adjust_salt uses a positive hard-coded minimum salinity

[Hallberg-NOAA]

When adjust_salt (a routine in MOM_diabatic_aux.F90) is used to deal with water with a negative (or exessively small) salinity, it uses a hard-code minimum salinity of 0.01 ppt (or psu).

Negative salinities are obviously very bad, but I think that a minimum value of 0 should be fine. In any case the minimum salinity should be a run-time configurable value. This particular code can be traced back to the initial check-in of the first version of MOM6, so there is no record from the person who first introduced this (I could have been that person!) was thinking. The capability in adjust_salt is important if we use a sea-ice model with a specified positive ice salinity (which is not necessary with SIS2). Does anyone recall why we set this minimum salinity to 0.01 instead of 0?

[MJHarrison-GFDL]

You and I did this together back in the prehistoric age! I'm not sure which one of us chose this value, but it was certainly arbitrary.

[Hallberg-NOAA]

This fixed value will be converted to a runtime parameter (MIN_SALINITY) when https://github.com/NOAA-GFDL/MOM6/pull/920 is completed. We will then test using 0 and consider changing the default to 0 in due course.

[Hallberg-NOAA]

The default value for MIN_SALINITY was changed to 0 with commit NOAA-GFDL@613a82e, and this was merged into the main branch of dev/gfdl with PR #1127. This issue is now considered resolved.