[MITgcm-devel] obcs & ptracers

[Martin Losch]

I was wrong. The reason, why the pickups do not work for obcs and 
ptracers in the way I though, is this:
In obcs_calc.F the "default" OBCS for ptracers is different from that 
of T and S: while the latter two are set to tRef and sRef respectively 
(so to the same values at every time step), ob?ptr = ptracer(just 
inside of boundary, so dependent on current time step). If you read 
ptracers-obcs from a file for all boundaries (so that the "default" is 
never used), my simple 5+5=10 (so 5 times as complicated as your test) 
passes, even with obcs_apply_ptracer in obcs_init_variables (after I 
have fixed another bug in obcs_external_fields_load.F that I found just 
now, oops).

The quick solution is to not use obcs_apply_ptracer (and 
obcs_apply_ts,uv, either), if we start from a pickup. (we do need them, 
when we start from 0, because then the obcs are not known to the 
initial fields). So I will (first fix the embarrasing bug that I have 
found) and then add an if statement that avoids calling the obcs_apply 
routines if we are starting from a pickup.

Martin

On Dec 14, 2005, at 4:26 PM, Martin Losch wrote:

> Hi Jean-Michel,
>
> I have to admit, that I never tested the restarts with obcs and 
> ptracers. I think you do need the to apply the open boundary 
> conditions at the very beginning, but not really after restarts (I was 
> probably to lazy to think about that). But the error is a different 
> one: there are exchanges missing for the ptracers (they are there for 
> uVel, vVel, theta, and salt). This is still the old problem, that in 
> obcs_prescribe_read, the "exchanges" for vertical slab are missing, so 
> that the obcs for ptracers are not correct on the halos. I should 
> probably look into that very soon, but I just don't have the time at 
> the moment :-(
>
> Martin
>
> On Dec 14, 2005, at 5:39 AM, Jean-Michel Campin wrote:
>
>> Hi,
>>
>> I found that when I comment out the CALL OBCS_APPLY_PTRACER
>> in obcs_init_variables.F, the differences vanish.
>> And now I understand why it's like this.
>>
>> Do we really need to apply the OBCs just after reading
>> a pickup file ?
>>
>> Jean-Michel
>>
>> On Tue, Dec 13, 2005 at 10:56:17PM -0500, Jean-Michel Campin wrote:
>>> Hi Martin,
>>>
>>> I run my favorite test (1+1 = 2) on exp4, and it
>>> works well for the dynamics part, but I get differences
>>> for the ptracer 1.
>>>
>>> I don't see clearly the reason ;
>>> When I decide not to prescribe the ptracer1 on
>>> the western OB (comment out OBWptrFile(1) = 'OBzonalS.bin',),
>>> I also get differences.
>>>
>>> Did you test the restart with obcs & ptracers
>>> in similar configs ?
>>>
>>> Thanks,
>>>
>>> Jean-Michel
>>> _______________________________________________
>>> 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
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel