shortwave modification for mushy thermo layers melting completely

[eclare108213]

The mushy-layer issue occurred in CESM in regions of very thick and rapidly melting ice in the Canadian Arctic Archipelago, and is likely symptomatic rather than the originating problem. We have two ways of addressing the issue, both of which are research projects. First, the model physics can be altered to stop such unphysically thick ice from appearing, which may involve the new tri-variate state space now being introduced to the sea ice model (floe size, thickness, macro-porosity). The second approach is to run mushy-layer thermodynamics as a super-parameterization (e.g., with extremely high resolution) within the sea ice component. While these approaches are explored, NCAR has a work-around external to the sea ice component, which may be appropriate for NOAA too.

@dabail10 writes:
There are two aspects to our so-called "workaround". The first was to implement the same shortwave absorption adjustment similar to what is used for BL99. That is, when the ice gets very close to the melting point, the excess shortwave is moved to the surface shortwave absorption and removed from the internal term. This is true in snow and sea ice. This reduces the amount of internal melt and hence the possibility of having big freshwater fluxes to the ocean. The fundamental issue is the vertical discretization in the ice. Imagine you have 10m of ice, then with 7 layers, the internal layer thicknesses are close to 1.5m. So, if you had an internal layer completely melt away, this is a very large flux. Our problem was that we were getting runaway ice growth in some of these small inlets in the CAA. So, we had points with 80m thick ice and hence 10m layers! That leads to the second aspect of our "workaround" which was to stop the runaway growth. Here, this was a land model change. We had both liquid and solid runoff fluxes from land going into the ocean. The solid fluxes are melted by the ocean model and hence take heat from the ocean. Thus cooling it down. In some locations this would cool the ocean enough to form sea ice. In these little inlets in the CAA, there was no way to flush the cold water or sea ice out and hence the runaway ice growth. The fix was to change the solid flux to liquid on land first (using heat from the atmosphere) and hence the ocean would not lose heat because of the runoff. It's still not perfect as we get 30m ice in the pre-industrial, but this doesn't cause the model to abort from large melt fluxes.