Open tomchor opened 1 year ago
We also can't get dissipation when the closure doesn't use an eddy diffusivity, right? So it's not possible to calculate diffusivity for any turbulence closure.
We also can't get dissipation when the closure doesn't use an eddy diffusivity, right? So it's not possible to calculate diffusivity for any turbulence closure.
True. Although we could consider throwing a useful error in those cases?
I think we should at least support the common cases (like (VerticalDiffusivity(), HorizontalDiffusivity())
) though, especially because it should be pretty straightforward.
Agree, it makes sense to try to support any AbstractScalarDiffusivity
since it's calculable and there's an interface. We can also support tuples of AbstractScalarDiffusivity
I think, via https://github.com/CliMA/Oceananigans.jl/blob/ce5ba71449cf16b8737dca10997a6119dee4c665/src/TurbulenceClosures/abstract_scalar_diffusivity_closure.jl#L61-L62
At the moment we can't get (pseudo)-dissipation when the model has a tuple closure.
I'm not sure if it's possible to get actual dissipation when the diffusivity isn't isotropic, but it should be straightforward to get pseudo-dissipation.