Compute salinity for the bottom boundary conditions at the top of the grid

[simone-silvestri]

Supposedly, this is the salinity of the surface of the ocean. Given the convention that a slab sea ice is at the top of the grid, we need to compute the salinity at size(grid, 3) instead of 1

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 70.44%. Comparing base (a922f71) to head (fdec6c9).
⚠️ Report is 47 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #73   +/-   ##
=======================================
  Coverage   70.44%   70.44%           
=======================================
  Files          27       27           
  Lines         829      829           
=======================================
  Hits          584      584           
  Misses        245      245           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[glwagner]

so grid is the ocean grid or the sea ice grid?

[glwagner]

confusing to find this inside ClimaSeaIce

[simone-silvestri]

I guess this is ocean salinity, so logically this should be the ocean's grid. However, in this case, we are passing to the bottom_temperature function the sea ice grid. Luckily, there is no ambiguity in case of a slab thermodynamic model as we have decided in #68 to have the momentum equations (and by extension any 2D field which in principle has no k-location) live at the top of the grid to comply with bathymetry and a possible coupled ocean. So if this function is used in the slab model (2D) it should be kᴺ = size(grid, 3) and not k = 1.
We still have to think about a strategy for a possible multi-layer model (is the k-index the index of the layers? Do we add another dimension?).

[glwagner]

I feel we likely want a 3D grid for multi-layer sea ice, because we need that for 3D fields and 3D kernels (though many of the kernels will be 2D)

[simone-silvestri]

Ok, I guess we can delay this until we decide the strategy for multi-layer sea ice. In the meantime, in ClimaOcean we can pass the ocean salinity to the boundary condition as

interior(S, :, :, grid.Nz : grid.Nz)

instead of

view(S, :, :, grid.Nz)

to make sure the indexing is correct.