unable to unstack in cfradial1 io due to duplicate value in elevation dim

[chiaweh2]

Hi All,

I am currently testing opening a SPol Radar data from https://data.eol.ucar.edu/dataset/342655. The file can be open fine when using the xarray cfradial1 engine or PyART but encounter the following error when doing xd.io.open_cfradial1_datatree(file)

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/test_xradar/.venv/lib/python3.11/site-packages/xarray/core/dataset.py", line 5606, in unstack
    result = result._unstack_once(d, stacked_indexes[d], fill_value, sparse)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/test_xradar/.venv/lib/python3.11/site-packages/xarray/core/dataset.py", line 5426, in _unstack_once
    new_indexes, clean_index = index.unstack()
                               ^^^^^^^^^^^^^^^
  File "/test_xradar/.venv/lib/python3.11/site-packages/xarray/core/indexes.py", line 1131, in unstack
    raise ValueError(
ValueError: Cannot unstack MultiIndex containing duplicates. Make sure entries are unique, e.g., by  calling ``.drop_duplicates('n_points')``, before unstacking.

The traceback error point to the following line of unstacking

xradar/xradar/io/backends/cfradial1.py

Lines 177 to 183 in 727434c

	rslice = slice(0, current_ray_n_gates[0].values.astype(int))
	ds = ds.isel(range=rslice)
	ds = ds.isel(n_points=nslice)
	ds_vars = ds[data_vars]
	ds_vars = merge([ds_vars, ds[[dim0, "range"]]], compat="no_conflicts")
	ds_vars = ds_vars.stack(n_points=[dim0, "range"])
	ds_vars = ds_vars.unstack("n_points")

And like the error mentioned the stacked dim elevation in the previous line does have duplicate values shown below

which resulted in the error of unable to unstack. Is the dimension here rarely or should not have duplicate values? Thank you all!

[kmuehlbauer]

@chiaweh2 The link does not resolve to a particular dataset. Would you mind having a look?

[chiaweh2]

@kmuehlbauer I am so sorry. I miss a 5 at the end. I fixed the link above. and just for convinent. The doi to the data is this https://doi.org/10.26023/QA4M-CDHH-MY0Z

[chiaweh2]

Just curious how pyART unpack/reshape the data and noticed this.
https://github.com/ARM-DOE/pyart/blob/15b6dcb5851ab4892ca92d0537d2b25969d73ab9/pyart/io/cfradial.py#L445-L452
This seems to side step and make sure even for repeating "coordinate value" it will still be able to reshape the data!

[chiaweh2]

Since different time may have different level, the time I got the error is this one cfrad.20220525_015443.665_to_20220525_015709.985_SPOL_PrecipRhi2_RHI.nc Sorry i cannot provide a link to this exact data. It is over the file size limit github set up.