(0.2%) (cm-3) Notice the scale! Total Number
Concentration (cm-3)

13. Model Results Total Number Concentration (cm-3) CCN(0.2%) (cm-3)
Tropical convection 
CCN removed with rain
Nucleation region
Emitted as well as grown into CCN sizes
Pressure [mbar]
Heavy pollution region
Latitude
Latitude
Total Number Concentration (cm-3)
CCN(0.2%) (cm-3)

14. Number Size Distribution
Model Results
Remote marine, Southern Ocean
Industrial region, China
Nucleated particles
Sulfate primary emission
Grow by condensation as they descend
Cloud processing
From cloud top down to surface
 Bimodal dist.
Sea-salt emission
Significant mass
(Larger particle)
Fewer numbers.
Number Size Distribution
dN/dLogDp (cm-3)

15. Number Size Distribution
L19
Model Results
Remote marine, Southern Ocean
L17
L10 L1
Number Size Distribution
dN/dLogDp (cm-3)

16. Upcoming Tasks & Challenges
Introducing 30-bin OC, EC, and dust
Running 2x2.5  memory need for (30*(n+2)) aerosol tracers, where n=number of mass species
OR … Running nested grid over ACE-Asia domain?
Comparing aerosol prediction with number size distribution measured from ACE-Asia

17. Question? Check out my poster about …
Comparison with Heintzenberg marine aerosol data
Inter-model comparison :
G-C, GISS GCM-II’ and GLOMAP