Reference state in HS state

[smarras79]

Why is there this check? Isn't the reference state always supposed to be in HS state?

https://github.com/climate-machine/CLIMA/blob/a1bc4b2e459bdb69a2c228feeffae4d64ec25b30/src/Atmos/Model/AtmosModel.jl#L136

[simonbyrne]

It is possible to run it without a reference state. Should we require a reference state?

[fxgiraldo]

We need a reference state for the linear IMEX solvers. Once we move to nonlinear IMEX solvers, we shouldn't need them. I should point out that, currently, we only plan on 1D-IMEX nonlinear solvers. For 3D-IMEX we still intend to use linear solvers (nonlinear is possible but has not been competitive with the linear version).

[simonbyrne]

Yes, if you attempt to use IMEX without a reference it will throw an error.

[tapios]

It is possible to run it without a reference state. Should we require a reference state?

We should not always require a reference state, although in practice we may (almost) always use it, for numerical stability reasons alone (factoring it out of the momentum equation). But it's good to keep the option to run standard flow models without gravity and without hydrostatic balance.

I see no harm in explicitly enforcing hydrostaticity in l. 136, even if right now reference states are hydrostatic (and most of the time will be, except perhaps in some flows without gravity).

[tapios]

Is this issue still open or can it be closed?

[smarras79]

(closed accidentally)