[MITgcm-support] OBCS for sea ice!

[Martin Losch]

Hallo Liqun,

to begin with, make sure that SEAICE_maskRHS=.FALSE. (this is the default). I introduced this option mainly for testing purposes, and it should not be used otherwise (and it probably does not work with the open boundary conditions anyway).

In the default case of SEAICE_maskRHS = .FALSE., the corresponding code in seaice_dynsolver.F is not used and seaiceMaskU/V are basically the same as the velocity masks maskW/S(:,:,1,:,:) for the surface ocean velocities. The masks are 2D and apply to the entire domain, and not only to individual boundaries. (and they are only used in the ice dynamics). I assume that your "problem" stems from a different source. 

Also the wrap-around for AREA (and all other variables) is intentional. The boundary conditions are cyclic/periodic by default and only putting a strip of "land" or an open boundary overrides this behavious. 

Judging from your figure (it was difficult to open, to be honest, something like a jpg or png would have been much better), you put your boundaries in the middle of the Arctic, which I would call an unfortunate choice. First of all, open boundary conditions are always problematic, because they are a discontinuity for the dynamics. In particular, you cannot expect too much from the open boundary implementation for seacie, because they were introduced to have a way to remove ice from the domain, e.g. think of a typical pan-Arctic domain, with an open boudary outside the usual ice cover, and the occasionally the ice might reach the boundaries and this case, you want the ice to be able to move out of the domain rather than getting stuck there. Honestly, I am surprised that the solution (at least the plot that you sent) does not appear to have serious issues, except near the boundary, where I would expect them. The noise near the coasts at approximate 125W is probably a consequence of your setting seaiceMaskU/V=1 everywhere (I would revert to the original code).

For more help on your specific problem you need to give us more information: version/checkpoint of the code, details of the configuration (what do you use in SEAICE_OPTIONS.h, OBCS_OPTIONS.h, data.obcs, data.seaice, etc). The problem description is also a little unclear.

Martin


On Jun 8, 2013, at 8:06 AM, <liqun_hu818 at yahoo.com.cn> wrote:

> Hi, all:
>     I am now trying setup a regional ice-ocean coupled application with MITgcm checkpoint64h, OBCs are prescribed from global model products. 
>     However, there is some problem with the Vvel at the southern boundary and the Uvel at the western boundary. These two boundary value are masked by the northern and eastern sea ice velocity mask. I checked the obcs_apply_uvice.F, and commented the seaiceMaskU/V, the problem still exist. And so I changed the value of seaiceMaskU/V to 1 in the seaice_init_varia.F, this kind problem could be solved.
>     After checking the code, I found that seaiceMaskU(i,Jsouth)=AREA(i,Jsouth)+AREA(i,Jsouth-1), and if Jsouth=1, it seems that the AREA(i,0) is set to be AREA(I,end), simila with the western boundary with Iwest=1. 
>     Though the seaiceMaskU/V are mainly used in the dynamic solver, could there be other problems?
> Best wishes,
> liqun
>  
> 
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support