new vertical coord decode and reordering

[kthyng]

The vertical decoding has been added for ocean_sigma_coordinate and the free surface variable has been expanded dimensionally for all the calculations so that the results preserve the standard ordering of time, vertical, lat/Y, lon/X. Three of the tests do not work anymore, though. Expanding dimensionally apparently bumps the attributes off of s_rho, for one. For another, the values are very similar but do not match anymore — not sure why. Interested to see what @dcherian thinks!

[kthyng]

Ah I see an immediate problem with my zname_in selection. It works for POM where there is one vertical coordinate but for FVCOM which also uses ocean_sigma_coordinate, there are two, and neither has _ in them. So, zname_in should be a list I guess? or just a prefix on the full s coordinate name?

[dcherian]

What are the variable names for FVCOM? Can you also add an FVCOM test dataset please?

I think the clearest solution might be a new outnames kwarg or maybe decode_to

ds.cf.decode_vertical_coords(outnames={"s_rho": "z_rho", "s_w": "z_w"})

which means that we should deprecate prefix by first changing the default to None, then

if prefix is not None:
    warnings.warn(..., DeprecationWarning)
    # set outnames here

[kthyng]

What are the variable names for FVCOM? Can you also add an FVCOM test dataset please?

FVCOM won't work currently with xarray because the vertical coordinates have the same names as the dimensions and can't be read in without dropping the coordinate information, which is what has the vertical coordinate info. But I am trying to set up to be ready to do it when it works!

I think the clearest solution might be a new outnames kwarg or maybe decode_to

ds.cf.decode_vertical_coords(outnames={"s_rho": "z_rho", "s_w": "z_w"})

which means that we should deprecate prefix by first changing the default to None, then

if prefix is not None:
    warnings.warn(..., DeprecationWarning)
    # set outnames here

I like this approach.

[kthyng]

@dcherian Sorry we overlapped a bit just now. Could you take a look now? I think I have it correct but not positive.

[dcherian]

Looks good to me now. Just 2 v. minor comments. Thanks!

The test failures are unrelated.

[kthyng]

@dcherian Anything else I should do here?

[dcherian]

There's another test that needs to be fixed by providing outnames

[kthyng]

@dcherian Ah ok how about now?

[dcherian]

Thanks!

[kthyng]

Wahoo!!! Thanks for all your help!