Top layer thickness & flux distribution

[dgwyther]

As discussed in Gwyther 2021, layer thickness has a big impact on melt rates; the method for mixing melt fluxes downwards is also important for melt rates. We compared ROMS, COCO and MPAS-O, but ISOMIP+ has more implementations. We should at least include some discussion of how the other models implement model z_top thickness/model boundary layer parameterisation, and the implications for melting.

[xylar]

Another relevant paper for this discussion is Scott et al. 2023. They show convergence with high vertical resolution (and without tuning heat and salt transfer coefficients) to a lower melt rate, consistent with ROMS. They assert that this is consistent with the constant background diffusivity prescribed by the experimental design (i.e. lack of enhanced turbulent mixing as one would presumably have in a resolved boundary layer).

Would be fun to discuss this further both in the Hackathon and in the paper.