Avoid OBC initialisation bug introduced in upstream PR1668

[dougiesquire]

Yesterday I received the following email from Theresa Cordero at GFDL:

Hi Dougie and Mike,

I wanted to let you know that I have found a bug that has been impacting our simulations with generic tracers and OBCs. I am not sure if either of you are running configurations like that or if you are aware of anyone at your institutions who is.

The problem changes were pulled into mom-ocean as a part of a recent PR from dev/gfdl: mom-ocean#1668
specifically the changes from the "+Rotated and simpler OBC initialization" PR that was included.

If you are running configurations with generic tracers and OBCs you may have noticed your answers changed after those code modifications were made. If that is the case, I recommend reverting your MOM6 version to before the commits in the "+Rotated and simpler OBC initialization" PR. Some changes in that PR incorrectly initialize the generic tracer OBCs.

Bob and I have a plan to fix the bug which I will share once it is available.
Apologies for letting this bug through onto main. We have been discussing how to enhance our testing at GFDL to limit these sorts of problems in the future.

Best Regards,
Theresa Cordero
Research Physical Scientist
Marine Ecosystems and Downscaling Division
Geophysical Fluid Dynamics Laboratory
[she/her/hers]

Our 2025.08 branch and 2025.08.000 release tag includes this bug. The latter was intended to be used in the next ACCESS-OM3 release. However, in light of Theresa's email, we have decided to do as she suggests.

The "+Rotated and simpler OBC initialization" PR to the NOAA-GFDL fork is here. The first commit in that PR was rebased and merged into dev/gfdl as commit NOAA-GFDL@3beb3a8.

I will set up a 2025.07 branch that includes commits up to, but not including 3beb3a8.