Upload
aizza
View
29
Download
0
Embed Size (px)
DESCRIPTION
Cube-based atmospheric GCMs at CSIRO: reversible staggering. John McGregor CSIRO Marine and Atmospheric Research Aspendale, Melbourne MetOffice, Exeter 24 October 2012 Acknowledgements: Marcus Thatcher and Martin Dix. Outline. CCAM formulation VCAM formulation Some comparisons Plans - PowerPoint PPT Presentation
Citation preview
CSIRO Marine and Atmospheric Research
Cube-based atmospheric GCMs at CSIRO: reversible staggering
John McGregor
CSIRO Marine and Atmospheric ResearchAspendale, Melbourne
MetOffice, Exeter 24 October 2012
Acknowledgements: Marcus Thatcher and Martin Dix
CSIRO Marine and Atmospheric Research 2
Outline
• CCAM formulation• VCAM formulation• Some comparisons• Plans
At Newton Institute
Sorted out VCAM advection
Solved “grid imprinting” problem
Cured 2-grid-point convection noise
CSIRO Marine and Atmospheric Research 3
Major problem in dynamical cores is how to provide suitable velocity components for the Coriolis terms, so as to give good geostrophic adjustment.
The approach in CCAM and VCAM is based on “reversible staggering” of velocity components.
CSIRO Marine and Atmospheric Research 4
OriginalSadourny (1972)
C20 grid
Equi-angular C20 grid
Alternative cubic grids
Conformal-cubicC20 grid
CSIRO Marine and Atmospheric Research 5
The conformal-cubic atmospheric model
• CCAM is formulated on the conformal-cubic grid
• Orthogonal• Isotropic
Example of quasi-uniform C48 grid with resolution about 200 km
CSIRO Marine and Atmospheric Research 6
CCAM dynamics
• atmospheric GCM with variable resolution (using the Schmidt transformation)
• 2-time level semi-Lagrangian, semi-implicit• total-variation-diminishing vertical advection• reversible staggering
- produces good dispersion properties• a posteriori conservation of mass and moisture
CCAM physics• cumulus convection:
- mass-flux scheme, including downdrafts, entrainment, detrainment
- up to 3 simultaneous plumes permitted• includes advection of liquid and ice cloud-water
- used to derive the interactive cloud distributions (Rotstayn 1997) • stability-dependent boundary layer with non-local vertical mixing• vegetation/canopy scheme (Kowalczyk et al. TR32 1994)
- 6 layers for soil temperatures- 6 layers for soil moisture (Richard's equation)
• enhanced vertical mixing of cloudy air• GFDL parameterization for long and short wave radiation
CSIRO Marine and Atmospheric Research 7
Location of variables in grid cellsAll variables are located atthe centres of quadrilateralgrid cells.
However, during semi-implicit/gravity-wave calculations, u and v are transformed reversibly to the indicated C-grid locations.
Produces same excellent dispersion properties asspectral method (see McGregor, MWR, 2006), but avoids any problems of Gibbs’ phenomena.
2-grid waves preserved. Gives relatively lively winds, and good wind spectra.
CSIRO Marine and Atmospheric Research 8
Reversible staggering
Where U is the unstaggered velocity component and u is the staggered value, define (Vandermonde formula)
•accurate at the pivot points for up to 4th order polynomials
•solved iteratively, or by cyclic tridiagonal solver
•excellent dispersion properties for gravity waves, as shown for the linearized shallow-water equations
| X | X * | X |m-1 m-1/2 m m+1/2 m+1 m+3/2 m+2 m+3/4
CSIRO Marine and Atmospheric Research 9
Dispersion behaviour for linearized shallow-water equations
Typical atmosphere case- large radius deformation
N.B. the asymmetry of the R grid response disappears by alternating the reversing direction each time step,giving the same response as Z (vorticity/divergence) grid
Typical ocean case- small radius deformation
CSIRO Marine and Atmospheric Research
Transformation of 2, 3, 4, 6-grid waves
10
CSIRO Marine and Atmospheric Research
Treatment of ps advection near terrain
Pressure advection equation
Define an associated variable, similar to MSLP
which varies smoothly, even over terrain. It is thus suitable for evaluation by bi-cubic interpolation, whilst the other term is found “exactly” by bi-linear interpolation (to avoid any overshooting effects). Formally, get
11
CSIRO Marine and Atmospheric Research
Treatment of T advection near terrainSimilarly to surface pressure advection, define an associated variable
which varies relatively smoothly on sigma surfaces over terrain. Again the second term can be found “exactly” by bi-linear interpolation. A suitable function is
Formally, get
This technique effectively avoids the requirement for hybrid coordinates.
