[MITgcm-devel] free drift ice model

[Martin Losch]

Hi,
to add to the confusion. I find that EVP does make a difference  
(performance wise) for spinup integration when you tracer time step  
is on the order 12h or so. EVP requires something like  
SEAICE_delatTevp=60s for my 2deg config, but this is already more  
efficient than LSR if the global timestep  
(deltaT=deltaTmom=deltaTtracer) = 3600s. (I use 18 cpus/tiles).

M.
On 30 Nov 2006, at 17:28, Dimitris Menemenlis wrote:

> Patrick, just to make sure I did not confuse you and Matt,
>
>> Re. points 1., 2.
>> I am planning on doing a similar ECCO-GODAE run as the v3.i0 (with  
>> new growth),
>> now using the C-grid EVP with suggested time-stepping.
>> But need to get some other runs through queue first...
>
> points 1 and 2 were mutually exclusive, that is, "SEAICE_deltaTdyn  
> = 3600" is only applicable when LSR solver is used.  For EVP to  
> work you need short time steps, which is why it may be attractive  
> for Matt's setup.  For ECCO-GODAE integrations, you are probably  
> best using LSR.
>
> D.
>
> -- 
> Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
> Jet Propulsion Lab, California Institute of Technology
> MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099
> tel: 818-354-1656;  fax: 818-393-6720
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel