[MITgcm-support] OBCs over topography!

[Martin Losch]

Van Thinh,

you should make it a habit to describe you  problems more concisely.  
Statements such as "I change the something and the model doesn't work"  
do not provide any information at all:
What does not work? Does the result look funny/noise, does the model  
blow up, after how many timesteps? Along with your increased number of  
vertical gridpoints, how did you change the vertical grid SPACING? As  
you can see, I can only take wild guesses at what you are doing.

The flow over bump problem involves significant vertical velocities, as  
you might guess. If (but only if) you are actually decreasing your  
vertical grid spacing to 4m (as your numbers imply), you may have run  
into a CFL-problem (w*deltaT/deltaZ >1), which you could easily check  
in the monitor output statistics (advcfl_wvel and advcfl_W_hf), e.g,  
with 4m and a time step of 600sec, you violate the vertical  
CFL-criterium with vertical velocities of 4e-4m/s. If these CFL-values  
explode BEFORE the model blows up, then that's your problem and the  
only remedy is to decrease your time step.

Hope that helps, and please describe your problems.

Martin

On Oct 21, 2005, at 5:30 PM, Van Thinh Nguyen wrote:

> Hi!
>
> As I told you in last email, I would like to simulate internal waves  
> due
> to a tidal flow over a depression Gaussian bump in a open channel at
> West/East. I did a lot test cases, I have found that the MITgcm model  
> in
> this case is quite sensitive with the vertical mesh density. My water
> depth is 120m (80m plat + 40m depth of depression Gausian bump). when I
> apply the verical grid delZ=20*6, it worked fine. when I tried to  
> increase
> delZ=30*4, it didn't work. I also tried to change hFacMin but it  
> doesn't
> help.
>
> Could someone please give a hint to fix this problem?
>
> Thanks so much!
>
> Van Thinh
> ---------------
>
>
> On Thu, 20 Oct 2005, Matthew Mazloff wrote:
>
>> Hi Van Thinh!
>>
>> You can give all cells and the masks will take care of it.  (e.g.
>> OB_Ieast=80*-1,OB_Iwest=80*1 is fine)
>>
>> Other things I have learned:
>> 1) You should make the topogrophy flat right at the obcs.  (e.g. make
>> the height at (i,j) = (i+1,j) for the western boundary)
>> 2) Do not use the CD scheme with obcs
>>
>> -Matt
>>
>>
>> Van Thinh Nguyen wrote:
>>
>>> Hi
>>>
>>> I'm confused how to set up the boundary condition at a OB where a
>>> topography intersects with the grid block (please see my figure  
>>> enclosed).
>>> I would like to simulate internal waves due to a tidal flow over a
>>> depression Gaussian bump in a channel, which has OBCs at West and  
>>> East,
>>> North & South are closed sides. It looks like the geometry in exp4,  
>>> the
>>> difference is only the bump (my case is depression bump!).
>>> Model is 3D, but the Fig. enclosed is a (X,Z) section at middle of  
>>> bump.
>>>
>>> At West/East OBs, topography intersects grid block at cell 55 (Z  
>>> direction
>>> with 80 cells). It means:
>>>
>>> - from free surface to cell 55 is OB
>>> - from cell 55 to the bottom is closed
>>>
>>> In this case, should I define cells (1-55) are OBCs in data.obcs or  
>>> I can
>>> define all cells at West/East are OBCs (with OB_Ieast=80*-1,
>>> OB_Iwest=80*1)
>>>
>>>
>>> Could someone please give a hint for this?
>>>
>>> Thanks so much!
>>>
>>> Van Thinh
>>>
>>> --------------------------------------------------------------------- 
>>> ---
>>>
>>> _______________________________________________
>>> 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