[MITgcm-support] MITgcm-support Digest, Vol 175, Issue 14

[Jody Klymak]

> On Feb 1, 2018, at  3:56 AM, Hadar Berman <hadarberman at gmail.com> wrote:
> 
> Thanks for your reply. I am only running for a few time steps so I did not save too many outputs. The offline is not interpolating between the saves, but reading exactly (I checked) the velocities from the output of the online model. So The velocities are exactly the same and the ptracer initial value is exactly the same. The differences between the advection terms of the offline vs. online model are very small but still exist and I was wandering what can cause this?

So again, just to be clear, you saved the velocity at every deltaT of the online model?  

Cheers,   Jody


> 
> Thanks again,
> Hadar.
>> 
>> 
>>> On 30 Jan 2018, at 07:50, Hadar Berman <hadarberman at gmail.com> wrote:
>>> 
>>> Dear MITgcm users,
>>> 
>>> I am working on a regional model, using the ptracers, kpp and obcs packages, advection scheme 33 and implicit diffusion. I am comparing an online and offline version, and I see differences in the ptracer concentration when running with the same time step in both offline and online runs. The velocities read by the offline model are exactly identical to those in the online one (meaning that I am starting at the right time step by using offlineTimeOffSet=-deltaT/2). 
>>> The differences arise already after one time step, small differences can be seen in the diagnostic ADVx/y/rTr01 and larger ones in the diagnostic DFrITr01. Why would there be any differences at all if my ptracer and velocities are identical?
>>> I can provide more information or data files if needed.
>> 
>> Are the velocities actually equal?  You must have saved a lot of velocity data, otherwise the offline version is interpolating between your saves, whereas the online version is using the new velocity calculated every time step.  
>> 
>> But maybe I?m misunderstanding.
>> 
>> Cheers,   Jody
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support