[MITgcm-devel] seaice water drag

[Martin Losch]

Hi Jean-Michel,

I was afraid that you would prefer option (2) (o: 
I agree it’s the best (but also most complicated) way of dealing with this, I’ll just have to pick good default values.
I’ll come up with a PR ...

Martin

> On 9. Jun 2020, at 17:46, Jean-Michel Campin <jmc at mit.edu> wrote:
> 
> Hi Martin,
> 
> Some thoughts:
> option 4 is not good.
> option 3 is better but:
>  a) since we have very little experience in realistic set-up with threshold==0, it's a bit risky.
>  b) we would like a way to reproduce old results.
> option 1 : I don't like it too much, because this value of 0.25 is a bit arbitrary.
>   I think the problem repported on support is an "extreme" case, but we may found
>   in other situations that this 0.25 threshold is too large.
> I prefer option (2). The default threshold can still be zero, if we want, but then
> we would have to change (many) data.seaice or (many) output.txt or a mix of both.
> 
> Cheers,
> Jean-Michel
> 
> On Mon, Jun 08, 2020 at 05:06:18PM +0200, Martin Losch wrote:
>> Hi there,
>> 
>> This
>> <http://mailman.mitgcm.org/pipermail/mitgcm-support/2020-May/012503.html>
>> support question is triggered by a bug in seaice_oceandrag_coeffs.F
>> The linear drag DWATN (CwatC in seaice_oceandrag_coeffs.F) is, probably for historical reasons, never below 0.25. I guess this is so because older versions of the implicit solver could not deal with zeros on the diagnogal and having a non-zero DWATN made sure that this never happend. Now the solver can deal with zeros on the diagnogal (also because of the scaleSurfStress parameter), but this part of the code is a leftover.
>> 
>> Changing the lower threshold of DWATN from 0.25 to 0. breaks all verification experiments, and I guess all setups out there, so I am a little hesitant to modify the code accordingly. I see these options:
>> 
>> (1) compute DWATN/CwatN only if SEAICE_waterDrag > 0. This will preserve all setups with non-zero waterDrag, but allows to have zero ice-ocean drag by setting SEAICE_waterDrag = 0 in data.seaice.
>> (2) replace the hard coded threshold of 0.25 by a runtime flag that defaults to 0.25 for SEAICE_waterDrag > 0 and to 0 for SEAICE_waterDrag = 0. This is probably the most flexible solution, but it add additional confusion about the plethora of unclear runtime flags.
>> (3) just change the threshold to zero and live with the broken setups
>> (4) do not do anything (I don???t think so).
>> 
>> Opinions are appreciated.
>> 
>> Martin
>> 
>> 
>> _______________________________________________
>> MITgcm-devel mailing list
>> MITgcm-devel at mitgcm.org
>> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-devel
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-devel