Zavisa Janjic 50 Years of NWP, June 14- 17, 2004 1 AN EVOLUTIONARY APPROACH TO NONHYDROSTATIC MODELING Zavisa Janjic, Tom Black, Mattew Pyle, Hui-ya Chuang, Eric Rogers and Geoff DiMego National Centers for Environmental Prediction, Camp Springs, Maryland
29
Embed
☞ Two Weather Research and Forecasting (WRF) Model dynamical cores:
AN EVOLUTIONARY APPROACH TO NONHYDROSTATIC MODELING Zavisa Janjic, Tom Black, Mattew Pyle, Hui-ya Chuang, Eric Rogers and Geoff DiMego National Centers for Environmental Prediction, Camp Springs, Maryland. ☞ Two Weather Research and Forecasting (WRF) Model dynamical cores: - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 1
AN EVOLUTIONARY APPROACH TO NONHYDROSTATIC MODELING
Zavisa Janjic, Tom Black, Mattew Pyle,Hui-ya Chuang, Eric Rogers and Geoff DiMego
National Centers for Environmental Prediction, Camp Springs, Maryland
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 2
☞ Two Weather Research and Forecasting (WRF) Model dynamical cores:
● NCEP Nonhydrostatic Mesoscale Model (NMM)
Janjic, Gerrity and Nickovic, 2001, Monthly Weather ReviewJanjic, 2003, Meteorology and Atmospheric PhysicsBlack, Tucillo, Parallelization, Optimization, WRF standards
● Instead of extending cloud models to larger spatial and temporal scales, NMM built on experiences of NWP (Janjic et al., 2001; Janjic, 2003), i.e.,
- Relaxing the hydrostatic approximation, while
- Using modeling principles (Janjic, 1977, 1979, 1984)proven in NWP and regional climate applications.
● No linearizations or additional approximations required, fully compressible system(Janjic, Gerrity, Nickovic, 2001, MWR; Janjic, 2003, MAP).
● The nonhydrostatic effects as an add–on nonhydrostatic module:
- Easy comparison of hydrostatic and nonhydrostatic solutions;
- Reduced computational effort at lower resolutions.
● Pressure based vertical coordinate.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 4
☞ Basic discretization principles set up in Janjic, 1977, Beitrage (also used in Eta):
● Conservation of major integral properties (Arakawa, “mimetic” approach, hot topic in Math);
- Controlled nonlinear energy cascade through Energy and Enstrophy conservation.
● Cancellation between contributions of the omega-alpha term and the PGF to KE, consistent transformations between KE and potential energy; ● Minimization of pressure gradient force error.
☞ Implementation of the basic principles evolved significantly over time.
● For historical reason, Arakawa E grid still in the initial NCEP formulation, more advanced B grid formulation (NMM-B) also exists (Janjic, 2003, MAP)].
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 5
● For rotational flow and cyclic boundary conditions, horizontal momentum advection on semistaggered grids B/E conserves (Janjic 1984, MWR) appropriate finite difference analogs of:
Enstrophy as defined in terms of on Arakawa staggered grid C.
A
dAyx
22
2
2
2)(
dAxA
2
2
2)(
dA
yA
2
2
2)(
A
dAyx 2
2
2
2
A
dAxy
])()[( 22 Energy as defined in terms of on Arakawa staggered grid C and semi staggered grids B/E.
A
dAy
2)(
A
dAx
2)(
Momentum as defined in terms of on Arakawa staggered grid C and semi staggered grids B/E.
A
dAyx 2
2
2
2
dAxA
2
2dA
yA
2
2
● A number of quadratic and first order quantities conserved in case of general flow.
Flat coordinate surfaces at high altitudes where sigma problems worst;Higher vertical resolution over elevated terrain;No discontinuities and internal boundary conditions.
● Time stepping.
- No redundant computations, high computational efficiency!- Adams-Bashforth for horizontal advection of u, v, T and Coriolis,- Crank-Nicholson/Matsuno for vertical advection of u, v, T (60 levels).- Forward-Backward (Ames, 1968; Gadd, 1974; Janjic and Wiin-Nielsen, 1977,
JAS; Janjic 1979, Beitrage) for gravity waves.- Implicit for vertically propagating sound waves (Janjic et al., 2001, MWR).- Split forward, long time steps for physics (Janjic, 1990, MWR)
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 7
☞ The formulation successfully reproduces classical 2D nonhydrostatic solutions.
