On vertical vs Z
Closed this issue · 5 comments
Also I realize that the vertical Axes and Coords are not what I was expecting but my eyes glossed over them earlier.
Probably s_rho should be the Axes Z (and I am setting this in xroms for this to appear here). But then the associated coord should be z_rho, in analogy to the relationship between xi_rho and lon_rho. Is there an adjustment I can make so that z_rho will become linked as a coord to s_rho in DataArrays?
Originally posted by @kthyng in #82 (comment)
I would add z_rho to the coordinates attribute of ds.temp and others to make the link explicit.
But the fundamental issue here is that Z and vertical are synonyms. One way to solve this would be to
- Have
verticalnot matchaxis: Z. - Delete the
positive: upattribute froms_rho(i don't like this but don't see an alternative) - then
ds.cf["Z"]will give yous_rhoandds.cf["vertical"]will give youz_rho.
cc @jthielen; for this model s_rho is a non-dimensional 1D vertical coordinate and z_rho is a 4D parameterized vertical coordinate; and this issue is about making it easy for users to pull out one or the other based on attributes using either Z or vertical. This is messy because positive is set on s_rho (which makes sense but is inconvenient).
What are your thoughts making Z and vertical not synonyms (similar to X vs longitude)?
Zwould matchaxisand maybeformula_termsand/or appropriatestandard_namefor parameterized vertical coordinates (looks like the standard names for meteorological variables are already in the criteria).verticalwould matchpositiveandstandard_names likeheight,depthetc.
From a quick look at http://cfconventions.org/Data/cf-conventions/cf-conventions-1.8/cf-conventions.html#vertical-coordinate I don't see how else to distinguish between Z and vertical (it isn't obvious that the conventions distinguish between the two)
Maybe @rsignell-usgs also has thoughts here?
Thank you, you explained my issue clearly. In the meantime I'll try your suggestion to add coordinate attributes, I haven't used those.
EDIT: I added this ds['temp'].attrs['coordinates'] = 'lon_rho lat_rho z_rho ocean_time' and ds.temp.cf.describe() still does not recognize z_rho as coordinates.
ah yes. Now that it is in coordinates, cf_xarray will look at the z_rho variable but not find any attrs matching the criteria for vertical. I would set positive: up on z_rho.
cc @jthielen; for this model
s_rhois a non-dimensional 1D vertical coordinate andz_rhois a 4D parameterized vertical coordinate; and this issue is about making it easy for users to pull out one or the other based on attributes using eitherZorvertical. This is messy becausepositiveis set ons_rho(which makes sense but is inconvenient)....
What are your thoughts making
Zandverticalnot synonyms (similar toXvslongitude)?* `Z` would match `axis` and maybe `formula_terms` and/or appropriate `standard_name` for parameterized vertical coordinates (looks like the standard names for meteorological variables are already in the criteria). * `vertical` would match `positive` and `standard_name`s like `height`, `depth` etc.
Sounds like a good idea! Both s_rho and z_rho are vertical coordinates, with s_rho being the axis-like coordinate and z_rho being the full/physical coordinate, which, as you said, is analogous to a grid_x versus longitude situation (where grid_x is the axis-like coordinate and longitude is the full/physical coordinate). So, having the distinction between Z and vertical makes sense.
With that in mind, would this be a correct/useful perspective, perhaps worth adding to the docs?
Z,Y, andXidentify axis-like coordinates (typically 1D dimension coordinates)vertical,latitude, andlongitudeidentify physical (fully-expanded) coordinates (typically non-dimension/auxiliary coordinates, unless on a particular grid type where they are 1D/dimension coordinates)