Evaluating Dissipation in NWP Models Using Kinetic Energy Spectra

1
Evaluating Dissipation in NWP Models Using
Kinetic Energy Spectra
Bill Skamarock (NCAR/MMM) Mike Baldwin
(NOAA/NSSL) Wei Wang (NCAR/MMM)
Nastrom and Gage (1985) Spectra computed from
GASP observations (commercial aircraft)
2
Global Models Koshyk and Hamilton (2001)
3
Cloud Models Vallis, Shutts and Gray (1997)
4
What can we do with NWP model spectra?

• Evaluate model resolution Where does the
  spectra deviate from its expected behavior?
• Tune model dissipation (and physics) to maximize
  resolution within a given model formulation.
• Examine model spin-up How long does it take to
  fill the spectra with energy at the fine scales.

5
What should the kinetic energy spectra look like
in an NWP model?
6
NWP Model Resolution and Spectral Tails
7
WRF mass coordinate model, BAMEX forecasts
BAMEX Goal Study the lifecycles of mesoscale
convective vortices and bow echoes in and around
the St. Louis MO area
Field program conducted 20 May 6 July 2003
8
Preliminary BAMEX Forecast Verification
Number of MCSs for each 36 h forecast initialized
at 00 UTC.
(Done, Davis, and Weisman)
9
Real-time WRF 4 km BAMEX Forecast
Valid 6/10/03 12Z
Composite NEXRAD Radar
4 km BAMEX forecast 36 h Reflectivity
4 km BAMEX forecast 12 h Reflectivity
10
Real-time 12h WRF Reflectivity Forecast
Valid 6/10/03 12Z
Composite NEXRAD Radar
4 km BAMEX forecast
10 km BAMEX forecast
22 km CONUS forecast
11
Spectra for WRF-mass BAMEX Forecasts, 1 3 June
2003
Average over approx. 4 9 km height, on
model surfaces. 22 km conus WRF-mass 24 - 48 h
forecast avg.
10 km WRF-mass 24 - 48 h forecast avg.
4 km WRF-mass 12 - 36 h forecast avg.
12
Spectra for WRF-mass BAMEX Forecasts, 1 3 June
2003 Nastrom and Gage GASP Spectra
13
WRF-mass BAMEX Forecasts, 1 3 June
2003, Effective Resolutions
14
Numerical Dissipation and Spectra
WRF-mass BAMEX forecast 1 June 2003
Filters 5th order horz. upwind advection in
the RK3 scheme
15
Numerical Dissipation and Spectra
WRF-mass BAMEX forecast 1 June 2003
Filters 5th order horz. adv. (solid
green) 5th order horz. adv., 3rd order vert.
adv., 2nd order horz. filter scaled to horz.
def. (dashed green)
16
Numerical Dissipation and Spectra
WRF-mass BAMEX forecast 1 June 2003
Filters 5th order horz. adv., 3rd order vert.
adv., 2nd order horz. filter scaled to horz.
def. (dashed green) 2nd order horz. filter
with constant eddy viscosity equal to 17,000
m2/s (solid red)
17
Numerical Dissipation and Spectra
WRF-mass BAMEX forecast 1 June 2003
Filters 2nd order horz. filter using eddy
viscosity 17,000 m2/s (solid red) As above
with horizontal divergence damping using an
eddy viscosity of 74,000 m2/s (dashed red)
18
Numerical Dissipation and Spectra
WRF-mass BAMEX forecast 1 June 2003
Filters 4th order horz. filter using eddy
viscosity 4,250 x 108 m4/s (solid blue) 4th
order horz. filter using eddy viscosity 17,000 x
108 m4/s (dashed blue)
19
WRF-mass BAMEX Forecasts, 2 June 2003 Spectra
Evolution (model spin-up)
20
NWP Forecast Spectra Conclusions
Uses for Spectra

• Evaluate model resolution.
• Evaluate model dissipation (and physics) to
  maximize resolution.
• Examine model spin-up.
• Verification?