When running the 30thdegree MOM6-EAC configuration, the output masks shown in ocean_geometry['wet'] and *.ocean_static['deptho'] are different to the input land_mask and topog files (see attached screenshots of TAS for example). The outputs have unexpected land cells scattered in various spots.
I expect that when you created the ocean mask, it used a sea level of 0: any non-zero depth will be water (maybe as you expect). The MINIMUM_DEPTH parameter in MOM6 is essentially doing masking itself; anything shallower than the minimum depth will be masked out. Those regions that differ between the inputs and the outputs are where the topography is less than 10-ish metres, which are water according to the tool that generated the masks, but land according to the model.
It looks like the smallest non-zero in your topography is 1m. I guess with the water being so shallow there, you might see that cell dynamically wetting/drying. Other than that, there could be some extreme values there. An alternative, if you’re happy with 10m being the minimum depth would be to instead of bumping the model’s minimum depth down, re-generate your ocean_mask, passing --sea_level 10 to the make_quick_mosaic tool and make the two consistent that way.
And a follow up question: with the wetting-drying capability of MOM6, does it really matter if some very shallow regions are masked? Will the model take-care of that naturally, or are their implications that I’m missing?