Open radolan (and other) files which contain crs information - interaction with wradlib package

[olbermann]

I want to open RADOLAN files (hdf5 from DWD, in ODIM_H5/V2_3 format) with the xarray engine from the wradlib package.

import xarray as xr
ds = xr.open_dataset(filename, engine="radolan")

Now I have to add the line

ds_with_crs = ds.rio.write_crs(ds.crs_wkt.attrs["crs_wkt"])

before I can

ds_with_crs.rio.reproject(...)

and it feels a bit awkward to me? The file contains crs info in a standardized format, so shouldn't it be ready for reprojection after opening?

[snowman2]

https://corteva.github.io/rioxarray/stable/getting_started/getting_started.html#xarray

it is recommended to use decode_coords=”all”. This will load the grid mapping variable into coordinates for compatibility with rioxarray.

[olbermann]

So you are saying that in the example on https://corteva.github.io/rioxarray/stable/examples/examples.html:
Converting NetCDF dataset array to GeoTiff one can omit the write_crsby adding the decode_coords="all" keyword argument?

In my case this does not help. I do not use the standard xarray engine or netcdf files, it is an xarray engine plugin provided by the wradlib package: https://github.com/wradlib/wradlib/blob/main/wradlib/io/backends.py

Would you agree with the description that this engine and the rioxarray package don't agree how the coordinate system should be stored in an xarray dataset?
Or should one say that rioxarray introduces its own separate accessor object which "no other package which does not depend on rioxarray can use" and stores the coordinate system info there?
Would it make sense / be possible for one of the two packages to change behaviour? (Without introducing a dependency of one of the two packages on the other, but rather "using a common standard"?)

[snowman2]

That's unfortunate that decode_coords="all" doesn't work for that engine. I see the decode_coords parameter in the open function. Maybe it just needs to be enabled.

Related: #287