Conversation
…to setup model itself
…igh order polynomials
| Nx = 400 | ||
| Ny = 450 | ||
| Nz = 70 # TODO: modify spacing if needed, 10 m at surface, 100m at seafloor |
There was a problem hiding this comment.
Can we start with a smaller resolution? It'd be nice to test this on our laptops first
|
|
||
| arch = GPU() | ||
|
|
||
| g_Earth = 9.80665 |
There was a problem hiding this comment.
It might be simpler to keep the defaults rather than change this. I made a comment though if we want to manually set this
| north_mask = GaussianMask{:y}(center=Ly, width=sponge_width) | ||
| south_sponge_layer = Relaxation(; rate=damping_rate, mask=south_mask) | ||
| north_sponge_layer = Relaxation(; rate=damping_rate, mask=north_mask) | ||
| sponge_layers = (south_sponge_layer, north_sponge_layer) |
There was a problem hiding this comment.
Let's try running without the sponge layer to start
| velocities = nothing, | ||
| pressure = nothing, | ||
| diffusivity_fields = nothing, | ||
| #auxiliary_fields = nothing, # NamedTuple(), |
| velocities = nothing, | ||
| pressure = nothing, | ||
| diffusivity_fields = nothing, | ||
| #auxiliary_fields = nothing, # NamedTuple(), |
There was a problem hiding this comment.
| velocities = nothing, | |
| pressure = nothing, | |
| diffusivity_fields = nothing, | |
| #auxiliary_fields = nothing, # NamedTuple(), |
| free_surface = ImplicitFreeSurface(gravitational_acceleration=g_Earth), | ||
| forcing = (u=sponge_layers, v=sponge_layers, w=sponge_layers, T=sponge_layers, S=sponge_layers), # NamedTuple() | ||
| closure = (horizontal_closure, vertical_closure), | ||
| boundary_conditions = (u=free_slip_surface_bcs, v=free_slip_surface_bcs, w=no_slip_field_bcs), # NamedTuple(), |
There was a problem hiding this comment.
Note we can't set boundary conditions on w in a hydrostatic model.
| coriolis = coriolis, | ||
| free_surface = ImplicitFreeSurface(gravitational_acceleration=g_Earth), | ||
| forcing = (u=sponge_layers, v=sponge_layers, w=sponge_layers, T=sponge_layers, S=sponge_layers), # NamedTuple() | ||
| closure = (horizontal_closure, vertical_closure), |
There was a problem hiding this comment.
We can rely entirely on WENO for horizontal dissipation. The only closure we need is for vertical mixing.
| # Diffusivities as part of closure | ||
| # TODO: make sure this works for biharmonic diffusivities as the horizontal, | ||
| horizontal_closure = HorizontalScalarDiffusivity(ν=0.1, κ=0.1) | ||
| vertical_closure = VerticalScalarDiffusivity(ν=3e-4, κ=1e-5) |
There was a problem hiding this comment.
We don't need biharmonic or a background diffusivity if we use WENO + CATKE, so we can eliminate these
| eos = TEOS10EquationOfState() | ||
|
|
||
| # Coriolis Effect, using basic f-plane with precribed reference Coriolis parameter | ||
| coriolis = FPlane(f=-1.3e14) |
There was a problem hiding this comment.
Perhaps insert the reference latitude here? Eg something like
coriolis = FPlane(latitude=-60)
| no_slip_bc = ValueBoundaryCondition(0.0) | ||
| free_slip_bc = FluxBoundaryCondition(nothing) | ||
|
|
||
| free_slip_surface_bcs = FieldBoundaryConditions(no_slip_bc, top=FluxBoundaryCondition(nothing)) | ||
| no_slip_field_bcs = FieldBoundaryConditions(no_slip_bc) |
There was a problem hiding this comment.
I don't think we want to impose no slip conditions here (even if it was used in the original Si and Stewart). We just need the momentum fluxes for u and v at the surface to start. We can also add bottom drag for u and v but this is a detail we can get to later
| filename = filename * ".jld2", | ||
| indices = (:, grid.Ny/2, :), | ||
| schedule = TimeInterval(1minute), | ||
| overwrite_existing = true) |
There was a problem hiding this comment.
The most interesting output will be a slice at the surface, eg using
indices = (:, :, grid.Nz)so make sure to always output that
|
Remember to prepend branch names with initials in the future, eg |
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## main #8 +/- ##
==========================================
- Coverage 22.80% 22.72% -0.08%
==========================================
Files 11 11
Lines 307 308 +1
==========================================
Hits 70 70
- Misses 237 238 +1
☔ View full report in Codecov by Sentry. |
| ice_consolidation_thickness :: MIT | ||
| # Numerics | ||
| advection :: A | ||
| architecture |
There was a problem hiding this comment.
| architecture | |
| architecture :: Arch |
This needs a type parameter to be concrete
but you should be able to get the architecture from the grid?
|
We can maybe revamp this experiment and run it as an example nightly? |
3 participants
This PR creates an idealized test case to help address issue #1. In particular, it's the beginning of setting up an ocean model based on the Antarctic Slope Current example presented here: http://eisenman.ucsd.edu/papers/Si-Stewart-Eisenman-2022.pdf.
TODO: