[MITgcm-devel] seaice adjoint and EVP

[Martin Losch]

Hi Dimitris, Jinlun,

I'll check in this change, once it proves to be a solution to our  
problems, okay? I'll have to test this in my runs an see, whether  
this causes any problems for me.

EVP and LSR do have slightly different solutions so I am so surprised  
that things happen for evp but not for lsr. Naturally these inlets  
are always a problem should be avoided (no advection in an out  
possible). I guess it's a good idea to have the flooding stuff turned  
on be default. (there are a few other defaults that I would like to  
see, but I need to revisit that, monday)

Let me know how it goes with zmin = 0.

Martin

On 17 May 2007, at 03:38, Dimitris Menemenlis wrote:

> Martin, as you guessed, EVP model blow-up appears to be unrelated  
> to stripes. Model crashes because snow thickness > 10^5 m in a  
> small inlet:
> http://ecco2.jpl.nasa.gov/data1/arctic/output/tests/evp/ETAN.jpg
> http://ecco2.jpl.nasa.gov/data1/arctic/output/tests/evp/HSNOW.jpg
>
> Weird that the LSR integration does not have the same problem in  
> the Arctic domain but I have seen that happen around Antarctica  
> with LSR.  Turning on SEAICEuseFlooding=.true. fixes this  
> particular problem.
>
> So I will set zMin=0, turn on SEAICEuseFlooding=.true., and restart  
> test integrations.  Will you check in the zMin=0 change?
>
> D.
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel