Development of WRF-CMAQ Interface Processor (WCIP)

1
Development of WRF-CMAQ Interface Processor (WCIP)

• Seung-Bum Kim and Daewon W. Byun
• University of Houston
• Air Quality Modeling and Monitoring Center

2
Development of WCIP

• Background NCAR/NOAA/Air Force is Developing
  Weather Research and Forecasting (WRF) model to
  replace MM5 NCEP Eta
• Goal Build consistent on/off-line WRF-chem model
• Objectives
• Demonstrate that CMAQs Fully-Compressible
  Governing Set of Equations (FCGSEs) is
  Dynamically Consistent with WRF Eulerian Dynamic
  Cores
• Mass (EM), Height (EH)
• Provide algorithms for the WCIP implementation
• Mass conservation test of MM5, WRF-EH, WRF-EM
  with CMAQ

3
CMAQ FCGSEs and WRF Dynamic Cores (1)
Horizontal Momentum Equation
Vertical Momentum Equation
horizontal wind vector on the reference
earth-tangential Cartesian coordinates
Contra-variant wind components used in CMAQ
4
CMAQs FCGSEs and WRF Dynamic Cores (2)
Conservation Equations
Air Density
Entropy Density
Pollutants
Entropy Density
Diagnostic Equations
Ideal Gas Law
Pressure
5
WRF Eulerian Mass (EM) Dynamic Core (1)
Vertical Coordinate terrain following, time
dependent hydrostatic pressure (p)
Wind components
Vertical momentum Eq.
6
WRF Eulerian Mass (EM) Dynamic Core (2)
Conservation Equations
WRF Eulerian Height (EH) Dynamic Core
7
WCIP Met. Algorithms for EM Core
Mass-Jacobian weighted Contravariant wind
components
Comparison of WRF- W with omega-equation needed
WCIP Met. Algorithms for EH Core
8
Mass conservation test of MM5, WRF-EH, and WRF-EM
with CMAQ

• Purpose
• To quantify the mass consistency of each model
• To find out possible problems in on/off-line
  WRF-chem modeling

9
Experimental Design
Integration 08/26/00UTC-08/27/00UTC, 2000 (24hr) 08/26/00UTC-08/27/00UTC, 2000 (24hr) 08/26/00UTC-08/27/00UTC, 2000 (24hr)
Grid size 4km 4km 4km
Time step 10 sec 10 sec 10 sec
IC/BC Eta AWIP analysis data Eta AWIP analysis data Eta AWIP analysis data
Model MM5 v3.5 WRF mass ver1.2.1 WRF height ver1.2.1
Horizontal grid 161X146 161X146 161X146
Vertical grid 43 43 42
Physics (the same physics options are selected) Physics (the same physics options are selected) Physics (the same physics options are selected) Physics (the same physics options are selected)
Microphysics Simple ice NCEP simple ice NCEP simple ice
Longwave rad. RRTM RRTM RRTM
Shortwave rad. Dudhia(1989) Dudhia(1989) Dudhia(1989)
Surface-layer MOS MOS MOS
Land-surface OSU-LSM OSU-LSM OSU-LSM
Boundary-layer MRF MRF MRF
Cumulus None None None
10
(No Transcript)
11
Major functions of current WCIP

• Read WRF data
• Reconcile coordinate
• Horizontal interpolation
• Compute Jacobian, entropy, density, etc.
• Current WRF does not provide
• enough PBL parameters needed in
• CMAQ. We had to use PBL diagnostic
• routine built in MCIP in this
• implementation

12
Treatment of missing met. variables in WCIP

• surface roughness, albedo, emissivity, surface
  moisture availability
• ? use MCIP2 values
• latlon and map scale factor on dot grids
• ? interpolation using those on cross grids
• Pressure, density at full layers
• ? diagnosed by using ideal gas law

To avoid errors from interpolation or
approximation, we better ask WRF group to make
special output procedure for AQ modeling group
13
Jacobian of WRF EM
06 UTC
18 UTC
Jacobian-weighted density varies with time, but
it is constant vertically in WRF EM coordinate
14
Jacobians of MM5 and WRF EH
15
Difference of Initialization process

• Although we tried to make comparable MM5 and WRF
  outputs in this experiment, we have found MM5
  initialization routine provides organized
  vertical wind field initially, but, for some
  reason, WRF SI routine does not generate any
  initial vertical motion.
• However, initial horizontal motion field was
  quite similar.

16
Vertical Velocity at surface layer
2000/08/26/00UTC (Initial Time)
MM5
WRF mass
zero field
17
Benefit of WRF EM

• The benefit of WRF EM over WRF EH is that the
  tendency of Jacobian-weighted density shown in
  coordinate transform becomes the tendency of
  surface pressure in WRF EM,
• so that the vertical wind can be determined
  based on the divergence in the layer and the
  tendency of surface pressure term.
• This can be still applied to non-hydrostatic
  fully compressible atmosphere as long as we rely
  on hydrostatic pressure as coordinate.

18
Contravariant Vertical Velocity in WRF EM
dynamical core
The tendency eq. for the surface hydrostatic
pressure from continuity eq. in the ideal case
using the boundary conditions at top and bottom

Since eta is a material coordinate, the air
density does not explicitly appear in the
continuity equation. Therefore, it can be used to
estimate the contravariant vertical velocity
component by integrating the wind divergence term
either from the bottom to a level eta or from the
top to eta
(downward integration)
(upward integration)
19
Difference of vertical momentum component in
generalized coordinate
Gravity wave pattern ? WRF may need normal
mode initialization, because this pattern is
not realistic !!!
20
Mass conservative temporal interpolation method
(I)
The Jacobian and density at a time
Wind components multiplied with
Jacobian-weight density are interpolated linearly,
21
Mass conservative temporal interpolation method
(II)
Finally, interpolated wind components are derived
with
22
Vertical velocity multiplied with
Jacobian-weighted density2000/08/26/06UTC
MM5
WRF mass
WRF height
23
PBL height
WRF mass
MM5
WRF height
24
Vertical Velocity at surface layer
2000/08/26/20UTC (14LST)
(31,50)
25
Normalized IC1_BC1 concentrationVertical
velocity (in WRF mass)
RED w-component on mass coordinate directly from
WRF mass BLACK vertical velocity on mass
coordinate in WCIP using omega equation
Hourly WRF EM data have mass consistency !!!!
26
Normalized IC1_BC1 concentrationMM5, WRF mass,
WRF height(No Collapsing)
In spite of existence of gravity wave mode, WRF
EM shows mass consistency characteristics as
good as MM5 or a little bit better.
27
Normalized IC1_BC1 concentrationEffects of
Collapsing
Collapsing damages mass conservation
characteristics significantly!!!
28
Are high frequency met. data always better?Time
Resolution Issue
This result shows us that high-frequcy met. data
might be worse for mass conservation ? need the
consistent numerical transport algorithm
between meteorological and chemistry-transport
model
29
Summary and Conclusions

• On the way we develop consistent on/off-line
  WRF-chem model,
• 1) Reliable WCIP has been developed.
• 2) We need to communicate with WRF group on the
  following issues
• Although many met. parameters needed in the CMAQ
  are calculated in the WRF, they are not included
  in standard output of WRF presently.
• According to the mass conservation test, we need
  build consistent transport numerical algorithms
  both in WRF and CMAQ