[MITgcm-support] BC's for flow in a box

Martin Losch Martin.Losch at awi.de
Tue Feb 24 10:52:03 EST 2009


Hi David,

the obcs-code is actually not so difficult, but it's quite hard to  
read at first, because things are often done more than once, and  
values are overwritten (and I am going to add a level to that soon (o:)

The simplest (and least flexible) approach to use OBCS is to make a  
local copy of obcs_calc.F, somehow generate the obc values for u,v,T,S  
in this file and then let the model apply them, along with a sponge- 
layer if required. If you don't mondify obcs_calc.F The default values  
for u,v are zero; the default values for T and S are Tref and Sref.
So if you don't do anything to the code and have some parameter like  
this in data.obcs:
&OBCS_PARM01
   OB_Jnorth=40*-1,
   OB_Jsouth=40*1,
   OB_Ieast = 30*-1,
   OB_Iwest = 30*1,
   useOBCSsponge=.true.,
   useOBCSprescribe=.false., (this is the default)
&

&OBCS_PARM02
&

&OBCS_PARM03
   Urelaxobcsinner = 2592000,
   Urelaxobcsbound = 86400,
   Vrelaxobcsinner  = 2592000,
   Vrelaxobcsbound = 86400,
   spongeThickness = 4,
&
You'll prescribe Tref for T, Sref for S, u=0 and v=0 along all open  
boundaries with a sponge layer of 4 grid points, the restoring  
timescale is 1day at the boundary, linearly increasing to 30days at  
the "inner" edge of the sponge-layer. If you don't want a sponge for T  
and S, you'll have to edit obcs_sponge (and set lambda_obcs_s/t=0), if  
you don't want to prescribe T/S along the boundary, you'll have to  
think of something or use the Orlanski OBCS (but that will also  
recompute OB?u/v). In the end you'll probably have to edit obcs_calc  
to suit your needs.

Martin



On Feb 24, 2009, at 3:24 PM, David Hebert wrote:

> Hi Martin,
>
> Thanks for the update. I do want a sponge layer for U/V. One thing I  
> am having trouble finding in the code is does this require  
> specifying boundary velocities of zero via OBWufile, OBWvfile, etc?  
> or by setting indices via OB_I and OB_J does the model set boundary  
> to zero unless otherwise specified with OBCS prescribe?
>
> Thanks,
>
> David
>
> Martin Losch wrote:
>> David,
>> closed boundaries with a sponge layer for for T/S are straight  
>> forward with the RBCS package, velocities are not affected by this  
>> package, using this package is quite straightforward.
>>
>> If you want a sponge layer for U/V, then you need to use OBCS and  
>> you need to put your boundaries inside the domain, that is at 1,nx,  
>> and 1,ny (OB_Jsouth=nx*1, OB_Jnorth=nx*-1, OB_Iwest=ny*1,  
>> OB_Ieast=ny*-1) and then use the parameters OBCSuseSponge, U/ 
>> Vrelaxinner/bound
>>
>> Martin
>>
>> On Feb 24, 2009, at 3:07 AM, David Hebert wrote:
>>
>>> I would like to simulate a flow similar to an experimental setup  
>>> of stratified fluid in a tank. In this configuration I would like  
>>> to specify "closed" boundaries with a sponge layer. In order to do  
>>> this, do I need to use OBCS prescribe to specify zero U,V at the  
>>> boundaries (and corresponding T,S)? Or is it possible (and easier)  
>>> by not specifying any open boundary indicies (that is, specify  
>>> OB_Jnorth=nx*0, OB_Iwest=ny*0, etc.).
>>>
>>> Thanks,
>>>
>>>
>>> David
>>> _______________________________________________
>>> 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




More information about the MITgcm-support mailing list