[MITgcm-support] useOBCSsponge

Matthew Mazloff mmazloff at MIT.EDU
Sun Aug 14 23:36:43 EDT 2005


Actually it does look like W is calculated on the open boundary if you 
use Orlanski boundary conditions.  I am not familiar with this option 
but it shows it can be done.

-Matt

Matthew Mazloff wrote:

> Hi Dmitri,
>
> I have used sponge layers in hydrostatic runs without any problems.
>
> Actually, I don't think (as the model is currently coded) you can 
> prescribe W on the open boundary for non-hydrostatic runs.  I would 
> think the model would crash at runtime if you tried; there is no W 
> file name in obcs_readparms.F
>
>       NAMELIST /OBCS_PARM01/
>      &          OB_Jnorth,OB_Jsouth,OB_Ieast,OB_Iwest,
>      &          useOrlanskiNorth,useOrlanskiSouth,
>      &          useOrlanskiEast,useOrlanskiWest,
>      &          OBNuFile,OBNvFile,OBNtFile,OBNsFile,
>      &          OBSuFile,OBSvFile,OBStFile,OBSsFile,
>      &          OBEuFile,OBEvFile,OBEtFile,OBEsFile,
>      &          OBWuFile,OBWvFile,OBWtFile,OBWsFile,
>      &          useOBCSsponge, useOBCSbalance, useOBCSprescribe
>
> It actually appears that W on the open boundary is set to zero in 
> non-hydrostatic runs.  I am not sure why the model forces W to zero on 
> the boundaries and maybe someone else can fill us in.  I am guessing 
> there is some technical reason or I have missed something and am just 
> wrong about this.
>
> After thinking about it, I don't think forcing w=0 on open boundaries 
> is extremely significant in ocean modeling.  It is definitely not the 
> most realistic way to deal with the boundary; you are correct that it 
> may cause sharp gradients in W.  But whether these unphysical 
> gradients are significant is another question.  For ocean modeling 
> with reasonable surface forcing, large W values are likely only to 
> occur in regions near solid boundaries where vorticity generated by 
> the vortex stretching associated with the vertical velocity can be 
> dissipated, or in fronts where the isopycnals are steeply inclined.  
> If the model adequately resolves these boundary and frontal regions, 
> unphysical gradients may have a large effect.  Open boundaries, 
> however, occur mostly in the open ocean (and hopefully not in fronts) 
> where it is likely that W is small.  If W is small next to the open 
> boundary, a zero W on the open boundary is likely insignificant.  If 
> the sponge layer is causing W and W gradients to be significant in the 
> ocean interior, I would guess the prescribed open boundary condition 
> is not very physical.  Perhaps this is why noone has taken the time to 
> find a way to calculate (restore) W in open boundary (sponge layer) 
> regions.  Just a guess.  Of course, I do not know what experiment you 
> are running.  Are you doing something where W is significant near your 
> open boundary and needs to be prescribed?  If so, I think you will 
> need to do some coding and i hope someone else can advise you on the 
> best way.
>
> -Matt
>
>
>
> Dmitri Leonov wrote:
>
>> Hello,
>>
>> Is the "useOBCSsponge" option ready to be used: in general and with 
>> non-hydrostatic dynamics in particular?
>>
>> Only u and v (also t and s but not w) are relaxed over 
>> 'spongeThickness' grid points.
>> But in the non-hydrostatic mode, w is prescribed at the boundary. 
>> That may cause a sharp gradient of w (?)
>> Does it mean that some modifications are needed (adding another 
>> subroutine to obcs_sponge.F, to be called from CALC_GW)?
>>
>> Thanks,
>> Dmitri
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>
>------------------------------------------------------------------------
>
>_______________________________________________
>MITgcm-support mailing list
>MITgcm-support at mitgcm.org
>http://mitgcm.org/mailman/listinfo/mitgcm-support
>  
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20050814/e8025d39/attachment.htm>


More information about the MITgcm-support mailing list