Open navidcy opened 1 week ago
Probably, either Oceananigans needs to have more built-in non-trivial grids (thereby motivating the implementation of OrthogonalSphericalShellGrid in Oceananigans the first place). Or, all of the orthogonal grid functionality could be moved here. A third option is to incorporate this package into Oceananigans.
I don't see what's the difference of the third option from the first.
The first option would add other orthogonal spherical shell grid options --- for example rotated lat lon grids --- to Oceananigans. Then we wouldn't need to invoke this package to test OrthogonalSphericalShellGrid
because we could use the native options instead, and this package could stand alone as implementation of the more complicated versions.
I like the rotated lat-lon grid option in Oceananigans. It's simple enough but very useful for things like arctic simulations
we can also implement a test utility that builds an OrthogonalSphericalShellGrid
from a lat-lon grid
Yeah, it's not very well thought out I think. The problem I think is basically that Oceananigans doesn't have much built-in functionality for orthogonal grids. So it helps to use this package to build a tripolar grid and test some of the functionality.
Probably, either Oceananigans needs to have more built-in non-trivial grids (thereby motivating the implementation of
OrthogonalSphericalShellGrid
in Oceananigans the first place). Or, all of the orthogonal grid functionality could be moved here. A third option is to incorporate this package into Oceananigans.My hunch is that it is important to have a general orthogonal grid implementation in Oceananigans because otherwise the general implementation of operators etc is not well-motivated.
This problem has been present from the beginning with
OrthogonalSphericalShellGrid
as I recall, it wasn't even possible to use it in stand-alone simulations at first.Originally posted by @glwagner in https://github.com/CliMA/OrthogonalSphericalShellGrids.jl/issues/33#issuecomment-2365246606