1. An Examination of WRF/Chem: Physical Parameterizations, Nesting Options, and Grid Resolution
Chris Misenis*, Xiaoming Hu, and Yang Zhang
North Carolina State University
Jerome Fast
Pacific Northwest National Laboratory
Georg Grell and Steven Peckham
NOAA Earth System Research Laboratory
*Now with N.C. DENR – Division of Air Quality

2. Outline Background and Motivation
Data Description and Model Configurations
Sensitivity to PBL and Land Surface Schemes
Sensitivity to Horizontal Grid Spacing and Nesting
Conclusions

3. Houston, TX
Courtesy: University of Texas

4. Process Interactions in WRF/Chem
Non-hydrostatic (with hydrostatic option) and fully mass-conserving.
Simulates trace gases and particulates “online” with meteorology.
Developed by the NOAA with contributions from NCAR, PNNL, NCSU, and BAMS.
Surface fluxes (sensible, latent heat) derived from LSM affect PBL scheme.
Surface meteorology from PBL affects LSM.
Both have direct impact on formation and transport of atmospheric pollutants.
For more information:
Grell et al., 2005, Atmos. Environ., 39
Fast et al., 2006, J. Geophys. Res., 111
Courtesy: UCAR (

5. Data Description – TexAQS-2000
Intensive field campaign in the Houston-Galveston area of East Texas during 8 to 9/2000.
Measured gaseous, particulate, and hazardous air pollutants at approximately 20 ground sites.
Measured vertical profiles by aircraft from several organizations.
Complex meteorological and geographical characteristics challenged capabilities of air quality models.
Courtesy: University of Texas (

6. WRF/Chem Configurations
Horizontal Grid Spacing: 12- and 4-km
Vertical Grid Spacing: 57 layers
Simulation Period: 28 August – 2 September, 2000
from TexAQS-2000
WRF (v.2.1.1) Options:
PBL: MYJ, YSU
LSM: RUC, Slab, NOAH
Surface Layer: Monin-Obukhov
Microphysics: Turned Off
Shortwave Radiation: Goddard
Longwave Radiation: RRTM (rapid radiative transfer model)

7. WRF/Chem Configurations (cont.)
Chemistry Options:
Gas-Phase Mechanism: RADM2
Aerosol Module: MADE/SORGAM
Ini. Cond.: Horizontally homogeneous
Emissions: TCEQ for gases
NEI ’99 v.3 for PM

8. WRF/Chem Simulation Design
12-km Sensitivity Simulations Nesting/Grid Option Simulations
Baseline: N_Y, 1W12
Physics Sensitivity: N_M, S_Y, R_Y
HGS/Nesting Sensitivity: 2W12, 1W4, 2W4
Simulation
LSM
PBL
N_Y
NOAH
YSU
S_Y
SLAB
R_Y
RUC
N_M
MYJ
Simulation
Grid Size
Nesting Option
1W12
12-km
None
2W12
Two-way
1W4
4-km
One-way
2W4

9. Normalized Mean Biases (NMBs), %
Sensitivity to PBL and LSM Schemes Time Series and Statistics for Meteorology
Normalized Mean Biases (NMBs), % T2 RH
WSP
WDR
PBLH
N_Y
-0.3
-27.4
12.6
6.6
54.4
N_M
-1.2
-21.7
27.1
6.4
22.7
S_Y
-4.1
2.5
1.7
7.5
24.3
R_Y
-30.5
24.2
5.7
53.7

10. Sensitivity to PBL and LSM Schemes Spatial Distributions of O3 and PM2
N_Y
S_Y
N_M
R_Y O3
PM2.5

11. Sensitivity to PBL and LSM Schemes Temporal Distributions of O3 and PM2.5

12. Normalized Mean Biases (NMBs), %
Sensitivity to PBL and LSM Schemes Vertical Distributions of O3 and Chemistry Statistics
Normalized Mean Biases (NMBs), % O3 NO
NO2 CO
PM2.5
N_Y
26.1
-80.2
54.9
-37.3
-1.0
N_M
10.5
-75.9
83.2
-24.1
6.1
S_Y
9.7
-74.7
110
-14.9
14.6
R_Y
13.4
-76.4
60.4
-33.2
0.5

13. Normalized Mean Biases (NMBs), %
Sensitivity to HGS and Nesting Time Series and Statistics for Meteorology
Normalized Mean Biases (NMBs), % T2 RH
WSP
WDR
PBLH
1W12
-2.7
-22.1
3.0
14.8
54.8
1W4
-7.4
-9.3
-4.8
12.1
2W12
-1.8
-18.4
22.8
10.6
59.3
2W4
-2.8
-21.6
6.4
13.1
53.8

14. Sensitivity to HGS and Nesting Spatial Distributions of O3 and PM2.5
1W12
1W4
2W12
2W4 O3
PM2.5

15. Sensitivity to HGS and Nesting Temporal Distribution of O3 and PM2.5

16. Normalized Mean Biases (NMBs), %
Sensitivity to HGS and Nesting Vertical Distribution of O3 and Chemistry Statistics
Normalized Mean Biases (NMBs), % O3 NO
NO2 CO
PM2.5
1W12
29.0
-80.1
105
-38.3
-32.1
1W4
19.1
-59.2
242
65.2
-22.3
2W12
27.7
-79.8
100
-32.8
2W4
25.8
-71.0
154
140
-35.1

17. Statistical Summary - MeteorologyRH
WSP
WDR
PBLH
N_Y U UU O
OOO
N_M OO
S_Y
R_Y
1W12
1W4
2W12
2W4
OOO: > 40%
OO: 15 to 40%
O: 0 to 15%
U: 0 to -15%
UU: -15 to -40%
UUU: < -40%
Normalized Mean Biases (NMB) in %

18. Statistical Summary - ChemistryO3 NO
NO2 CO
PM2.5
N_Y OO
UUU
OOO UU U
N_M O
S_Y
R_Y
1W12
1W4
2W12
2W4
OOO: > 40%
OO: 15 to 40%
O: 0 to 15%
U: 0 to -15%
UU: -15 to -40%
UUU: < -40%
Normalized Mean Biases (NMB) in %

19. Summary
No one simulation seems to greatly outperform the others for this particular episode.
Statistically, S_Y performs better for O3, NO, and CO, while N_Y performs better for NO2, and R_Y for PM2.5 (in terms of NMB). 1W4 performs better for O3, NO, and PM2.5, while 2W12 performs better for NO2 and CO.
Temporal variability of O3 is fairly well-captured, while PM2.5 is worse, though not as poor as CO or NOx species.
Computational efficiency is a major factor only for nesting options.
Two-way significantly slower than one-way.
Further understanding of model parameterizations and atmospheric processes is needed.
Large biases in PBLH, NOx, and, CO. How well current model parameterizations handle processes that influence meteorology and chemistry should be further examined.

20. Acknowledgements Pacific Northwest National Laboratory:
Drs. William Gustafson and Rahul Zaveri
Group Members:
Air Quality Forecasting Lab (NCSU)
Funding:
NSF Award No. Atm
NOAA # DW