Science Assessment of CAMx and CMAQ: I. Comparison of Model

Simulation of Houston-Galveston
Airshed Ozone Episode with
EPA’s CMAQ
Daewon Byun: PI
Soontae Kim, Beata Czader, Seungbum Kim
Emissions input
Chemical Mechanisms
Vertical Mixing
Objectives – Evaluation of modeled HRVOC effects with
an alternative modeling tool
• Air quality models based on first-principle description of nature are
extremely complex and depends on various inputs and model
assumptions
•TCEQ – utilizes Environ’s CAMx (Comprehensive Air quality Model–
Extended)
•Model being compared: EPA’s CMAQ (Community Multiscale Air Quality)
model
Benefits
• Comparative evaluation of two models provides tremendous insights on the
validity of model inputs, model configurations and results
• Help identify strengths/shortcomings of the many components in the system
• Can provide “weight of evidence” information for the present SIP modeling
Emissions Inventory
• Standard vs. imputed (base5b/psito2n2)
HRVOC emissions
– CB-IV mechanism
– SAPRC mechanism
• CAMx and CMAQ both use the same
emissions EI but some minor differences
– Different plume rise methods cause
different vertical distributions of elevated
source emissions.
– Some chemical species for CAMx are not
used in CMAQ. Ex) MEOH, ETOH
Institute for Multi-dimensional Air Quality Studies
Transport Algorithms Used in
CMAQ and CAMx
Process
UH CMAQ
TCEQ CAMx
Horizontal advection
PPM (Piecewise
Parabolic Method)
PPM
Vertical advection
PPM
Semi-implicit (CrankNicholson)
Horizontal diffusion
K-theory, constant
K-theory, variable
Vertical diffusion
K-theory with PBL
similarity method for
Kv calculation
K-theory with O'Brien
(1970) scheme for Kz
calculation
Mass adjustment
Yes
Yes
X22: CAMX
CAMx 4.03
TAMU&ATMET
Base5b regular
C_a01, TCEQ
Supersite: LaPorte
with base Texas Emissions
Two models
are quite
comparable
Q22: CMAQ
CMAQ4.2.2
TAMU (M_a02)
Base5b regular
C_a01, TCEQ
X20: CAMX
CAMx 4.03
TAMU&ATMET
Base5b psito2n2
C_a01, TCEQ
Supersite: LaPorte
with Imputed HRVOC (ETH, OLE)
CAMx
responds to
the imputed
data much
more
Q20: CMAQ
CMAQ4.2.2
TAMU (M_a02)
Base5b psito2n2
C_a01, TCEQ
X22: CAMX
CAMx 4.03
TAMU&ATMET
Base5b regular
C_a01, TCEQ
Supersites: LaPorte/Clinton
with Base Texas Emissions
Q22: CMAQ
CMAQ4.2.2
TAMU (M_a02)
Base5b regular
C_a01, TCEQ
Some missing peaks with base emissions
Not much bias
X20: CAMX
CAMx 4.03
TAMU&ATMET
Base5b psito2n2
C_a01, TCEQ
Supersites: LaPorte/Clinton
with Imputed HRVOC emissions
Some improvement here
Q20: CMAQ
CMAQ4.2.2
TAMU (M_a02)
Base5b psito2n2
C_a01, TCEQ
Often overpredicted
Mostly overpredicted
Aug. 28th Comparison with
NOAA Aircraft
CMAQ/CB-4 with imputed HRVOC
Good correlation with observation;
(model prediction somewhat lower)
X20: CAMX
CAMx 4.03
TAMU&ATMET
Base5b psito2n2
C_a01, TCEQ
Comparison with NCAR Aircraft
with Imputed HRVOC (ETH, OLE)
Still significant
underprediction
In ETH conc.
Q22: CMAQ
CMAQ4.2.2
TAMU (M_a02)
Base5b psito2n2
C_a01, TCEQ
Is there any other way to predict
High ozone productivity in the
model?
• Problem in the vertical
distribution of the imputed
HRVOC emissions?
• Different vertical mixing?
• Different chemical mechanism?