12
CSIRO Marine and Atmospheric Research
Helmholtz solver
3-colour scheme used for SOR solution of Helmholtz equations(conformal octagon grid has 4-colour scheme)
By suitable manipulation, the SLSI leads to a set ofHelmholtz equations for each of the vertical modes, on a 5-point stencil.The Helmholtz equations may be solved by simple successive over-relaxation. A vectorized solution is achieved by solving successively on each of the following 3 sets of sub-grids.A conjugate-gradient solver is also available, and is usually used.
13
CSIRO Marine and Atmospheric Research 14
MPI implementation
Remapping of off-processor neighbour indices to buffer region
Indirect addressing is used extensively in CCAM - simplifies coding
Original
Remapped region 0
CSIRO Marine and Atmospheric Research 15
Typical MPI performance
Showing both Face-Centred and Uniform decomposition for global C192 50 km runs, for 1, 6, 12, 24, 48, 72, 96, 144, 192, 288 CPUs (strong scaling example)
VCAM slightly slower, but is still to be fully optimised
CSIRO Marine and Atmospheric Research 16
An AMIP run 1979-1995
CCAM
Obs
Tuning/selecting physics options:• In CCAM, usually done with 200 km AMIP runs, especially paying
attention to Australian monsoon, Asian monsoon, Amazon region• No special tuning for stretched runs
DJF JJA
CSIRO Marine and Atmospheric Research 17
Variable-resolution conformal-cubic grid The C-C grid is moved to locate panel 1 over the region of interestThe Schmidt (1975) transformation is applied•this is a pole-symmetric dilatation, calculated using spherical polar coordinates centred on panel 1•it preserves the orthogonality and isotropy of the grid•same primitive equations, but with modified values of map factorPlot shows a C48 grid (Schmidt factor = 0.3) with resolution about 60 km over Australia
CSIRO Marine and Atmospheric Research 18
C48 8 km grid over New Zealand
C48 1 km grid over New Zealand
Grid configurations used to support Alinghi in America’s Cup Also Olympic sailing for Beijing and Weymouth (200 m)
Schmidt transformation can be used to obtain even finer resolution
CSIRO Marine and Atmospheric Research 19
Downscaled forecasts60 km
8 km
1 km
When running the 8 km simulation, a digital filter is used to diagnose large-scale and fine-scale fields. The large-scale fields are then inherited every 3 hours from the 60 km run.
CSIRO Marine and Atmospheric Research
Miller-White nonhydrostatic treatment
Being a semi-Lagrangian model, CCAM is able to absorb the extra phi terms into its Helmholtz equation solver, for “zero” cost
The new dynamical core (VCAM) uses a split-explicit treatment, so the Miller-White treatment would need its own Helmholtz solver, so may need another treatment for VCAM
20
CSIRO Marine and Atmospheric Research
CCAM simulations of cold bubble, 500 m L35 resolution, on highly stretched global grid
21
CSIRO Marine and Atmospheric Research 22
Gnomonic grid showing orientation of the contravariant wind components
Illustrates the excellent suitability of the gnomonic grid for reversible interpolation – thanks to smooth changes of orientation
CSIRO Marine and Atmospheric Research
New dynamical core for VCAM - Variable Cubic Atmospheric Model
• uses equi-angular gnomonic-cubic grid - provides extremely uniform resolution - less issues for resolution-dependent parameterizations
• reversible staggering transforms the contravariant winds to the edge positions needed for calculating divergence and gravity-wave terms
• flux-conserving form of equations– preferred for trace gas studies– TVD advection can preserve sharp gradients– forward-backward solver for gravity waves (split explicit)– avoids need for Helmholtz solver– linearizing assumptions avoided in gravity-wave terms
23
CSIRO Marine and Atmospheric Research 24
CSIRO Marine and Atmospheric Research 25
Horizontal advection
Flow=qyVj+1/2
Vj-1/2
ucovUi-1/2
vcov
(qx, qy)
q
Flow=qxUi+1/2
Transverse components (included in both low/high order fluxes)calculated at the centre of the grid cells (loosely following LeVeque)
Low-order and high-order fluxes combined using (lively) Superbee limiter
Cartesian components (U,V,W) of horizontal wind are advected
CSIRO Marine and Atmospheric Research 26
Problem caused by spurious vertical velocities at vertices!