The cold bubble test. Potential temperatures after 300 s, 600 s and 900 s in the right hand part of the integration domain extending from the center to 19200 m, and from the surface to 4600 m. The contour interval is 10 K.
10
00
0
0
0
0
0
0
0
5
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0 0
0 0
0
0
0
0
0
0
0
0
5
5
5
5
0
0
0
0
0
0
0
0
0
0
00
0
0
0 00
0
Potential temperature after 360 s, 540 s, 720 s and 900 s. The area shown extends 16 km along the x axis, and from 0 m to 13200 m along the z axis. The contour interval is 10K.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 8
-2.5
-2.5-2.5
-2.5
00000
00
0 0000
00
00 000
0
0 0 0 0
00
00
000
000 0 0 00000 000
00
0000
0000
00
00
000 000
00
00
000
000
00
00
00 00
00
00
000 0
00
0
0
0
0
000 00
0
00
0
00
0
00
00
00000000
0
0
0
0
00
0
0
0
0
00
00
00
00000
00
000
0
00
0000 000
00
2.52.5
2.5
2.5
2.5
2.5
2.5
Full Analytical (Boussinesque)
Deviation of horizontal wind from basic state (uniform 10 m s-1) after 9000 s. The area shown extends 18400 m on each side of the center of the mountain, and from 0 m to 8000 m in the vertical. The contour interval is 0.5 m s-1 and the dashed contours indicate negative values.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 9
☞ WRF NMM model = Nonhydrostatic dynamics+Upgraded physical package of the NCEP Eta model (Janjic 1990, 1994, MWR; Chen, Janjic and Mitchell, 1997, BLM; Janjic 2000, JAS; Janjic 2001, NCEP Office Note 437), NCAR physics as well.
● Computationally robust, reliable in operations;
● THREE TIMES faster than most established NH models, can be sped up;
● Little noise, no Rayleigh damping and associated extra computational boundary condition at the top with real data with resolutions down to 100 m.
● NWP, convective cloud runs, PBL LES, with resolutions from 50 km to 100 m;
● Operational at NCEP (small domains, initialized and driven by the Eta):
- No filtering of mountains.- HiRes Windows, Fire Weather, On Call.
● Quasi-Operationally run elsewhere.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 10
☞ Atmospheric spectrum.
● The WRF-NMM and the NMM-B well qualified for investigating numerical spectra:
- Conservation of rotational energy and enstrophy, more accurate nonlinear energy cascade,
- Conservation of total energy provides stable integrations without excessive dissipation (either explicit or built-in finite-difference schemes) that could affect properties of model generated spectra,
- Hybrid pressure-sigma vertical coordinate system relatively free of errors associated with representation of mountains in the upper troposphere and in the stratosphere where the sigma coordinate errors are largest,
- Explicit formulation of dissipative processes allows precise “dosage” of dissipation.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 11
● Nastrom-Gage (1985, JAS) 1D spectrum in upper troposphere and lower stratosphere from commercial aircraft measurements.
● No spectral gap.
● Transition at few hundred kilometers from –3 slope to –5/3 slope.
● 0.01-0.3 m s-1 in the –5/3 range, high up, not to be confused with severe mesoscale phenomena!
104
108
103
105
102
102 101103
-3
-5/3
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 12
0
1
2
3
4
5
6
7
8
-6.0 -5.5 -5.0 -4.5 -4.0 -3.5 -3.0
isabelnmm3648 k -̂3 k -̂5/3
630 km 63 km
“Isabel”NMM,8 km, 60 lev. -3
-5/3
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 13
0
1
2
3
4
5
6
7
8
-6.0 -5.5 -5.0 -4.5 -4.0 -3.5 -3.0
oceanphy3648 k -̂3 k -̂5/3
Atlantic,NMM-B,15 km, 32 lev.
630 km 63 km
-5/3
-3
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 14
No Physics With PhysicsThe Atlantic case, NMM-B, 15 km, 32 Levels
No time (and space) for downscale nonlinear energy cascade, physical or spurious energy source needed on small scales!
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 15
Decaying 3D turbulence, Fort Sill storm, 05/20/77.
NMM-B, Ferrier microphysics, 1km, 32 levels,112km by 112km by 16.4km, double periodic.
