VertCoord getting `NVertLayers` from wrong stream and not reading `RefLayerThickness`

[xylar]

I could be mistaken but in implementing:
E3SM-Project/polaris#544
and reviewing
#382
I got the impression that VertCoord is reading NVertLayers from the horizontal mesh rather than the expected InitialVertCoord stream.

And looking through the code for VertCoord, it also seems like RefLayerThickness isn't being read in for that stream, which is important because that is the field that would provide NVertLayers.

[xylar]

@brian-oneill, I'm assigning this to you but please reassign if that's wrong.

[xylar]

In general, it seems like maybe we're not fully wired up for the p* coordinate yet?

[brian-oneill]

As I mentioned in #382, I'll make a PR to set NVertLayers from the InitialVertCoord stream. And yes, p* is not implemented yet.

Looking at MPAS, seems like refBottomDepth is what's read in from stream and then refLayerThickness is computed from it. I could add this to VertCoord as well in the PR.

[xylar]

@brian-oneill, the MPAS-Ocean implementation isn't what is needed in Omega. We want to be reading in the full RefLayerThickness field from the InitialVertCoord stream because that provides the level of flexibility we need. We need to be able to support initial conditions that are either:

• constant z-tilde-levels except for partial cells at the bathymetry (implemented in Polaris)
• match MPAS-Ocean z-levels with partial cells, but converted to z-tilde (so we can match existing MPAS-Ocean initial conditions; to be implemented by @katsmith133 in a separate workflow)

[xylar]

@brian-oneill, I have a draft PR (E3SM-Project/polaris#573) that will write out RefLayerThickness and VertCoordMovementWeights, which are the expected p* variables according to https://docs.e3sm.org/Omega/Omega/design/VertCoord.html. Feel free to use it in testing your p* branch or I can do it. Either way is fine.

[brian-oneill]

@xylar PR #401 is up. It successfully passes the omega_pr suite with both the current polaris-dev as well as your draft PR (I renamed RefLayerThickness to RefPseudoThickness in my PR, so I made that change locally in your tasks/ocean/init.py). Currently in Omega, VertCoordMovementWeights isn't being read in, but is initialized via the MovementWeightType config option. If the plan is to read those in, I can add that field to the stream in the PR.