[MITgcm-devel] Re: [MITgcm-support] obcs_apply_ptracer

[Martin Losch]

OK, I can do this:
1. change the default to OBCSfixTopo=.true.
2. override the default in exp4 (OBCSfixTopo=.false. in data.obcs + a  
comment in this file, why we do this) in order to reproduce the old bug
3. update all other experiments that are affected (dome, seaice_obcs,  
internal_wave, tutorial_plume_on_slope)

Do it now (tomorrow) or wait for Dimitris and the seaice_obcs  
experiment? Other option is to use obcsfixtopo=.false. for  
seaice_obcs for now until he fixes the sensitivity/timestepping scheme?

Martin


On 24 Apr 2008, at 16:59, Jean-Michel Campin wrote:

> Hi Martin,
>
> Am i right regarding exp4, that this is less an issue since
> the gradients of bathy near obc regions are small (=> much less
> that 1 level => no "step") + uses free slip.
> I would then suggest to keep this one with OBCSfixTopo=.FALSE.
> (with the comment: for backward compatibility).
> And otherwise, kind of agree with Dimitris, we could put as a default
> OBCSfixTopo=.TRUE., as long as we have a way to reproduce old bugs ...
>
> Jean-Michel
>
> On Thu, Apr 24, 2008 at 06:31:00AM -0700, Dimitris Menemenlis wrote:
>> Martin, thanks!  I will definitely set OBCSfixTopo=.TRUE. for  
>> seaice_obcs,
>> since I need to change output.txt anyways.  It seems to me that  
>> this switch
>> should be .TRUE. by default and .FALSE. only if it is needed to  
>> preserve
>> backward compatibility with existing verification experiments.  D.
>>
>>
>>> I have now "fixed" the problem that taka had; and that fix should  
>>> also fix
>>> the problem that you were having with seaice because my fix will  
>>> put land
>>> on
>>> the boundary (online), where there is land just inside the domain  
>>> (that's
>>> what we concluded to do at our secret MAD/NSA conference call).
>>>
>>> All: I suggest that we modify at least one of the verification  
>>> experiments
>>> to
>>> use this fix (OBCSfixTopo is false by default). Which one do you  
>>> guys
>>> suggest, maybe even seaice_obcs? Or should I change the default.  
>>> We didn't
>>> talk about that on the phone.
>> _______________________________________________
>> 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