Spectrum of w2 at 700 hPa, hours 3-4 average.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
-4.5 -4.0 -3.5 -3.0 -2.5 -2.0
k -̂5/3 ferrier700180240
-5/3
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 16
Eta NMM
Verification
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 17
Eta NMM
Mountain waves in the Eta (left) and the NMM (right)
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 18
951.72 mbObs. 953 mb
http://www.nhc.noaa.gov/2003isabel.shtml
“Isabel”
NMM 8 km, 60 lev.
GFS data
Sea levelPressure
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 19
NMM 8 km, 60 lev.GFS data3 hourAccumulation
NCAR geometrical progression color code
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 20
OTHER PHYS1OTHER PHYS2WRF NMM PHYS1WRF NMM PHYS2
AlaskaOctober 2002Vector windAll verification times
Large scale errors clustering by dynamical cores, implications for ensembles!
● NCEP standard verification package, verification against observations.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 21
OTHER PHYS1OTHER PHYS2WRF NMM PHYS1WRF NMM PHYS2
WestFebruary 2003Vector windAll verification times
Large scale errors clustering by horizontal discretization!
● NCEP standard verification package, verification against observations.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 22
☞ Mesoscale Convective Systems, NSSL-SPC Spring Program 2004
● NMM WRF, 4.5 km, 35 levels, about Central domain, Ferrier microphysics, no parameterized convection, Eta initial and boundary conditions, 00Z, up to 30 hours, available in the morning.
● NMM starting from Eta data needs up to 6 hours to spin up convective systems.
NMM
NCAR,CAPS
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 23
24 hour 4.5 km forecast of 1 hour accumulated precipitation valid at00Z April 21, 2004 (better than 12 hour forecasts by operational models).
Verifying 2 km radar reflectivity. Courtesy Jack Kain.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 24
Radar,2 km
NCAR EMC,4 km, 35 lev.,Lin
NCEP NMM,4.5 km, 35 lev.,Ferrier
NSSL/SPC2004 SpringProgram
04/05/28, 00Z ∼ +24 hours
Courtesy:Jack Kain,Steve Weiss
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 25
☞ Conclusions
● Promising results of high resolution NWP on meso scales.
● NWP on near-cloud scales successful more frequently and with stronger signal than if only by chance.
● Convective systems direct circulations spun-up by the model, predictable?
● Reemphasized importance of forcing and (micro)physics on meso scales?
● Full potential of mesoscale NWP not yet developed.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 26
Sep 04,2003RMS H500 RMS RH700
W est 6 18 30 42 W est 6 18 30 42Central 12 24 36 48 Central 12 24 36 48
East 6 18 30 42 East 6 18 30 42Alaska 12 24 36 48 Alaska 12 24 36 48
9 w ins NMM 7 w ins NMM6 w ins Eta 5 w ins Eta1 w in Other 4 w ins Other
RMS T850 RMS V250
W est 6 18 30 42 W est 6 18 30 42Central 12 24 36 48 Central 12 24 36 48
East 6 18 30 42 East 6 18 30 42Alaska 12 24 36 48 Alaska 12 24 36 48
11 w ins NMM 9 w ins NMM5 w ins Eta 7 w ins Eta0 w ins Other 0 w in Other
56.25% NMM35.94% Eta7.81% Other
● NCEP standard verification package, verification against observations.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 27
Jan 24,2004, 1 month snapshotRMS H500 RMS RH700
West 6 18 30 42 West 6 18 30 42Central 12 24 36 48 Central 12 24 36 48
East 6 18 30 42 East 6 18 30 42Alaska 12 24 36 48 Alaska 12 24 36 48
12 wins NMM 1 win NMM2 wins Eta 15 wins Eta2 wins Other 0 wins Other
RMS T850 RMS V250
West 6 18 30 42 West 6 18 30 42Central 12 24 36 48 Central 12 24 36 48
East 6 18 30 42 East 6 18 30 42Alaska 12 24 36 48 Alaska 12 24 36 48
5 wins NMM 10 wins NMM9 wins Eta 1 win Eta2 wins Other 5 wins Other
43.75% NMM42.19% Eta14.06% Other
● NCEP standard verification package, verification against observations.
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 28
Radar,2 km
NCAR EMC,4 km, 35 lev.,Lin
NCEP NMM,4.5 km, 35 lev.,Ferrier
NSSL/SPC2004 SpringProgram
04/06/01, 00Z ∼ +24 hours
Courtesy:Jack Kain,Steve Weiss
Zavisa Janjic 50 Years of NWP, June 14-17, 2004 29