Jeff McQueen, Pius Lee, Marina Tsildulko, Youhua Tang,

1
Recent Upgrades of the NOAA WRF-CMAQ Air Quality
Forecast System 14 February 2007
Jeff McQueen, Pius Lee, Marina Tsildulko, Youhua
Tang, Geoff DiMego NOAA/NCEP Environmental
Modeling Center Rohit Mathur, Hsin-Mu Lin,
Daiwen Kang, Shocai Yu, George Pouliot U.S.
EPA/ORD/ASMD and NOAA/OAR/ARL Paula
Davidson NOAA/NWS/OST
2
2006 Developments

• WRF Transition
• Improved vertical coupling
• Continued lateral boundary condition tests
• CMAQ Improvements
• Optimized Advection scheme
• Emissions upgraded for 2006
• AQF system retrospective Real-time testing
• July 2005 with experimental CONUS configuration
• Real-time Verification
• Spatial map comparisons to observations (03 PBL
  hgt)
• Inclusion of AIRNOW PM 2.5 observations
• Improved Analyses Visualizations
• Focus group, TEXAQS06 projects

3
NAM-WRF Implementation

• Major Components of Upgrade Package for NAM
• Replace Eta Model with WRF version of NMM
  (Nonhydrostatic Mesoscale Model)
• WRF Common Modeling Infrastructure
• Non-hydrostatic dynamics
• Use of hybrid sigma-pressure vertical coordinate
  with top at 2 mb
• Refined advection, diffusion, numerics and
  physics
• Replace Eta 3D-Var analysis with Gridpoint
  Statistical Interpolation (GSI) analysis
• Unified (regional global) 3D-Variational
  analysis adapted to WRF
• Begin use of background errors based on WRF-NMM
  to 2 mb
• Use of new variable for moisture analysis
• Use of tendency in constraint terms
• Use of dynamically retuned observational errors

4
Forecast Domains (2006)
259 grid cells
CONUS 5x Domain 2. Experimental WRF-CMAQ-ozone
3. Developmental WRF-CMAQ- aerosol
1. Operational East 3x Domain
268 grid cells
5
2006 SystemsWRF-CMAQ 48 h forecasts
6
Summer 2006 PerformanceOverall biases
http//www.emc.ncep.noaa.gov/mmb/aq/fvs/web-test1/
html/regular.html
Courtesy M. Tsidulko
Temp interpolation Bug fixed (5X)
Isoprene Fixed (5X)
NAMY
7
California AQ Forecastsunder-prediction
http//www.emc.ncep.noaa.gov/mmb/aq/fvs/web-test1/
html/airnow.html
8
Isoprene Correction improved forecast Max.-8hr
O3 7/19/06
Loose
Old Tight
Isoprene Fix Tight
Improvements but further retrospective
experiments needed
9
Summer 2006 Performance1h errors after
Isoprene/NAM-Y corrections
1h RMSE by Forecast Hour
1 h Max Bias by day

• Avg RMSE reduced by 2-4 ppb in experimental CONUS
  system
• CONUS biases also improved over operational E.
  U.S. system

10
Summer 2006 PerformanceSub region 1h avg O3 RMSE
NE US 3X, 5X SE US 3X, 5X
Mid-West 3X, 5X Low. Miss 3X, 5X

• Avg RMSE reduced by 2-4 ppb in experimental
  CONUS (5X) system
• Little improvement in Mid-West, Miss. Valley

11
Retrospective Run Case Study 1 h Max O3 Biases
c
August 2, 2006
August 4, 2006
August 3, 2006

• Best performance over the Northeast and Midwest.
• High bias in Southeast. Courtesy Pius Lee
  Shocai Yu

12
Developmental Case Study Aerosols Daily Biases

• Best performance over the Northeast
• Low Bias in Southeast U.S.
• Courtesy P. Lee, D. Kang, S. Yu

13
PBL Height Verification
WRF/NMM (green) and estimated with RAOBS (RED)
NAMY
NAMY
M. Tsidulko
Western U.S.
Eastern U.S.
14
Isoprene Correction improved Experimental
Forecast Max.-8hr O3 7/19/06
Loose
Old Tight
Isoprene Fix Tight
Improvements but further retrospective
experiments needed Rohit Mathur
15
Mid-Atlantic Ozonesondes August 2, 2006 at 1700
UTC NAM-Y CMAQ vs Obs With CMAQ Isoprene
Corrections
sgma
Beltsville, MD
Wallops, VA
16
Western Mountain Ozonesondes August 4, 2006 at
2000 UTC NAM-Y CMAQ vs Obs With CMAQ Isoprene
Corrections
sgma
Boulder, CO
Socorro, NM
17
California Ozonesondes August 4, 2006 at 2000
UTC NAM vs Obs With CMAQ Isoprene Corrections
Default
RAQMS LBC
sgma
Table Mtn,CA
Holtville, CA
18
Trinidad Head, California
August 2, 2006
August 4, 2006
CMAQ static LBC CMAQ RAQMS LBC
19
Summary

• Experimental O3 testing (5X) revealed significant
  errors in WRF-Post, biogenic emissions, PREMAQ
  code and emission files
• Biogenic errors had largest impact  when
  correctly included, produced genl incr. in
  surface ozone
• Corrections to NAM improved performance in all
  AQF products
• operational, experimental, and developmental
  testing
• NAM-Y-CMAQ More localized but strong impacts
• Warmer, well mixed PBL esp. during stable
  conditions
• Positive impact on CMAQ performance except
  continued under-prediction over California
  South Texas
• Further retrospective testing being conducted
  with WRF-Launcher and CMAQ