[MITgcm-devel] Re: downslope pkg

[Martin Losch]

Hi Jean-Michel,

basically I agree with you. I had a potiential control variable  
"R_shelfiice" and an adjoint model in mind, where I thought it would  
be convenient  to have these things separate.
Now I had another close look at the ctrl_ini_depth/update_masks_etc  
and I think it's safe to go ahead as you suggest.

Also, you are right and the upper BC is different from the lower one.  
Actually, initially I had thought to implement the entire stuff with  
the help of the surface following coordinates z*, but this would have  
meant pressure gradient errors and I would have had to deal with  
them. However, this route would allow a dynamic ice shelf on top of  
the ocean. We should keep this option in mind (even if it means  
coding higher order gradient schemes for horizontal pressure).

If you agree, I'll create a new file shelfice_update_masks.F, in  
which I combine shelfice_update_masks and shelfice_ini_depth and  
clean up ini_depth.

But I'll wait for your approval before I do that.
Sorry for the confusion,
Martin

On 11 Sep 2008, at 03:43, Jean-Michel Campin wrote:

> Hi Martin,
>
> In summary, we agree on:
> * split of shelfice_init in 2: init_fixed & init_varia is a good thing
> * not good to have those 2 (hidden) S/R :
>   SHELFICE_INI_DEPTH & SHELFICE_UPDATE_MASKS
>   in file shelfice_init_varia.F
>
> Remaining question: what to do with those 2 subroutines:
> given that they don't fit into "standard" pkg initialisation
> (read_params/init_fixed/init_varia)
> a) keep 2 separate S/R. or
> b) move SHELFICE_INI_DEPTH (reading of SHELFICEtopoFile into  
> R_shelfIce array)
>  at the beggining of SHELFICE_UPDATE_MASKS, so that we end up with  
> only 1 S/R.
>
> I prefer option (b):
> - it minimizes the number of non-standard shelfice entry in the
>   initialisation part.
> - don't see the upper BC (ice-shelf base) as being so much related  
> to ini_depth
>   (dealing with fixed lower BC). In fact, from a dynamical point of  
> view,
>   it's more the mass of the ice-shelf which would be the upper BC.  
> And would
>   probably have to deal with mass if a dynamical ice-shelf component
>   was considered.
>   The code is not ready for that, but since the link to ini_depth  
> is not
>   so clear, the simpler (just 1 call) might be better.
> - none of those S/R is very long, sothat it's little bit easier to  
> follow
>   what's going on all together in 1 S/R.
> but if you insist on having 2 S/R (in the same file or not), it's  
> not a
> big deal (as long as they are not hidden with a 3rd S/R, it's fine  
> with me).
>
> Cheers,
> Jean-Michel
>
> On Wed, Sep 10, 2008 at 06:09:19PM +0200, Martin Losch wrote:
>> I agree, that putting the new subroutines all into  
>> shelfice_init_varia.F
>> is a bit stupid and I am willing to change that, but in the end I  
>> felt it
>> was more consistent to read the topography, where all topographies  
>> are
>> read, and modify the hfacs in a different routine. Maybe we talk  
>> about
>> this some more tomorrow, need to mow the lawn now ...
>>
>> Martin
>>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel