[MITgcm-devel] eosType ... POLY3 vs. MDJWF

[Martin Losch]

Hi Patrick,
I am using MDJWF as the default EOS (unless I want a linear EOS) and I 
didn't have any problem so far (after this bug in find_alpha had been 
fixed).
I successfully used it for, e.g.,
- global_ocean.90x40x15 (it ran for at least ten years with KPP, have 
not tried without KPP)
- globa_ocean.90x40x15 plus dynamic seaice-package and KPP (1000 years 
of integration)
- global_ocean.cs32x15 with dynammic seaice-package and KPP (1000 years 
of integration)
- very high resolution non-hydrostatic convection experiments 
(dx=dz=10m)

The only possible reason for instabilities could be that with this 
eosType=MDJWF and JMD95p, the true pressure is used instead of rho*g*z 
(lagged by one timestep). So you could try JMD95p and see whether you 
have the same problems (in fact I used JMD95p for all of my 
Boussinesq/non-Boussinesq experiments and never had any problems with 
that).

JMD95z uses p=g*rho*z, which, according to Dewar etal (1998, JPO 28(4), 
577ff), can cause errors of order 3Sv in the Gulf Stream, for example.

Martin

On Wednesday, May 26, 2004, at 05:34 AM, Patrick Heimbach wrote:

> Hi there,
>
> how much experience is there using the "new"
> EOS a la MDJWF?
> Martin, did u use it extensively?
> Dimitris, are u using it for the cubed sphere?
> Jean-Michel?
> It seems to me the model gets quite unstable
> when using it, but I may be missing something
> on how to use it properly.
> I think, even the global_ocean.90x40x15
> crashes fairly soon when using MDJWF instead of POLY3.
> I played around a bit with diffusivities to see
> whether it can be compensated, not sure it can.
> Any input would be welcome.
>
> Thanks
> -Patrick
>
> _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
> Patrick Heimbach     Massachusetts Institute of Technology
> FON: +1/617/253-5259                    EAPS, Room 54-1518
> FAX: +1/617/253-4464               77 Massachusetts Avenue
> mailto:heimbach at mit.edu                 Cambridge MA 02139
> http://www.mit.edu/~heimbach/                          USA
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-devel