Eastwards solid body rotation in 900 time stepsUsing superbee limiter
CSIRO Marine and Atmospheric Research
| X | X * | X |m-1 m-1/2 m m+1/2 m+1 m+3/2 m+2 m+3/4
27
Usual pivot velocity (in terms of staggered u) isum+3/4 = (um+1/2+um+3/2)/2
In terms of unstaggered U, it isUm+3/4 = (2Um+6Um+1)/8
But adjacent to panel edge it is better to useUm+3/4 = (-Um-1 + 3Um + 7Um+1 - Um+2)/8
which is derived by using an estimate for Um+1/2 provided by averaging 1-sided extrapolations of U. These extrapolations will be very accurate for velocities such as solid body rotation
Improved treatment of pivot points at panel edges
edge
CSIRO Marine and Atmospheric Research
Reduction of “grid imprinting”
28
Spurious vertical velocities reduced by factor of 8 by improved calculation of pivot velocities near panel edges
CSIRO Marine and Atmospheric Research 29
With better staggered velocities at panel edges (avoiding the spurious vertical velocities)
CSIRO Marine and Atmospheric Research 30
Solution procedure
• Start loop
Nx(t/N) forward-backward loop Stagger (u, v) +n(t/N) Average ps to (psu, psv) +n(t/N) Calc (div, sdot, omega) +n(t/N) Calc (ps, T) +(n+1)(t/N) Calc phi and staggered pressure gradient terms, then unstagger these Including Coriolis terms, calc unstaggered (u, v) +(n+1)(t/N) End Nx(t/N) loop
Perform TVD advection (of T, qg, Cartesian_wind_components) using average ps*u, ps*v, sdot from the N substeps
Calculate physics contributions
• End loop
Main MPI overhead is the reversible staggering at each substep, but this just needs nearest neighbours in its iterative tridiagonal solver.
Message passing is also needed in the pressure gradient and divergence calcs
CSIRO Marine and Atmospheric Research
500 hPa omega (avg. Jan 1979)
Hybrid coordinates introduced
Non-hybrid
31
CSIRO Marine and Atmospheric Research
Example of effect on rainfall of hybrid coordinates
32
Hybrid coords
Non-hybrid coords
32
CSIRO Marine and Atmospheric Research
Avoidance of 2-grid rainfall over Indonesia
33
Small Indonesian islands provide convective forcing at 2-grid length scale – was a persistent feature.Problem solved by spreading the convective heating over the forward-backward time steps.
CSIRO Marine and Atmospheric Research
Absence of noise problems
• Some groups report noise problems with split-explicit methods, requiring diffusive suppression methods
• No divergence damping or other noise suppression is needed in VCAM, thanks to- use of reversible staggering (N.B. significant noise is seen
if simple interpolation of velocity components is used in Coriolis terms)
- hybrid coordinates
- application of convective heating over the forward-backward time steps
34
CSIRO Marine and Atmospheric Research
250 hPa windsin a 1-year run- similar
35
VCAMCCAM
CSIRO Marine and Atmospheric Research 36
DJF JJA
VCAM1-year
CCAM1-year
Same physics
ObsClimate(rainfall
& MSLP)
CSIRO Marine and Atmospheric Research 37
Comparisons of VCAM and CCAM
VCAM advantages• No Helmholtz equation needed• Includes full gravity-wave terms (no T linearization needed)• Mass and moisture conserving• More modular and code is “simpler”• No semi-Lagrangian resonance issues near steep mountains• Simpler MPI (“computation on demand” not needed)
VCAM disadvantages• Restricted to Courant number of 1, but OK since grid is very
uniform• Some overhead from extra reversible staggering during sub
time-steps (needed for Coriolis terms)• Nonhydrostatic treatment will be more expensive
CSIRO Marine and Atmospheric Research
Tentative conclusions
• Reversible staggering works well for both CCAM and VCAM
• VCAM seems to perform better than CCAM in the tropics- better rain over SPCZ and Indonesia
- possibly by avoiding linearizing ps term in pressure gradients, and better gravity wave adjustment by not using semi-implicit
- rainfall needs improving over China
38
CSIRO Marine and Atmospheric Research 39
• CABLE canopy/carbon scheme has been included• mixed-layer-ocean available• aerosol scheme added• urban scheme added• TKE boundary scheme available• new GFDL radiation scheme available• new version on gnomonic grid (VCAM), in flux-conserving
form (being able to achieve conservation is another advantage of stretched global models) – now working
• coupling to PCOM (parallel cubic ocean model) of Motohiko Tsugawa from JAMSTEC - underway (3:way: CSIRO + JAMSTEC + CSIR_SouthAfrica)
Model developments
Equi-angular gnomonic C20 grid
CSIRO Marine and Atmospheric Research 40
Thank you!