Application of MomentSectional Algorithms to Emerging Problems in Aerosol Science

1
Application of Moment-Sectional Algorithms to
Emerging Problems in Aerosol Science
Center for Atmospheric Particle Studies (CAPS)

• Peter J. Adams
• JaeGun Jung, Yunha Lee, Jeffrey R. Pierce, Win
  Trivitayanurak, and Spyros N. Pandis
• Funding EPA, NASA, NSF
• IAMA Conference
• December 6, 2007

2
Outstanding Issues in Aerosol Science

• Indirect radiative forcing
• Global CCN concentrations and composition
• CCN sources and budget
• Human exposure to ultrafine aerosols
• Near-source processing/behavior
• Prediction of CN in air quality models
• Nucleation
• testing proposed mechanisms against observations
• evaluating impact on CN/CCN regionally and
  globally
• relative contribution compared to primary
  particles

3
Common Themes

• Number concentrations and distribution
• as important as mass
• Need to conserve number
• conserve number concentration during condensation
• know number loss rates during coagulation
• construct budgets of number distributions
• Want to be flexible about size distribution
• new modes may appear, grow and merge into
  existing (e.g. nucleation)
• cloud processing may split an existing mode (i.e.
  Hoppel gap)

4
TOMAS Overview

• TwO-Moment Aerosol Sectional algorithm
• Moments 1) aerosol number and 2) aerosol mass
• Average particle size within section not constant
• Size range usually 10 nm 10 µm
• 30 bins segregated by dry mass per particle
• Sometimes extended down to 1 nm (nucleation mode)

5
TOMAS Theory
General Dynamic Equation

• Coagulation

Condensation

• Original theory application cloud microphysics
  Tzivion et al. 1987, 1989
• Application to aerosol microphysics Adams and
  Seinfeld, 2002

6
TOMAS Theory

• Closure Issue Nk and Mk insufficient for unique
  solution ? need to specify sub-bin distribution
• Complete higher-order moments approximated in
  terms of tracked moments in coagulation integrals
• Generally piecewise linear but top hat an
  option for condensation

mo 2mo Mass
mo 2mo Mass
7
Advection of Moments in 3D Models

• Care must be taken that Nk and Mk advected in a
  consistent fashion (e.g. Piecewise Parabolic
  Method)

i-1 i i1
i-1 i i1
i-1 i i1
Spatial location

• As Mk spatial gradient generally differs from Nk,
  parabolas are different/independent
• Locally, implied partice size (Mk/Nk) may be
  unphysical and/or out of bin boundaries
• Solution advect Nk normally, override behavior
  for Mk (assume uniform Mk/Nk throughout grid cell)

8
Accuracy
(size-independent Kij)
(constant Cgas)
Predicted size distributions accurate Total mass
and number conserved to machine precision (lt10-5)
9
Comparison to Other Algorithms

• Traditional (mass-based) sectional
• dont predict/conserve number concentrations
• Modal
• assume modal structure
• Moment-based
• ambiguous about shape of size distribution
• Moving Center Sectional
• closest analog
• treats size distribution within each bin as
  monodisperse
• Advantages flexible size distributions,
  accuracy, mass/number conservation
• Disadvantage computational time
• 1 month global simulation 3 days computer time
• includes detailed gas-phase photochemistry
• GEOS-Chem single processor 2.66 GHz

10
Application Nucleation Mechanisms
Particle Diameter
Particle Diameter
Time of Day
11
Application UF Concentrations
2 x 104 cm-3
CN3 without nucleation Maximum 20,000 cm-3
2 x 105 cm-3
CN3 ternary nucleation Maximum 200,000 cm-3
July 27, 2001
12
Application CCN Prediction
13
Application Number Budgets
Ultrafine Mode
CCN Mode
80 nm
Nucleation 56 cm-3 day-1
Condensation 2.2 cm-3 day-1 Cloud processing
4.5 cm-3 day-1
Coagulation 101 cm-3 day-1
Emissions 84 cm-3 day-1
Deposition 31 cm-3 day-1
Emissions 1.4 cm-3 day-1
Deposition 8.3 cm-3 day-1
GEOS-Chem CTM global average Sulfate and
sea-salt only
14
Nucleation and CCN
Binary CCN(0.2) cm-3 STP
Binary nuc rate cm-3 s-1
Binary total num cm-3 STP
Pressure mb
Ternary nuc rate cm-3 s-1
Ternary total num cm-3 STP
Ternary CCN(0.2) cm-3 STP
CN increases 2x
CCN0.2 increases 10
Latitude
400
7000
40
1E-14
1E-4
1E6
0
0
700
15
Sensitivity of CCN to nucleation and primary
emissions

• Anthropogenic primary emissions
• Nucleation (binary nucleation case)
• Nucleation (ternary nucleation case)

16
POA-SOA Split
Predicted CCN(0.2) (100 POA) All runs have same
total OA source POA varied 0, 10, 50, 90,
100
17
Conclusions

• Hybrid moment-sectional algorithm is useful for
  problems where aerosol number (distribution) is
  important
• Conserves aerosol number and mass to machine
  precision
• Highly accurate, flexible for size distribution
• Still fast enough for application in 3D models
  (GISS GCM, GEOS-Chem, PMCAMx)
• Applied to nucleation, primary particles,
  ultrafine, and CCN production

18
Acknowledgments / Co-Authors
Jeff Pierce (AGU talk/poster on global nucleation)
Win Trivitayanurak (AGU poster on POA-SOA split)
Yunha Lee
Spyros Pandis
Jae Gun Jung