How to support time-depedennt metric terms?

[willirath]

(This picks up christophrenkl@1c273fc)

Currently, Xorca assumes that all the metric fields (e{1,2,3}{t,u,v,w}) are time-independent. We should loosen this requirement, because with VBL (?), Nemo now has a variable vertical grid.

[christophrenkl]

I came across this issue yesterday and as you've noticed I introduced a hack that seems to be working for me. However, it probably is not a universal solution. Since I changed the dimensions of the e3{t, u, v, w} scale factors, the time-independent fields from the mesh_mask file will get inflated along the time axis and thus become unnecessarily big.

[NoeLahaye]

Hello @willirath (and thanks for this very useful library). As I was searching for this kind of information, I came across this issue. Is this still the case? I need to process some data with variable volume, so I wanted to know wether it xorca will handle this natively or if there is some work to be done. One hack I have in mind -- instead of reading the time-variable vertical grid -- is to use an approximate variable grid computation from the grid at rest + ssh anomaly. Any suggestion?