[MITgcm-support] how to control obcs balance flow

[Martin Losch]

I think in a previous email I wrote, that you need double precision (real*8) to make sure that your properly balanced boundary flow is not offset by numerical truncation. Even with real*8 you still have numerical round-off (but that should not have a large effect).
So, if your boundary flow is properly balanced, real*8 will help with your eta problem.

I can’t really do more that repeat the text:
“-1” means that the net flow through the particular boundary corrected to be 0.
“0.” means that there is no correction to this boundary
a value >0 sets the relative size of the correction. 
In your case with a strong current entering in the south and exiting in the east, these boundaries should not have “-1”, because that will remove the flow.
If you don’t want to have any flow accross N and W boundaries then these should get -1. If you think that there should be some flow across those boundaries, you should set them 0. or larger.
If you then set the S and E boundaries both to 1, then the correction that balances the net flow will be split equally between both boundaries (provied that the N and W boundaries have -1 or 0). If set one to 0 and the other to 1, you will have the correction applied only to one of the boundaries, etc.
Alternatively you can decide, that you Kuroshio in- and outflow should not be modified at all, but all imbalance should be compensated by flow accross one of both of the other two boundaries, then you set S and E to 0., and have values > 0 for the N and/or W boundaries.

Martin

On May 1, 2014, at 7:31 PM, 李志远 <oceanlizy at 163.com> wrote:

> 
> From: LIZHI<oceanlizy at 163.com>
> To: mitgcm-support at mitgcm.org
> Subject: Re: [MITgcm-support] how to control obcs balance flow ?
> 
> 
> Hi,Martin ,thank you for your kindness to help me solve my problems . but I still have some questions ,so I can not but to bother 
> 
> you again .
> 
>     when I use precision="real*4" in data.obcs ,  the eta of model results is abnormally too high or too low after a long run , even I have balanced the inflow and outflow offline according to the way you told me before . Then I changed the data precision to double precision "real*8" , the new model is running , I did not check the output yet . Will this be of some help for the eta result ?  Because this eta result is crucial for my study ,I have to adjust it to a normal value .
> 
> I see the document at " http://mitgcm.org/public/r2_manual/latest/online_documents/node236.html ", I am not so clear about what do the value of -1,1,2,0,mean  below  :
> 
> OBCS_balanceFacE = 1.,
> 
> OBCS_balanceFacW = -1.,
> 
> OBCS_balanceFacN = 2.,
> 
> OBCS_balanceFacS = 0., 
> 
>    if such values are defined in data.obcs , the strong inflow at each open boundarys will be removed ? But in my model,the kuroshio current flow into south open boundary and flow out at east open boundary, and such a strong in flow has great influence on my model fields .so, I think in my case , remove such a inflow at boundarys will not be a good choice. So ,I am puzzled how to set such values to make sure the eta results are not so abnormal ,and the balance flow will not change the strong inflow at all boundarys . 
> 
> Thank you ,sir !
> 
> Yours LI ZHI
> 
> Have a good vacation  !
> 
>  
> 
> 
> 
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support