[MITgcm-devel] possible problem with sigmaR and shelfice

[Martin Losch]

OK, I'll wait until the next checkpoint.
M.

On Dec 16, 2011, at 4:29 PM, Jean-Michel Campin wrote:

> Hi Martin,
> 
> Will change the sigmaR masking in grad_sigma.F ; I also agree
> to change calc_ivdc the way you propose, but was wondering if
> we should not make a checkpoint before (was going to make one
> soon, since there have been enough changes since the last one).
> 
> Cheers,
> Jean-Michel
> 
> On Thu, Dec 15, 2011 at 01:41:50PM +0100, Martin Losch wrote:
>> Hi there,
>> 
>> I have stumble across a potential problem with sigmaR. It's computed as maskC(k)*(sig(k)-sig(k-1)) in grad_sigma.F or zero for k=1. For ice shelves we can have k>0 and sigmaR does not make sense at the interface between ice and ocean, but usually that should not be problematic, but for example in gmredi_calc_tensor, sigmaR is vertically averaged to give dSigmaDr at the cell centers, and here the wrong sigmaR is used (and then implicitly for slope computations, etc.)
>> In the same way calc_ivdc.F will compute the wrong IVDConvCount at the interface.
>> 
>> I am not sure if this really causes problems, but I suggest to compute sigmaR like this maskC(k)*maskC(k-1)*(sig(k)-sig(k-1)) instead, and likewise modify calc_ivdc (and use sigmar in that routine as suggested by the comment in do_ocean_phys.F?). What do you think?
>> 
>> I do not see any effects on testreport (but there are effects with ice shelves and ivdc_kappa>0).
>> 
>> Martin
>> 
>> _______________________________________________
>> MITgcm-devel mailing list
>> MITgcm-devel at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel