[MITgcm-support] Linear internal waves evolution under a free surface and a rigid lid.

[Jean-Michel Campin]

Hi Sergey,

Did you try commenting out "tempImplVertAdv=.TRUE.," ?
The thing is I am not sure how well the implicit vertical
advection is doing with a highly non-linear advection scheme
like tempAdvScheme=77,

You could also try to set:
 cg2dUseMinResSol=1,
in 2nd namelist (PARM02) of the main parameter file "data".
This can help to pick the cg2d solution corresponding to the
minimum residual (instead of the last iteration solution when
it does not converge). But I am not sure that this will help you
since it's seems to be the 3-D solver that first shows some 
convergence problems.

And happy new year to you !
Cheers,
Jean-Michel

On Tue, Dec 31, 2013 at 07:44:35AM +0400, Fancer Lancer wrote:
> Dear MITgcm,
> 
> *First of all I want to wish everyone a happy new year, let all your plans
> come successfully true, models converge, researches turn out well!*
> 
> And here is my question.
> I am trying to run a few setups of quasi-monochromatic internal waves in a
> two-layers environment with very sharp transition layer. The target of the
> experiments is to research a transformation process of linear internal
> waves over a sharp bottom obstacle. The linearity of the waves means very
> small amplitudes of the disturbances comparing with a depth. So to speak we
> calculate a small amplitude quasi-sinusoidal internal waves on a tiny
> pycnocline over a sharp bottom obstacle. Technically there is no pycnocline
> at all, but the transition between two layers is sharp.
> That setup converges only if I specify the free surface, in case of the
> rigid lid the system suddenly blows up.The monitor output shows a small
> fluctuation of rhsMax in the middle of calculations:
>  cg2d: Sum(rhs),rhsMax =  -1.00649220260016E-04  2.80476672625558E-08
>  cg3d: Sum(rhs),rhsMax =  -4.36678712900714E-03  2.73727850080651E-11
>  cg2d: Sum(rhs),rhsMax =  -1.00473220760270E-04  2.80986212383509E-08
>  *cg3d: Sum(rhs),rhsMax =  -1.02540856095068E-06  1.16576772289106E-07*
>  cg2d: Sum(rhs),rhsMax =  -9.74440421204186E-05  2.81449166440040E-08
>  cg3d: Sum(rhs),rhsMax =  -4.22056718404259E-03  2.75142884748998E-11
>  cg2d: Sum(rhs),rhsMax =  -1.00470277023852E-04  2.81813454193756E-08
>  cg3d: Sum(rhs),rhsMax =  -4.35243360164479E-03  2.75448882646779E-11
> ...
> And after several time-steps the model blows up:
>  cg3d: Sum(rhs),rhsMax =  -2.95322925916202E-15  1.00918335046307E+15
>  STOP GAD_IMPLICIT_R: error when solving 3-Diag problem
> That doesn't happen if the free surface is specified. I didn't find any
> problem with CFL, but a surface elevation turns to an enormous value during
> the calculations:
> (PID.TID 0000.0001) %MON dynstat_eta_max              =
>  -1.6182199742936E+01
> (PID.TID 0000.0001) %MON dynstat_eta_min              =
>  -1.6182199808065E+01
> ...
> (PID.TID 0000.0001) %MON dynstat_eta_max              =
>  -1.2229692791603E+01
> (PID.TID 0000.0001) %MON dynstat_eta_min              =
>  -1.2229692856491E+01
> ...
> ...
> (PID.TID 0000.0001) %MON dynstat_eta_max              =
>  -4.3612306317841E+04
> (PID.TID 0000.0001) %MON dynstat_eta_min              =
>  -4.3612307712367E+04
> 
> I don't know how it affects my calculations, since the surface is blocked
> by a rigid lid.
> Could you please tell me what is wrong here and give some guide how to fix
> the problem?
> (data file is in attachment)
> 
> Sincerely,
> -Sergey


> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support