An Exploration of Model Concentration Differences Between CMAQ and CAMx

1
An Exploration of Model Concentration Differences
Between CMAQ and CAMx

• Brian Timin, Karen Wesson, Pat Dolwick,
• Norm Possiel, Sharon Phillips
• EPA/OAQPS
• Timin.brian_at_epa.gov
• October 3, 2007

2
Introduction

• OAQPS conducted 2001 base case modeling with CMAQ
  and CAMx
• Both models used the same raw emissions and
  meteorological data
• Large differences were seen in predicted ozone
  concentrations as well as other precursors and
  species
• We conducted several analyses to help examine
  differences in the models
• Sensitivity model runs with CAMx and CMAQ
• Analysis of existing information

3
2001 Model Platforms

• Both models were run with a very similar setup

4
CMAQ vs. CAMx- Ozone Concentration
July 17, 2001 at 21Z
CAMx (w/OBrien)
CMAQ
5
CMAQ vs. CAMx- CO Concentration
July 17, 2001 at 21Z
CAMx (w/OBrien)
CMAQ
6
CMAQ vs. CAMx- FORM
July 17, 2001 24-hour average
CAMx (w/OBrien)
CMAQ
7
CMAQ vs. CAMx- Sulfate Concentration
July 19, 2001 24-hour average
CAMx (w/OBrien)
CMAQ
8
Analyses

• Chemical mechanism
• Photolysis rates
• Cloud attenuation of radiation
• Vertical mixing
• Dry deposition

9
Analysis of Chemistry and Clouds

• CAMx (mechanism 4) uses a hybrid version of CB-IV
  which contains additional reactions (CB-IV)
  compared to CMAQ CB-IV
• Photolysis rates are generally higher in CAMx and
  with CB05 compared to CMAQ CB-IV
• Cloud attenuation of radiation differs between
  the models
• These differences between the models were judged
  not likely to cause significant regional ozone
  differences between the models

10
Vertical Mixing

• Vertical mixing is governed by vertical diffusion
  coefficients (Kv)
• CMAQ v4.5 used ACM mixing
• CAMx used OBrien Kvs
• There is an option in MM5CAMX to generate CMAQ
  like Kvs
• Comparison of actual CMAQ Kvs and CMAQ like
  Kvs confirmed similar magnitudes and spatial
  patterns
• We conducted a CAMx sensitivity run which used
  CMAQ like Kvs and compared the results to
  OBrien
• CMAQ like Kvs (and actual CMAQ Kvs) are
  generally much higher than OBrien Kvs
• Expect higher ozone with CMAQ like Kvs in NOx
  limited areas

11
CAMx Ozone Change-CMAQ-like Vs. OBrien KVs
Change in CAMx hourly ozone at 15Z on July 17,
2001
Change in CAMx hourly ozone at 20Z on July 17,
2001
Blue lower ozone with CMAQ-like Kvs
12
CAMx KVs and Ozone- Atlanta Example
CMAQ-like Kvs are (almost) always higher than
OBrien and tend to drop off at a higher layer
Ozone concentrations in CAMx and CMAQ are similar
at 15z, but CAMx becomes much higher at 20z
13
Maximum Daytime PBL Comparison

• We compared maximum PBL heights in Atlanta from
  observations, predictions from MM5 (MCIP), and
  from CMAQ and CAMx
• CMAQ tends to mix to a higher layer compared to
  the PBL heights from MCIP
• This example for Atlanta is not representative of
  all days and areas

Note CAMx and CMAQ mix up to the top of discrete
model layers (as defined in the table above)
14
Dry deposition

• CAMx uses a Wesely based dry deposition scheme
• CMAQ uses the M3Dry scheme
• Closely tied to the Pleim-Xiu land surface model
• Accounts for enhanced deposition to wetted
  surfaces (soluble species)
• Contains more recent science
• RADM dry deposition scheme (similar to Wesely) is
  optional in CMAQ (MCIP 3.2 and prior)
• Examination of dry deposition velocities (Vd)
  revealed large differences between models

15
Dry Deposition VelocitiesCAMx vs. CMAQ- Ozone
CMAQ (M3Dry) ozone Vd at 16Z on July 17, 2001
CAMx ozone Vd at 16Z on July 17, 2001
16
Dry Deposition VelocitiesCAMx vs. CMAQ- CO
CMAQ (M3Dry) CO Vd at 16Z on July 17, 2001
CAMx CO Vd at 16Z on July 17, 2001
17
Dry Deposition VelocitiesCAMx vs. CMAQ- NO
CAMx NO Vd at 16Z on July 17, 2001
CMAQ (M3Dry) NO Vd at 16Z on July 17, 2001
18
Dry Deposition VelocitiesCAMx vs. CMAQ- NO2
CAMx NO2 Vd at 16Z on July 17, 2001
CMAQ (M3Dry) NO2 Vd at 16Z on July 17, 2001
19
Dry Deposition VelocitiesCAMx vs. CMAQ- FORM
CAMx FORM Vd at 16Z on July 17, 2001
CMAQ (M3Dry) FORM Vd at 16Z on July 17, 2001
20
Dry Deposition Sensitivities

• Two CMAQ sensitivity runs were conducted to
  examine dry deposition issues
• Alternative mesophyll resistance values with
  M3Dry
• Alternative dry deposition scheme (RADM)
• The platform for these CMAQ runs was CMAQ v4.6
  with CB05 chemistry
• Ran CMAQ for 2 weeks in August 2002 (plus 7 day
  ramp-up)

21
CMAQ Dry Deposition Sensitivity No. 1 Mesophyll
Resistance

• M3Dry Vd values for CO, NO, and NO2 were found to
  be too high
• Added a mesophyll resistance value in MCIP for
• NO 9400 S/M
• NO2 500 S/M
• CO 100,000 S/M
• Ran MCIP and CMAQ with the new values
• August 2002 period

The mesophyll resistance values for NO, NO2, and
CO were later incorporated into MCIP 3.3
22
CMAQ CO Vd- M3Dry vs. M3Dry w/modified Mesophyll
Resistance
M3Dry CO Vd at 16Z on August
10, 2002 (w/mesophyll resistance)
M3Dry CO Vd at 16Z on August
10, 2002
23
CMAQ CO Concentration and Change in Concentration
Due to Mesophyll Resistance
Change in CMAQ CO concentration
on August 5, 2002 (w/mesophyll resistance) (24
hour avg.)
CMAQ CO concentration on
August 5, 2002 (24 hour avg.)
24
CMAQ Ozone Concentration and Change in
Concentration Due to Mesophyll Resistance
Change in CMAQ ozone concentration
on August 5, 2002 (w/mesophyll resistance)
(8 hour max.)
CMAQ ozone concentration on
August 5, 2002 (8 hour max.)
25
CMAQ Dry Deposition Sensitivity No. 2 RADM Dry

• The RADM dry deposition routine is an option in
  MCIP
• Formulation is based on Wesely, 1989
• Very similar to CAMx
• Ran MCIP and CMAQ with RADM dry
• August 2002 period

RADM Dry was removed from MCIP v3.3
26
CMAQ Ozone Vd- M3Dry vs. RADM
M3Dry ozone Vd at 16Z on
August 10, 2002
RADM Dry ozone Vd at 16Z on
August 10, 2002
27
CMAQ FORM Vd- M3Dry vs. RADM
M3Dry FORM Vd at 16Z on
August 10, 2002
RADM Dry FORM Vd at 16Z on
August 10, 2002
28
CMAQ Ozone Concentration and Change in
Concentration Due to RADM Dry Deposition
CMAQ ozone concentration on
August 5, 2002 w/M3Dry (8 hour max.)
Change in CMAQ ozone concentration
on August 5, 2002 (w/RADM Dry) (8
hour max.)
29
CMAQ Sulfate Concentration and Change in
Concentration Due to RADM Dry Deposition
Change in CMAQ sulfate concentration on August
5, 2002 (w/RADM Dry) (24 hour avg.)
CMAQ sulfate concentration
on August 5, 2002 w/M3Dry (24 hour avg.)
30
Conclusions

• We examined numerous differences between CMAQ and
  CAMx
• The majority of the ozone differences can be
  attributed to different implementations of
  vertical diffusion and dry deposition
• Numerous other smaller differences were also
  identified
• CO concentrations were too low in CMAQ due to
  high CO Vd (corrected by adding a mesophyll
  resistance value)
• Other species (including secondary aerosols) are
  also affected by mixing and dry deposition

31
Recommendations

• Further testing of vertical mixing is needed in
  both models
• Need more comparisons between observed PBL and
  CMAQ/CAMx mixing
• Does CMAQ overmix compared to MM5 predicted
  PBL?
• Does OBrien have too little mixing?
• More vertical layers may be needed in the AQM
  boundary layer
• Further examination of dry deposition velocities
  is needed
• Evaluate diurnal pattern of Vd
• Are afternoon Vd values too high in CMAQ?
• Does the Wesely scheme need to be replaced?

32
Recommendations

• Various combinations of chemical mechanisms
  (CB-IV, CB05, SAPRC), vertical diffusion
  (OBrien, ACM, ACM2) and dry deposition (M3Dry,
  Wesely, AERMOD) can give very different results
• Each process needs to be individually evaluated
• Operational ozone evaluation should not be used
  to determine the best model formulation