Melt potential does not have pressure dependence

[Hallberg-NOAA]

MOM6 has the option to calculate the ocean melt potential (i.e., the heat content of the seawater in a near-surface region relative to the freezing point) within extract_surface_state(), but the code that does this does not include the pressure dependence of the freezing point, instead using a pressure of 0 (effective 1 atmosphere?) the the freezing point calculations. Under sea-ice ice using a pressure of 0 for the freezing point calculation is not such a bad approximation, but under ice shelves or icebergs such an assumption is wildly inaccurate.

It would be easy enough to implement this pressure dependence for the calls to extract_surface_state() that occur at the end of step_MOM, as the pressure exerted by the ice (and atmosphere) are provided within the forcing or mech_forcing types. However there are also calls to extract_surface_state() that occur from within the various drivers at the end of the ocean model initialization, and the ocean surface pressure from the atmosphere, sea-ice, icebergs and ice shelves are not provided to these calls. If the pressure dependency were to be included in the melt potential calculation within extract_surface_state(), retaining reproducibility across restarts would seem to require significant changes in the arguments to the ocean model initialization calls and the overall sequence of the calls to various ocean and ice components for any cases that use the melt potential. A more straightforward approach might be to separate out the melt potential calculation from extract_surface_state() into a separate routine that can be called by the drivers that choose to use the melt potential, with the surface pressure being an optional argument that would be provided in cases that use the pressure dependent freezing point calculation for the melt potential calculation. This simpler approach would not impose any changes on drivers that do not use the melt potential.

The melt potential is not used with the FMS coupler, and it does not appear to be used in other configurations. As such, I would be interested in hearing the opinions of anyone who is using the melt potential on how to handle this issue. In particular, @gustavo-marques originally added the melt potential code for use with some coupled models using the MCT coupler.