A Case Study Using the CMAQ Coupling with Global Dust Models

1
A Case Study Using the CMAQ Coupling with Global
Dust Models

• Youhua Tang, Pius Lee, Marina Tsidulko, Ho-Chun
  Huang, Sarah Lu, Dongchul Kim
• Scientific Applications International
  Corporation, Camp Springs, Maryland
• Jeffery T. McQueen, Geoffrey J. DiMego
• NOAA/NWS/National Centers for Environmental
  Prediction, Camp Springs, Maryland.
• Robert B. Pierce
• NOAA/NESDIS Advanced Satellite Products Branch,
  Madison, Wisconsin
• Patricia K. Quinn, Timothy S. Bates
• NOAA Pacific Marine Environmental Laboratory,
  Seattle, WA
• Hsin-Mu Lin, Daiwen Kang, Daniel Tong, Shao-cai
  Yu
• Science and Technology Corporation, Hampton, VA.
• Rohit Mathur, Jonathan E. Pleim, Tanya L. Otte,
  George Pouliot, Jeffrey O. Young, Kenneth L.
  Schere
• EPA National Exposure Research Laboratory,
  Research Triangle Park
• Paula M. Davidson

2
Objective

• Current operational WRF-NMM/CMAQ forecast still
  uses static profile lateral boundary condition
  (LBC). Our testing shows dynamic ozone LBCs from
  global models have significant impact on air
  quality prediction in upper and middle
  troposphere. What is the impact on particulate
  matter prediction?
• During Texas Air Quality Study 2006, the model
  inter-comparison team found all 7 regional air
  quality models missed some high-PM events that
  can not be reasonably interpreted with any local
  or regional factors. Here we revisit these events
  by coupling a regional model with global models.

3
WRF-NMM/CMAQ Model Configuration

• Driven by hourly meteorological forecasts from
  the operational North America Mesoscale (NAM)
  WRF-NMM prediction system.
• The operational CMAQ system covering Continental
  USA in 12km horizontal resolution
• Carbon Bond Mechanism-4 (CBM4) with AERO3
• 22 vertical layers up to 100hPa.
• vertical diffusivity and dry deposition
  based on Pleim and Xu (2001),
• scale J-table for photolysis attenuation
  due to cloud
• Asymmeric Convective Scheme (ACM) (Pleim
  and Chang, 1992).

4
Global Models as CMAQ LBC Providers
RAQMS (Real-time Air Quality Modeling System, Pierce et al, 2003) GFS-GOCART (offline dust)
Horizontal Resolution 2??2? T126 (1?x1?)
Meteorology GFS analysis GFS retrospective run
Anthropogenic emissions GEIA/EDGAR with updated Asian emission (Streets et al. 2003) Not active
Biomass burning emissions ecosystem/ severity based Not active
3-D Var Data Assimilation OMI/TES/MODIS assimilation Not applicable
Input frequency to CMAQ Every 6 hours Every 3 hours
5
GOCART has 5 dust bins in diameter 0.2-2 µm,
2-4 µm, 4-6 µm, 6-12 µm, 12-18 µm Which are
mapped into CMAQ with PM2.5bin10.4187bin2 PM_
Coarse0.5813bin2bin30.7685bin4
6
GFS-GOCART prediction for a dust intrusion event
around Aug 2, 2006
7
Surface weather map on July 28, 2006
Dust Intrusion Path
8
GFS-GOCART and RAQMS exhibit differences in
altitude and concentration of dust along the
eastern lateral boundary of CMAQ that causes
differences in PM prediction over Texas
9
GFS-GOCART LBC
Dust Intrusion Path
RAQMS LBC
10
GFS-GOCART LBC
RAQMS LBC
CMAQ surface PM2.5 (µg/m3) Compared to AIRNOW at
18Z, 08/02/2006
11
Comparison for surface stations over Texas
12
Florida
13
The NOAA ship Ron Brown measurements also showed
the dust signal in marine boundary layer.
Julian day 212 is July 31
Ron Brown dust mass is calculated as Dust
2.2Al 2.5Si 1.63Ca 2.2Fe 1.9Ti
This equation includes a 16 correction factor
to account for the presence of oxides of other
elements such as K, Na, Mn, Mg, and V. Also, the
equation omits K from biomass burning by using Fe
as a surrogate for soil K and an average K/Fe
ratio of 0.6 in soil. (Malm et al., JGR, 99,
1347, 1994)
14
Another method to use GFS-GOCART output CMAQ
base GOCART DUST PM2.5 (Bin10.4187Bin2)
15
Aug 17
Aug 18
Aug 19
Aug 20
Aug 21
Aug 25
Aug 22
5 km
Aug 28
3 km
Aug 23
CALIPSO images provided by Dave Winker
16
GFS-GOCART prediction for another dust intrusion
event around Aug 28, 2006
17
Another intrusion event around Aug 28
CMAQ predictions compared to Ron Brown data
18
Model simulations compared to AIRNOW hourly PM2.5
data
All Stations South of 38?N, East of -105?W
CMAQ base S0.418 R0.462 MB -4.65 S0.301 R0.431 MB -7.94
CMAQ with GFS-GOCART LBC S0.607 R0.538 MB -2.98 S0.709 R0.542 MB -4.11
CMAQ with RAQMS LBC S0.458 R0.402 MB -2.25 S0.386 R0.480 MB -6.64
CMAQ GOCART S1.092 R0.492 MB -0.783 S1.828 R0.458 MB 1.93
Period of 20060729 to 20060807
All Stations South of 38?N, East of -105?W
CMAQ base S0.339 R0.273 MB -3.24 S0.270 R0.336 MB -6.08
CMAQ with GFS-GOCART LBC S0.396 R0.315 MB -2.73 S0.375 R0.447 MB -5.36
CMAQ with RAQMS LBC S0.494 R0.289 MB -1.10 S0.326 R0.281 MB -3.97
CMAQ GOCART S0.459 R0.347 MB -2.46 S0.492 R0.480 MB -4.34
Period of 20060827 to 20060902
Other than the circled cases, the regional
predictions coupled with global models show
improvement over the CMAQ base prediction.
S is regression slope, R is correlation
coefficient, and MB is mean bias in µg/m3
19
Summary

• Appropriate LBCs are necessary for successful
  regional PM prediction during dust intrusion
  events. For summer 2006 events, dust LBC
  sometimes dominated the influence on regional PM
  prediction.
• The model results shows that there is strong
  sensitivity of the surface PM prediction to the
  entry height of the dust intrusion. (elevated
  lower troposphere versus near surface)
• These coupling experiments mainly reflect the
  long-range transport impact on certain local
  receptors. The model prediction can be very
  sensitive to accuracies of dynamical, physical
  and chemical processes in both global and
  regional models.