[MITgcm-devel] SEAICEuseEVP and SEAICE_deltaTevp

[Martin Losch]

Usually 120 subcycling time steps should be OK, as long as SEAICE_elasticParm is order 1/3.
But deltaT/120 will not work when deltaT is not divisible by 120 (at least the code will stop).
You might want to use more or less sub-cycling time steps, depending on your problem. 

I am totally against making EVP easy to use, because EVP requires that you know what you are doing (remember all the noise in the solutions), and the longer I am dealing with this model, the more I come to the conclusion most users do not know what they are doing. I think that the current implementation is fine, and that people just need to read the documentation (and giving users an excuse not to read the documentation is just plainly idiotic, we have enough annoying questions at mitgcm-support as it is).

And in reply to Chris: I do not know how anyone can set SEAICEuseEVP, unless he/she has hacked seaice_readparms.F. SEAICEuseEVP is not part of the seaice-namelist(s).

I do not think that having to set two parameters to turn on parts of the code is not very useful either. I do not see the point of making SEAICEuseEVP a runtime parameter.

Am I being stubborn?

Martin

On Jul 26, 2011, at 2:57 PM, Menemenlis, Dimitris (3248) wrote:

> Martin, what formula do you recommend for estimating a reasonable deltaTevp?
> 
> Under what circumstances would deltaT/120 be an unreasonable choice?
> 
> Dimitris Menemenlis
> 818-625-6498
> 
> On Jul 25, 2011, at 11:49 PM, "Martin Losch" <Martin.Losch at awi.de> wrote:
> 
>> It's not very clear, which value to use for deltaTevp. deltaT/120 is a reasonable choice (and often recommended), but by no means universal.
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel