1. 7th Annual CMAS Conference
Comparison of Two Annual PM2.5 Modeling Results for the South Coast Air Basin
7th Annual CMAS Conference
October 8, 2008
Bong Mann Kim and Joe Cassmassi

2. Annual PM2.5 Simulations CAMx and CMAQ
January 1, 2005 – December 31, 2005
Same input files
Emissions
Meteorological data
Boundary and initial files
Compared with 2005 PM2.5 data measured for the Multiple Air Toxics Exposure Study (MATES III)

3. Modeling Domain

4. Input Files Emissions Meteorological data Boundary and initial files
Monthly weekday, Saturday, Sunday
Monthly biogenic emissions
Meteorological data
MM5 initialized from NCEP w/ one-day “ramp up” period and run for 5-day w/o FDDA
Boundary and initial files
Generated from WRAP visibility global modeling results

5. Ambient Data PM2.5 sampling conducted as part of MATES III program
PM2.5 data collected once every 3 days
Sampled from April 2004 to Mach 2006 at ten locations
PM2.5 mass, ions, carbon, metals
PM2.5 data from eight sites are used in the model performance evaluation

6. Model Comparison CAMx CMAQ Model version v 4.2 v 4.6
Gas phase chemistry
CB4
CB05
Aqueous chemistry
RADM
Aerosol chemistry
ISORROPIA
AE4
Secondary organic chemistry
SOAP
Horizontal advection
BOTT
PPM
Vertical advection
Vertical diffusion
Eddy
Gas phase chemistry solver
CMC
EBI
Aerosol size distribution
2 sec. EFC
3 L-N Modes
Minimum vertical diffusivity
1.0 m2/s

7. Comparison of Model Performance
PM2.5 mass, NH4, NO3, SO4, OC, EC, Others
Statistics
Mean bias (MB)
Mean error (ME)
Normalized mean bias (NMB)
Normalized mean error (NME)
Graphics
Error plots
Scatter plots
Time series plots

8. Error Plots
CAMx
CMAQ

9. Performance Statistics
PM2.5 Mass
NO3
SO4 OC EC
Others
CAMx
CMAQ LA MB
10.23
13.18
3.05
2.57
-0.61
0.51
1.28
2.31
-0.02
0.37
1.35
2.87 ME
12.47
14.68
4.39
4.20
1.94
2.12
1.81
2.64
0.68
0.70
4.02
NMB
0.53
0.69
0.52
0.44
-0.16
0.14
0.27
0.49
-0.01
0.19
0.39
0.83
NME
0.65
0.77
0.76
0.72
0.56
0.38
0.34
0.35
1.16
RUB
5.21
6.30
2.01
2.13
-0.46
0.23
0.42
0.50
-0.54
-0.44
1.22
1.66
10.32
10.91
4.81
4.91
1.59
1.38
1.34
1.43
2.45
2.78
0.24
0.29
0.26
0.28
-0.15
0.07
0.10
0.12
-0.32
-0.26
0.47
0.63
0.64
0.33
0.36
0.40
0.78
LGB
10.41
12.03
2.48
2.49
1.20
0.73
0.82
2.36
2.73
11.22
12.87
3.38
2.18
2.75
1.85
0.91
0.95
3.36
3.60
0.60
0.61
0.62
0.00
0.17
0.18
0.54
0.57
0.80
0.74
0.84
0.75
0.43
0.66
0.99
1.06
All ST
6.52
7.57
2.05
-0.21
-0.06
1.11
1.61
10.49
11.50
3.98
3.84
2.03
2.11
2.02
3.27
-0.12
0.09
0.03
0.08
-0.11
-0.03
0.31
0.59
0.55
0.79
0.90

10. Scatter Plots for each species at Los Angeles
CAMx
CMAQ

11. Scatter Plots for each species at Rubidoux
CAMx
CMAQ

12. Scatter Plots for each species at Long Beach
CAMx
CMAQ

13. Scatter Plots for each species at Los Angeles
CAMx
CMAQ

14. Scatter Plots for each species at Rubidoux
CAMx
CMAQ

15. Scatter Plots for each species at Long Beach
CAMx
CMAQ

16. Conclusions CAMx and CMAQ produced similar results
Both models tend to over-predict NH4, NO3, OC, others, and PM2.5 mass
CAMx predicted better for OC, others and PM2.5 mass
CMAQ predicted better for NH4, NO3, and SO4
CAMx is about 2 times faster than CMAQ
CAMx takes ~3 days with one CPU
CMAQ takes ~7 days with one CPU