cmorization of REMO output

[larsbuntemeyer]

Summary

For the new netcdf remo output, we should simplify the cmorization process, using typcically xr.resample functions. Right now, there are no real CORDEX cmor tables in any standard format. We can try to reformat the output as close as possible to cmor standards and incorporate the python API at another time.

To Do

• Create a mapping table that maps remo codes to cf standard names
• Create CORDEX mip tables (this should be a community effort)
• Add _bounds variable in space and time (if not available)
• Check for Cordex cmor tables in json format (not available yet, see Added 1hr frequency for variable pr as requested from CORE PCMDI/cordex-cmor-tables#1)

Maybe we can also use cf-xarray that should handle all cf related issues.

see also: euro-cordex/py-cordex#26

[larsbuntemeyer]

@ludwiglierhammer

Maybe this could come in handy for handling bounds: https://cf-xarray.readthedocs.io/en/latest/generated/xarray.Dataset.cf.add_bounds.html

[larsbuntemeyer]

function that can be used to determine frequency of a time-axis: xr.infer_freq
works mostly for subdaily and daily frequencies, for monthly, it has to be first of month.

[larsbuntemeyer]

example for grid mapping: PCMDI/cmor#82

[ludwiglierhammer]

in remo_cmor.py function _define_axes the time axis name is 'time'. This has to be more flexible because in the test CORDEX tables (CORDEX_coordinate.json) there are two time axis defined: time and time1. Time1 is used e.g. for 3hourly data (CORDEX_3hr.json).
The differnec between time and time1 is:
time -> "must_have_bounds": "yes",
time1 -> "must_have_bounds": "no"

[larsbuntemeyer]

This effort has been moved to: euro-cordex/py-cordex#26