[MITgcm-support] obcs/exf interactions

[Alistair Adcroft]

Several answers to several questions:

I've sorted out other descrepancies between the runs and now have 
established that the line appears due to the old set-up; when running 
for longer the line is visible in the surface flow in level 1 of the old 
set-up but not visible deeper down until later. It is not apparent in 
pressure. I suspect it is a difference in stress near the ice-edge. The 
discrepancy between the runs is detectable earlier on when looking at 
the difference maps, which is what I was doing at first. I suspect the 
bugs that DM mentioned are in the older run.

In the longer runs, the straight line does not remain straight for very 
long.

I didn't use the pickups because the checkpoint.F seems very different 
and I'm not sure if they are compatible; unfortunately we don't have 
version numbers in the pickup files and the version of pickup file 
involves the tag name on the main branch and not ecco branch. We'll have 
to sort that out if you want to use existing pickup files; I believe 
some recent changes to pickups will make that harder than it sounds. 
We'll need JMC's help on that front.

A.



Tom Haine wrote:
> Jinlun, Alistair,
> 
> Barotropic modes are forced by windstress variability so I'm not very
> surprised there's evidence in 300m velocities of the ice edge. In some
> animations I occasionally see hints of it. 
> 
> Alistair, are your figures from iteration 41400 or so?  I guess you
> start from iter=0 when the ice edge is much sharper (yes, I realize I'm
> the source of all that's not beautiful in the MITgcm!). Maybe this
> explains your result?  If you start from iter=41400 maybe your velocity
> anomalies will have the smooth, natural-looking curves of my ice edge at
> that time?
> 
> We should clearly favor the new pkg/seaice solver.
> 
> Tom.
> 
> 
> On Thu, 2004-10-07 at 15:42, Jinlun Zhang wrote:
> 
>>Tom,
>>The results look good to me, no straight line here. Ice model runs ok even with some
>>bugs. Not sure why a straight line would show up at 300m and looks like using the
>>bug-fixed solver would not help.
>>Jinlun
>>
>>
>>Tom Haine wrote:
>>
>>
>>>Jinlun,
>>>
>>>Attached are figures of various surface fields, forcing fields, and ice
>>>fields from close to the time Alistair is starting his integrations.
>>>These results are from the old code.
>>>
>>>I guess pkg/exf changes (presumably in the stress computation) might
>>>explain the difference Alistair sees, although I don't know why they
>>>would be concentrated at the ice edge.
>>>
>>>Tom.
>>>
>>>
>>>On Thu, 2004-10-07 at 14:32, Jinlun Zhang wrote:
>>>
>>>>I am puzzled by the straght line after some time steps. The initial ice
>>>>condition is a straght line, but after a while it should change. Here is my
>>>>guess:
>>>>(1) Ice model might not be excuted so the initial ice edge is kept unchanged.
>>>>Need to look at ice output.
>>>>(2) Different ocean model versions might get surface flux differently over areas
>>>>partially covered by ice (exf to blame?)
>>>>Jinlun
>>>>
>>>>Tom Haine wrote:
>>>>
>>>>
>>>>>Dear Jinlun,
>>>>>
>>>>>I don't know. Alistair's plots are differences between our code (hacked
>>>>>ecco_c50_e29 with 2/2003 pkg/seaice) and the latest HEAD release. The
>>>>>plots he showed are difference plots at (guessing) 300m after a few
>>>>>steps. There is a difference in 300m velocity at the position of the ice
>>>>>edge at that time. My guess is that it persists but I don't know for
>>>>>sure. The ice is moderately thick (about 1m) and dense northwest of the
>>>>>line you see (my ugly attempt to put seaice on the shelf). Elsewhere
>>>>>there is no seaice.
>>>>>
>>>>>Of course, I don't know which velocity field is better (do you have a
>>>>>preference?). But it would be nice to locate the origin of the
>>>>>discrepancy.
>>>>>
>>>>>Tom.
>>>>>
>>>>>On Thu, 2004-10-07 at 13:15, Jinlun Zhang wrote:
>>>>>
>>>>>>Tom,
>>>>>>
>>>>>>OK you are using the LSR solver.  I don't think the solver has been changed
>>>>>>since 2/2003,  only some cosmetic changes. Do you see the ice edge just at
>>>>>>initial 1 or 2 time steps or the ice edge persists?
>>>>>>
>>>>>>JInlun
>>>>>>
>>>>>>Tom Haine wrote:
>>>>>>
>>>>>>
>>>>>>>Hi Jinlun,
>>>>>>>
>>>>>>>We use lsr:
>>>>>>>
>>>>>>>      SEAICEuseLSR = .TRUE.,
>>>>>>>
>>>>>>>in data.seaice. Without this line (i.e., with the adi solver) it doesn't
>>>>>>>work in parallel (following discussion on this list in April this year).
>>>>>>>
>>>>>>>We're using pkg/seaice from around February 2003. I think several of the
>>>>>>>key routines have been updated since then. Can this account for the
>>>>>>>velocity differences Alistair sees?
>>>>>>>
>>>>>>>Tom.
>>>>>>
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support


-- 
Dr Alistair Adcroft            http://www.mit.edu/~adcroft
MIT Climate Modeling Initiative        tel: (617) 253-5938
EAPS 54-1624,  77 Massachusetts Ave,  Cambridge,  MA,  USA