Title: Linkage between WRFNMM and CMAQ
1Linkage between WRF/NMM and CMAQ
- Daewon Byun (PI)
- C.K. Song P. Percell
- University of Houston
- Institute for Multidimensional Air Quality
Studies (IMAQS) - Coauthors
- Jon Pleim, Tanya Otte, Jeff Young, Rohit Mathur
- ASMD, Air Resources Laboratory, NOAA
- In partnership with U.S. EPA
- and many others
- Hsin-Mu Lin, David Wong, etc
2What are the main science issues of the NWP AQM
coupling?
Off-line
- Consistent governing set of equations state
variables - Consistent coordinates and grid structures
- Consistent numerics physics, and
parameterizations - Flexible able to help diverse stake holders
(research regulatory application use of
different emissions inputs) - Allow studying effects of using different basic
input data (e.g., Land Use/Land Cover,
topography, emissions, etc) separately - Same () numerics physics, and
parameterizations - Same () coordinates and grid structures
- Same () governing set of equations state
variables
On-line
Need to check how closely the dynamics
variables and trace species are matched
3Components of Off-line Coupled system
WRF/nmm
Spatial interpolation
WRF/nmm
WRF/nmm Postprocessors (vertical/horizontal)
WRF-CMAQ Interface Processor
PREMAQ (consistent vertical coordinate)
CMAQ/E-grid
CMAQ
On rotated lat/long E-grid coordinate Consistent
vertical coordinate
Lambert conformal projection C-grid
Loose coupling
Tight coupling
4Fully Compressible Atmosphere (OOyama, 1990) used
for CMAQ
Proper Coupling Requires
- Follow coordinates/grid of met model
- Reproduce Jacobian
- Couple state variables consistently
5WRF/NMM http//www.dtcenter.org/wrf-nmm/users/
Nonhydrostatic Mesoscale Model (NMM) core of the
Weather Research and Forecasting (WRF) system was
developed by NOAA/NCEP
WRF/NMM Hybrid sigma-pressure coord.
Arakawa-E Conserves mass, momentum,
enstrophy, TKE and scalar
ARW (Advance Research WRF) Terrain following
hydrostatic P coord. or Terrain following
sigma (ARW) Arakawa-C Conserves mass,
momentum, dry entropy, and
scalar
6Hybrid Sigma-Pressure Coordinate
7Define J for the Generalized Vertical Coordinate
Initial Terrain-Following Hydrostatic Sigma
coordinate
Method 1
sigma interface of the lower and upper
layers PD pressure of top of lower layer
Method 2
8Vertical Jacobian Discontinuity Problem
Solution
One way to remove discontinuity
For example,
SIGMA LEVELS 1.0000, .9976, .9948, .9920,
.9890, .9858, .9825, .9790, .9754, .9718, .9679,
.9637, .9590, .9538, .9480, .9415, .9340, .9251,
.9144, .9020, .8883, .8736, .8582, .8420, .8253,
.8079, .7900, .7714, .7523, .7326, .7124, .6915,
.6699, .6477, .6248, .6015, .5779, .5540, .5300,
.5057, .4812, .4566, .4319, .4070, .3822, .3576,
.3333, .3100, .2881, .2679, .2494, .2316, .2135,
.1936, .1707, .1445, .1159, .0863, .0569, .0282,
.0000,
Case 1) Surface pressure 101300 Pa
sigma(kc)0.3822,
Case 2) Surface pressure 70000 Pa
sigma(kc)0.3822,
9Horizontal E-Grid System of WRF/nmmRotated
lat./long Arakawa-E grid -gt C-grid for CMAQ
If we use diamond grid C(C,R,L,S) -gt C(CR, L,S)
10Dynamics with Semi-Staggered Arakawa E grid
The E grid is essentially a superposition of two
C grids.
Advantages of using E-grid with dynamics solution
- When only the adjustment terms in the equations
of motion and continuity are considered, two
large-scale solutions from each C grid may exist
independently, and a noisy total solution
results. - So, employ the forward-backward time differencing
scheme to prevents gravity wave separation and
thereby precludes the need for explicit filtering
(Mesinger 1973 Mesingerand Arakawa 1976 Janjic
1979).
11Consistent coordinates and grid structures
WRF/EM CMAQ utilize Arakawa-C Grid
Arakawa-B Grid (MM5) is linearly interpolated
onto Arakawa-C Grid (CMAQ)
Dimension for Grid Point
What to do with NMM E-grid data?
12How to Utilize Arakawa-E for CMAQ?
