MOM6 fails if Coriolis force switched off

[MHBalsmeier]

I am using MOM6 for short-term predictions of the surface current at high resolution.

During experimentation I found that MOM6 fails immediately after the initialization if the Coriolis force is switched off via OMEGA = 0.0 (commenting this line leads to a successful run),

How to reproduce the issue.
Attached you find all files needed to reproduce the issue.

I am at commit bf0bba1 on Ubuntu 24.

Disclaimer: I am new to MOM6 and do not fully understand all the options yet, so this might totally be a problem in MOM_input. I also do not know if this is the correct repo for this issue. Feel free to move it if it is not.

issue_cold_start.zip

[adcroft]

I tried your config with floating-point exceptions enabled; there are some divisions by Omega (or "f") in some of the sub-grid parameterizations. The division by f is normally paired with another term that would normally avoid the denominator vanishing but in your config those other terms also seem to be zero. For instance setting GUST_CONST=1e-6 (smallish number) gets past one. However, there's a calculation TKE_to_Kd that relies on either stratification and rotation and for some reason it has no stratification. While those division-by-zeros are clearly edge cases we should account for, it seems the easiest thing to do is keep the planet rotating. Just because you are doing short-term prediction doesn't mean rotation doesn't matter. At high-resolution you might want to turn off the mixed-layer restratification which is one of the parameterizations that assumes non-zero rotation.

[MHBalsmeier]

@adcroft Thanks for your detailed reply, much appreciated!
For my operational setups I for sure keep the planet rotating. I am just experimenting with different setups (wind forcing on/off, tides on/off, Coriolis on/off) and encountered this error and found it worth reporting.
For now I can of course do my experiments with OMEGA set to a smallish value for example.
For the future it would be indeed great if this could be fixed.

[MHBalsmeier]

@adcroft I have found a similar problem in coupled mode. I wanted to see what happens when starting from a state at rest and only turning on the tides (with Coriolis force switched on).
Running

mpiexec -np 4 ../build/ice_ocean_SIS2/MOM6

fails with number of barotropic step (nstep+nfilter) is 0.
I am aware I can test this case with ocean_only mode, but again, I found it worth reporting.
Starting from a state at rest and only turning on the tides works with the above config file, but not in coupled mode. Setting GUST_CONST=1e-6 also did not solve it.
Appended you find everything to reproduce it.

mom6-coupled-init.zip