Vertical re-distribution of imputed HRVOC emissions
Regular EI: includes Area/Nonroad, Mobile, Point and Biogenic emissions
Imputed EI: Regular +
Additional VOC emissions
OSD (Ozone Season Day) emissions: ~ 130 tons/day
Hourly emissions: 30 ~ 70 tons/day
Most of the imputed HRVOC emissions are treated as
fugitives and thus ends up in the lower model layers
Speciated OSD emissions mapped into the CB-4
species
60
Emissions (tons/day)
50
ISOP
40
ETH
ALD2
30
FORM
PAR
20
OLE
10
0
E
EN
E
Y
TH
L
N
PE
N
TE
E
)
(1
)
)
)
E
-1
-1
E
E
E
1)
E
,2
,1
(
1,3
N
L
L
NE
N
1
N
N
N
E
E
E
,
Y
Y
E
E
E
E
E
E
I
I
(
T
EN
D
YL
NE
PR
TH
TH
E
EN
EX
AD
C
P
A
E
E
E
T
O
N
BU
H
I
P
E
T
O
S
E
U
I
D
-M
-M
B
AD
PR
RO
DI
BU
(3
(2
P
T
A
T
NE
NE
BU
N
E
E
E
T
T
P
BU
BU
Stack parameters
Species
ETHYLENE
PENTENE (1)
BUTENE
PROPYLENE
BUTENE (1)
BUTADIENE
BUTENE (3-METHYL-1)
BUTENE (2-METHYL-1)
ISOPRENE
HEXENE
DECENE,1PENTADIENE (E-1,3)
BUTADIENE, 1,2PROPADIENE
Average
# of
Stacks
Emissions
(tons/day)
3131
1343
76
3623
1266
338
415
538
463
522
1
10
1
1
51.881
3.343
3.120
52.831
7.213
6.433
0.321
1.107
2.018
2.733
0.000
0.315
0.014
0.018
Mean Ht.
(H>0.5m)
Mean Dia.
(D>0.01m)
Mean
Temp
(T>293K)
Mean Velo.
(V>0.0001m/s)
12.4
10.5
20.7
14.2
14.2
13.6
11.9
11.1
11.3
12.0
15.2
12.6
68.6
1.5
12.9
0.7
0.9
0.1
0.8
0.8
0.7
0.9
0.9
0.9
0.9
1.8
1.5
0.1
0.8
449.7
342.5
572.2
430.2
417.9
481.0
311.8
313.5
371.7
346.6
417.0
486.2
1089.0
294.0
411.0
4.992
1.222
0.005
4.957
2.464
6.659
0.076
0.080
1.140
0.967
8.232
1.251
0.001
0.001
3.56
He = Hs + a * F**b / U
F = 0.25*g * ( Ts-T )/T * V * D**2
Vertical re-allocation
Ozone concentrations predicted
Surprisingly not much difference…..
But improves ETH
CMAQ Kz Sensitivity Experiments
With
• CMAQ Eddy scheme
• CAMx Kz scheme (Louis79 & OB70)
• Holtslag and Boville (1993)
CAMx
Aug. 28th Comparison with NOAA Aircraft
Peaks matches well
CO shows serious mixing problem
CMAQ
Missing peaks in plumes
CO compares quite well
Column (21, 34) in East Houston and just north of Ship Channel
CMAQ Eddy scheme
CAMx Kz scheme (Louis79 & OB70)
Holtslag and Boville (1993)
2500
CMAQ
CAMx 2500
HB93
2000
1500
1000
2000
1500
1000
I=21 J=24
21UTC(15CST) 08/25/2000
I=21 J=24
21UTC(15CST) 08/25/2000
3000
3000
2500
2500
HEIGHT (m AGL)
3000
HEIGHT (m AGL)
I=21 J=24
21UTC(15CST) 08/25/2000
3000
HEIGHT (m AGL)
HEIGHT (m AGL)
I=21 J=34
21UTC(15CST) 08/25/2000
2000
1500
1000
2000
1500
1000
500
500
500
500
0
0
0
0
0
100 200 300 400 500 20
Kz (m2s-1)
40
60
80 100 120 140
O3 (ppbV)
0
10
20
30
40
50
0 2 4 6 8 10 12 14 16 18
NOx (ppbV)
HCHO (ppbV)
O3 against NOAA AL aircraft data
CMAQ
CAMxKz
HB93
08/25
08/30
CB-4 vs. SAPRC mechanisms
• Lumped VOC emissions in raw EI need to be
speciated into individual model species prior
to input to AQMs. Ex) CB4, SAPRC99, RADM2
• As an alternative chemical mechanism,
SAPRC99 includes more explicit chemical
species than CB4, but still cooperates with
grouped VOC species.
• It may not be enough to explain the roles of a
variety of VOC species in petrochemical plant
plumes over the HGA during the high ozone
formations.  Extended SAPRC
Institute for Multi-dimensional Air Quality Studies
Effects of chemical mechanism
CB-4 mechanism
SAPRC mechanism
CMAQ/SAPRC
Still missing some plume
peaks; but the correlations are
quite good (ozone)
CMAQ/SAPRC shows
promising results for NOy
Conclusive Remarks
Ozone reactivities in the air quality models
are significantly affected by
–
–
–
–
–
HRVOC emissions
Vertical mixing
Chemical representation
Meteorological inputs (not shown today)
Model configuration (not shown today)
Uncertainties in the HRVOC emissions data must be evaluated in
conjunction with all other key modeling factors