Evaluation of CMAQ v4.7 Sulfate Predictions for 2002

1
Evaluation of CMAQ v4.7 Sulfate Predictions for
2002 2006

• K. Wyat Appel and Shawn J. Roselle
• 8th Annual CMAS Conference, Chapel Hill, NC
• October 21, 2009

2
Acknowledgements

• EPA (CDC PHASE Project)
• Alice Gilliland
• Fred Dimmick
• Eric Hall
• Tom Pierce
• Norm Possiel
• Tyler Fox
• EPA AMAD
• Rohit Mathur
• Prakash Bhave
• Computer Sciences Corporation
• Lucille Bender
• Nancy Hwang
• Lara Reynolds

3
CMAQ Simulations

• Consistent annual simulations from 2002-2006
• 36-km CONUS and 12-km Eastern U.S. annual
  simulations
• MM5 Meteorology with 34 vertical layers
• GEOS-CHEM boundary conditions
• Based on 2002 GEOS-CHEM simulation
• Vary monthly/spatially, but same set of monthly
  values used for each year
• Emissions based on 2002 National Emissions
  Inventory
• Year-specific updates to fires, mobile and EGU
  point (CEMS data) emissions
• CMAQ v4.7
• 24 vertical layers
• CB05 Chemical Mechanism
• Base model only (i.e. criteria pollutants only)

4
CASTNET SO42- 2002 through 2006
2002
2003
2004
2005
2006
At the CASTNET sites, Sulfate is underpredicted
in the summer of all years, with the largest
overpredictions in 2002 and 2005.
5
IMPROVE SO42- 2002 through 2006
2002
2003
2004
2005
2006
At the rural IMPROVE sites, Sulfate is
underpredicted in the summer, with the largest
overpredictions in 2002 and 2005.
6
CSN SO42- 2002 through 2006
2006
2002
2003
2004
2005
At the urban CSN sites, Sulfate is underpredicted
in the summers of 2002 and 2005, but nearly
unbiased in the summers of 2003, 2004 and 2006
7
SO42- Monthly Average Normalized Mean Bias
2003
2005
2006
2002
2004
Since SO42- is most important during the summer
months, our focus has been on the summertime
underprediction of SO42-.
Of 180 month/network observations, only 34 show a
positive bias Of those 34, 26 occur between
September and December.
8
(No Transcript)
9
NADP SO42- Wet Deposition 2002 through 2006
2002
2003
2004
2005
2006
10
NADP Precipitation 2002 through 2006
However, biases in SO42- wet deposition do not
account for all the underprediction in ambient
SO42- during the summers of 2002 through 2006.
2002
2003
2004
2005
2006
11
Current/Future Work To Address Sulfate
Underprediction

• Gas-phase sulfate production
• Issues with cloud predictions
• Clear-sky photolysis sensitivity

12
CMAQ Sulfate Predictions with Clear Sky
PhotolysisJuly 2005
Absolute Difference
Ratio
Average increase at observation sites of 3.5 -
4.5.
13
Current/Future Work To Address Sulfate
Underprediction

• Gas-phase sulfate production
• Issues with cloud predictions
• Clear-sky sensitivity
• Results indicate that too little gas-phase SO42-
  production not the culprit
• Sulfate particle size distribution
• Comparisons to MOUDI data from 2003 and 2004

14
MOUDI CMAQ
See poster by Bhave et al. for more details
regarding comparisons of CMAQ with MOUDI data.
15
Current/Future Work To Address Sulfate
Underprediction

• Gas-phase sulfate production
• Issues with cloud predictions
• Clear-sky sensitivity
• Results indicate that too little gas-phase SO42-
  production not the culprit
• Sulfate particle size distribution
• Comparisons to MOUDI data from 2003 and 2004
• Results show particle size distribution too large
  and broad in some regions
• Sensitivity to vertical structure

16
Ratio of 34 layer / 14 layer CMAQ predicted
sulfate
SO42- concentrations higher with more vertical
layers. Average increase in SO42- at observation
sites is 5-6.
17
Current/Future Work To Address Sulfate
Underprediction

• Gas-phase sulfate production
• Issues with cloud predictions
• Clear-sky sensitivity
• Results indicate that too little gas-phase SO42-
  production not the culprit
• Sulfate particle size distribution
• Comparisons to MOUDI data from 2003 and 2004
• Results show particle size distribution too large
  and broad in some regions
• Sensitivity to vertical structure
• More vertical layers results in slightly higher
  SO42- concentrations
• Vertical distribution of SO42-

18
Regionally-Averaged Vertical ProfilesICART Time
Period (Summer 2004)
CMAQv4.7
Eta-CMAQ (v4.5)1
SO42- (ug/m3)
SO2 / Total S
SO42- (ug/m3)
SO2 / Total S
from Mathur et al., 2008
from CDC PHASE simulations
Overprediction in SO42- is lower with CMAQv4.7,
but still largely overpredicted aloft.
19
Current/Future Work To Address Sulfate
Underprediction

• Gas-phase sulfate production
• Issues with cloud predictions
• Clear-sky sensitivity
• Results indicate that too little gas-phase SO42-
  production not the culprit
• Sulfate particle size distribution
• Comparisons to MOUDI data from 2003 and 2004
• Results show particle size distribution too large
  and broad in some regions
• Sensitivity to vertical structure
• More vertical layers results in slightly higher
  SO42- concentrations
• Vertical distribution of SO42-
• Analyses show too much SO42- aloft
• New cloud scheme based on Grell cloud model
• Still in development

20
Summary

• Ambient SO42- underpredicted during the summer,
  particularly in the summers of 2002 and 2005
• Underprediction appears to be at least in part
  related to dry/hot summers
• Summer/Fall of 2005 was an active tropical year,
  which may have contributed as well
• Ambient SO42- overpredicted in the fall of 2003,
  2004 and 2006
• This issue remains to be investigated, but is a
  lower priority than summer underprediction
• SO42- wet deposition fairly well predicted in the
  summer
• Some overpredictions could contribute to
  underpredictions in ambient SO42-
• However, errors in wet deposition are not the
  main factor contributing to underprediction in
  ambient SO42-
• Near-term investigations into summertime SO42-
  underprediction
• Clear-sky photolysis sensitivity resulted in
  small increase in SO42-
• Errors in SO42- particle size distribution are
  still being investigated
• Vertical resolution plays a small role, with more
  vertical layers resulting in more SO42-
• Limited analysis showed too much SO42- aloft
  however, aloft prediction is improved from
  previous version of CMAQ.
• Future Work
• Additional sensitivities and analysis related to
  meteorological predictions
• Implementing a new cloud scheme (Grell) in CMAQ
  will force additional analysis into this issue.

21
Questions?

• I was gratified to be able to answer promptly.
• I said I don't know. - Mark Twain

22
Supplementary Slides
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
IMPROVE
CSN
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
(No Transcript)