- Develop a horizontal advection algorithm in CMAQ
for Arakawa E-grids - Split 2-D horizontal advection operator into 1-D
operators and use CMAQ-proven 1-D schemes, such
as PPM, with alternation between appropriate X
and Y directions - Work directly with meteorological variables on
the E-grid - avoid spatial interpolation
Use rotated square cells (rotated B-grid then on
C-grid)
Spatial distribution of dependent variables for a
uniformly spaced Arakawa E-Grid
E-Grid with rotated square cells. Scalar
variables are considered to be constant on each
grid
13Advantages
- Makes the E-Grid look like a B-grid whose rows
and columns are along diagonal SW?NE and SE?NW
lines - Can use 1-D algorithm, e.g. PPM, along these
lines - CMAQ (and preprocessors) are familiar with
turning B-grid data into C-grid flux point data
Disadvantages
- Diagonal lines of cells have variable lengths,
which requires non-trivial extra book-keeping (in
EGRID_MODULE.F) - Requires interpolation of wind velocities to get
flux point values - Jagged boundary effect
- Parallelization could be more difficult
14Bookkeeping issues
Grid geometry changes depending on whether
the number of columns or rows is even or odd
Partitioning for parallelization
15Jagged Boundary Effect
rotated B-grid then on C-grid
Boundary values propagate into the domain because
boundaries are angled 45 degree
16Comparison between regular CMAQ and Option 1
Option 1 rotated B-grid then on C-grid
CMAQ C-grid
17Calculation Flow of WCIP/NMM
Mapping Variables
START
get env./IOAPI variables
- define grid/coord.
- rotated Lat./Lon coord.
- E-grid structure
- calculate Dx Dy
- allocate memory xgrid and cgrid
get met. data
calculation for WRF/NMM - Eta1 Eta2 -
geopotential height - hydrostatic pressure -
hydrometeor
GRIDOUT
derive dynamic fld.
METCRO/DOTOUT
continue
END
18- TEST Run
- Target Period 00Z June 28 - 06Z June 30, 2006
- Horizontal Resolution 12 km
-
19Model Configuration
C-Grid E-Grid --------------------------------
--------------------------------------------------
------- Met. MM5 v3.6.1 WRF/NMM v2.1 (w/ Eta
forecast) (w/ Eta forecast) MCIP MCIP
v3.0 WCIP/NMM v1.0 BCON BCON/Standard BCON/E-g
rid v1.0 ICON ICON/ Standard ICON/ E-grid
v1.0 CMAQ CMAQ v4.4 CMAQ/ E-grid
v1.0 ---------------------------------------------
--------------------------------------------
I.C. C-Grid UH-AQF/CMAQ 12km resolution output
00Z June 28, 2006 B.C. C-Grid UH
-AQF /CMAQ 36km resolution output
00Z June 28 06Z June 30, 2006 Emisson
None Chem. Mech. CB-IV
20Domain Configuration
C-Grid E-Grid --------------------------------
--------------------------------------------------
------- Met. (MM5) (WRF/NMM)
nx(dx) 100(12 km) 85(0.0780 deg.)
ny(dy) 100(12 km) 135(0.0724 deg.) nz 43
sigma 44 hybrid sigma-P CMAQ
nx(dx) 89 57 ny(dx) 89 113 nz 23
(see COORD_23L.EXT) 23 (JP Dis.) --------------
--------------------------------------------------
------------------------- dssqrt(dx2dy2)
12 km As for DOT case of MCIP, nx and ny
should be 90 As for CRO/DOT case of WCIP/NMM,
nx(ny) should be 59(115)
21Recommended Model Physics for WRF/NMM
Microphysics Ferrier Cumulus Convection
Betts-Miller-Janjic Shortwave Radiation GFDL
Longwave Radiation GFDL Lateral diffusion
Smagorinsky PBL, free atmosphere
Mellor-Yamada-Janjic Surface Layer Janjic
Scheme Land-Surface 4-layer soil model
22(No Transcript)
23CMAQ Results No emissions, Transport Chemistry
Only 12Z (06 CST) June 28, 2006 (12 hrs after
initial time)
24C-Grid E-Grid
Wind PBLHCO O3
25C-Grid E-Grid
Wind PBLHCO O3
hr18
26C-Grid E-Grid ZH
JabobianAir temp. U-wind
---- 13000 m
---- 13000 m
discontinuity
27C-Grid E-Grid
CO
CO
28C-Grid E-Grid
O3
O3
29Conclusion Presented a method to cast the WRF
meteorological data on CMAQ grid coordinate
structures to represent transportation of
pollutants. Developed WCIP/NMM, BCON/E-grid,
ICON/E-grid, and CMAQ/E-grid Performed
simulation (WRF/NMM -gt CMAQ/E-grid) was
successfully done A simple evaluation with
transport and chemistry was performed Results of
CMAQ/E-grid simulation is generally consist with
CMAQ/C-grid but reveal properly the discrepancy
of meteorological fields
Future Work To solve some unsolved problems
(WRF/NMM IOAPI, etc) More Evaluations
Documentation Deliver the developed codes to
NOAA/EPA for National AQF