[MITgcm-support] Unstructured open boundaries

[Martin Losch]

Hi Estanislao,

I found your email in the online archives: <http://mailman.mitgcm.org/pipermail/mitgcm-support/2019-September/012194.html>

I think I understand the problem now, and your solution is the correct one, i.e. logically there should be flux through the interface between i=2 and 3 at j=2.

It maybe worth extending section "8.3.1.4. Defining open boundary positions” in the manual to give a more complex example like this. It would be great if you could start doing that (since you already spent time and though on this) and create a pull request. The instructions for that can be found here: <https://mitgcm.readthedocs.io/en/latest/contributing/contributing.html#contributing-to-the-manual>
and more generally section 5 <https://mitgcm.readthedocs.io/en/latest/contributing/contributing.html#contributing-to-the-mitgcm>. It’s not hard at all. Once you’ve created the pull request, we can work on this together.

Martin

> On 9. Oct 2019, at 04:34, Estanislao Gavilan Pascual-Ahuir <dramauh at hotmail.com> wrote:
> 
> Hi Martin,
> 
> I wrote an email two weeks ago about this matter with a diagram. If you did not get it to try to check the junk box, for some reason the mails from mitgcm are finishing in my junk box. I found out that the obsc_check file has issues when there is a boundary point at the corner. In those cases, the model thinks that there are two boundaries (e.g. upper and lateral). Even if the west and east boundary indexes are set 0, the model sends you an error related with the mask. Martin, do you think is it worth to write a small section about this in the manual?
> 
> Kind regards,
> 
> Estanislao