[MITgcm-devel] recent changes in model/scr

[Jean-Michel Campin]

Hi Martin,

I would try to comment out:
> exactConserv = .true.,
This should return to previous results.

I am still not yet done with this option (momStepping=F & exactConserv=T)
so might need to add some stop for unsafe options combination.

Cheers,
Jean-Michel

On Tue, Nov 22, 2011 at 01:21:01PM +0100, Martin Losch wrote:
> Hi Jean-Michel,
> 
> I have a simple channel configuration with ice, where I start from some initial velocity field and  basically turn off the ocean dynamics (mom/temp/saltStepping = .false.), and let the ice evolve. This configuration is used in a manuscript (Losch+Danilov, I told you about it, it will appear soon in Ocean Modelling, finally). 
> Ironically, now that this manuscript is finally accepted, I cannot reproduce the results with the latest code (the model blows up very quickly). I think I have isolated the change that did it. With checkpoint63e things still work, but after changing integr_continuity (I updated starting from checkpoint63e, timestep.F calc_g[s,t,w].F thermodynamics.F adams_bashforth?.F the_main_loop.F, then diags_rho.F do_oceanic_phys.F set_parms.F config_summary.F (and PARMS.h), no problem, but then momentum_correction_step.F initialise_varia.F integr_continuity.F, forward.F), the problem starts.
> 
> Where does my configuration (with mom/temp/saltStepping = .false.) care about changes in the ocean? I have no clue, do you?
> 
> Martin
> 
> # ====================
> # | Model parameters |
> # ====================
> #
> # Continuous equation parameters
>  &PARM01
>  tRef = 3*-1.9,
>  sRef = 3*34.,
>  viscAr=1.E-3,
>  viscAh=0.E4,
>  diffKhT=0.0,
>  diffKrT=3.E-5,
>  diffKhS=0.0,
>  diffKrS=3.E-5,
>  rhonil=1035.,
>  gravity=9.81,
>  eosType = 'MDJWF',
>  f0 = 0.0e-4,
>  beta = 0.,
>  implicitDiffusion=.TRUE.,
>  implicitViscosity=.TRUE.,
> #useOldFreezing=.TRUE.,
>  exactConserv = .true.,
> #useRealFreshWaterFlux=.true.,
>  useRealFreshWaterFlux=.false.,
> #convertFW2salt = -1.,
> #useCDscheme=.TRUE.,
>  useNHMTerms=.TRUE.,
> # turn on looped cells
>  hFacMin=.1,
>  hFacMindr=20.,
> # set precision of data files
>  readBinaryPrec=64,
>  momStepping = .false.,
>  tempStepping= .false.,
>  saltStepping= .false.,
> # use a flux-limiting advection scheme to avoid overshoots
> #tempAdvScheme = 77,
> #saltAdvScheme = 77,
>  useSingleCPUio = .true.,
>  debuglevel=-1,
>  &
> 
> # Elliptic solver parameters
>  &PARM02
>  cg2dMaxIters=500,
>  cg2dTargetResidual=1.E-13,
>  &
> 
> # Time stepping parameters
>  &PARM03
>  niter0 = 0,
> #niter0 = 864000,
> #niter0 = 36000,
> #ntimeSteps = 864000,
>  ntimeSteps = 10,
> # 100 years of integration will yield a reasonable flow field
> #startTime  =          0.,
> #endTime    = 31104000.E1,
> #endTime    = 62208000.E2,
> #deltaT    = 600.,
>  deltaT = 3600.0,
>  abEps = 0.1,
>  forcing_In_AB  = .false.,
>  pChkptFreq= 311040000.,
>  dumpFreq=   864000.,
>  dumpFreq=   0.,
>  dumpInitAndLast = .false.,
>  monitorFreq=0.,
>  dumpFreq=   1.,
>  monitorFreq=1.,
> # restoring time scale of only 12 days = 60/5, because surface
> # layer is only 10m as opposed to 50m in the 4x4 deg-run.
> #tauThetaClimRelax = 1036800.,
> # restoring time scale of only 36 days = 30*6 days/5
> #tauSaltClimRelax = 3110400.,
>  periodicExternalForcing=.TRUE.,
> # these should not have any effect, when exf is used
> # still I am not allowed to set them to zero
>  externForcingPeriod=2592000.,
>  externForcingCycle=31104000.,
>  &
> 
> # Gridding parameters
>  &PARM04
>  usingCartesianGrid=.TRUE.,
>  usingSphericalPolarGrid=.FALSE.,
>  delR= 3*1000.,
>  dySpacing=5.e3,
>  dxSpacing=5.e3,
>  &
> 
> # Input datasets
>  &PARM05
>  bathyFile=      'topog.strait8',
> #hydrogThetaFile='lev_t.bin',
> #hydrogSaltFile= 'lev_s.bin',
>  uVelInitFile= 'u10.init',
>  &
> 
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel