Add ocean horizontal viscosity to 1 deg ocean simulation example

[glwagner]

@jlk9 @heimbach

[simone-silvestri]

Looks good! I have found a bit more success using the biharmonic viscosity, with something like $\nu = \frac{Az^2}{15days}$

something like

@inline νhb(i, j, k, grid, ℓx, ℓy, ℓz, clock, fields, λ) = Oceananigans.Operators.Az(i, j, k, grid, ℓx, ℓy, ℓz)^2 / λ
horizontal_viscosity = HorizontalScalarBiharmonicDiffusivity(ν=νhb, discrete_form=true, parameters=15days)

[glwagner]

Looks good! I have found a bit more success using the biharmonic viscosity, with something like ν = A z 2 15 d a y s

something like

@inline νhb(i, j, k, grid, ℓx, ℓy, ℓz, clock, fields, λ) = Oceananigans.Operators.Az(i, j, k, grid, ℓx, ℓy, ℓz)^2 / λ
horizontal_viscosity = HorizontalScalarBiharmonicDiffusivity(ν=νhb, discrete_form=true, parameters=15days)

Ok let's try it. We may want to implement this function in Oceananigans.

[navidcy]

https://clima.github.io/ClimaOceanDocumentation/previews/PR697/literated/one_degree_simulation/

Simulation results look OK.
But what are all those JLD2 warnings after run!(simulation)?

[glwagner]

Looks like it has something to do with using a function in ScalarBiharmonicDiffusivity. I think the results like much better. The grid imprinting seems to be gone.

[navidcy]

I've never seen that warning before. Did we change something recently related to ScalarBiharmonicDiffusivity?

[glwagner]

I've never seen that warning before. Did we change something recently related to ScalarBiharmonicDiffusivity?

It's the kind of message that occurs when we try to load something after it has been serialized with JLD2. I don't know why JLD2 is trying to serialize model.closure.

[navidcy]

It's because the JLD2Writer is trying to save the function

ClimaOcean.jl/examples/one_degree_simulation.jl

Line 57 in bfae2f7

@inline νhb(i, j, k, grid, ℓx, ℓy, ℓz, clock, fields, λ) = Oceananigans.Operators.Az(i, j, k, grid, ℓx, ℓy, ℓz)^2 / λ

[navidcy]

Happy to merge this. But do we want it with all those JLD2 warnings?

[navidcy]

I don't think we can avoid them...

[glwagner]

I think we can write including = [] in the constructor for JLD2Writer:

https://github.com/CliMA/Oceananigans.jl/blob/58c6b7a7b4760abc4da76ddef1ae8d198ea82b6b/src/OutputWriters/jld2_writer.jl#L35

the stuff saved is pretty arbitrary though. We should probably change that to just save the grid.

[glwagner]

@simone-silvestri presumably you must hve seen this?