FZJ-CMIP-ozone-1-2 has inconsistent long_name for pressure coord

[penguian]

Dataset source ID(s)

FZJ-CMIP-ozone-1-2

Describe the issue

Inconsistent long_name for coordinate with var_name=plev.

Expected data

long_name should either consistently be pressure or air_pressure.

Screenshots

When the CMIP7 NetCDF files for FZJ-CMIP-ozone-1-2 are loaded into a Jupyter notebook using Iris, and the following cell code is run:

for cube in [ozone_hi_cube, zmta_hi_cube, ozone_pi_cube, zmta_pi_cube]:
    print(cube.metadata.attributes["source_id"])
    print(cube.metadata.long_name)
    print(cube.metadata.attributes["comment"])
    print(cube.coord(var_name="plev").metadata)
    print()

the result is

FZJ-CMIP-ozone-1-2
vmro3
CMIP7 historical ozone forcing with nudged QBO signal, historical solar forcing, and historical emissions
DimCoordMetadata(long_name=pressure, var_name=plev, units=hPa, climatological=False, circular=False)

FZJ-CMIP-ozone-1-2
zmta
CMIP7 pi-control zonal mean temperature historical simulation
DimCoordMetadata(long_name=pressure, var_name=plev, units=hPa, climatological=False, circular=False)

FZJ-CMIP-ozone-1-2
vmro3
CMIP7 pi-control ozone forcing climatology averaged over 21-year simulation with QBO signal (182901-184912) and repeating 1850 emissions
DimCoordMetadata(long_name=air_pressure, var_name=plev, units=hPa, climatological=False, circular=False)

FZJ-CMIP-ozone-1-2
zmta
CMIP7 pi-control onal mean temperature climatology averaged over 21-year simulation with QBO signal (182901-184912) and repeating 1850 emissions
DimCoordMetadata(long_name=air_pressure, var_name=plev, units=hPa, climatological=False, circular=False)

which shows that var_name=plev corresponds to two different long_name values, pressure for historical data, and air_pressure for climatological data.

Additional context

This complicates input processing, since the same code can only be used for mon (historical) and monC (climatology) data if the var_name is used for the coordinate, rather than the long_name.

[znichollscr]

Thanks for raising @penguian. This is unfortunate. However, I am tempted to label this with wont-fix and close. The long_name is, in my experience, not controlled and therefore a free for all. While consistency would be nice, the thing that gets checked carefully is var_name I believe, so that is a much safer thing to rely on. What do you think @durack1 ?

[znichollscr]

One other note: if a consistent long-name is critical for your processing, you can always just update the long name after loading the data in. You'll have one (maybe two?) lines that differ for processing mon vs. monC, but otherwise the code can stay the same.