dataarray arithmetics restore removed coordinates in xarray 0.15

[mraspaud]

MCVE Code Sample

import xarray as xr
import numpy as np
arr2 = xr.DataArray(np.ones((2, 2)), dims=['y', 'x'])
arr1 = xr.DataArray(np.ones((2, 2)), dims=['y', 'x'], coords={'y': [0, 1], 'x': [0, 1]})

del arr1.coords['y']
del arr1.coords['x']

# shows arr1 without coordinates
arr1

# shows coordinates in xarray 0.15
arr1 * arr2

Expected Output

<xarray.DataArray (y: 2, x: 2)>
array([[1., 1.],
       [1., 1.]])
Dimensions without coordinates: y, x

Problem Description

In xarray 0.15, the coordinates are restored when doing the multiplication:

<xarray.DataArray (y: 2, x: 2)>
array([[1., 1.],
       [1., 1.]])
Coordinates:
  * y        (y) int64 0 1
  * x        (x) int64 0 1

Output of xr.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.8.1 | packaged by conda-forge | (default, Jan 29 2020, 14:55:04)
[GCC 7.3.0]
python-bits: 64
OS: Linux
OS-release: 4.18.0-147.0.3.el8_1.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_GB.UTF-8
LOCALE: en_GB.UTF-8
libhdf5: 1.10.5
libnetcdf: 4.7.3

xarray: 0.15.0
pandas: 1.0.0
numpy: 1.18.1
scipy: 1.4.1
netCDF4: 1.5.3
pydap: None
h5netcdf: 0.7.4
h5py: 2.10.0
Nio: None
zarr: 2.3.2
cftime: 1.0.4.2
nc_time_axis: None
PseudoNetCDF: None
rasterio: 1.1.2
cfgrib: None
iris: None
bottleneck: 1.3.1
dask: 2.10.1
distributed: 2.10.0
matplotlib: 3.1.3
cartopy: 0.17.0
seaborn: None
numbagg: None
setuptools: 45.1.0.post20200119
pip: 20.0.2
conda: None
pytest: 5.3.5
IPython: 7.12.0
sphinx: 2.3.1

[keewis]

the issue here is that if you call __delitem__ on DataArray.coords (which returns a dict-like object named DataArrayCoordinates), it deletes the coordinates, but leaves the indexes intact:

In [13]: arr1_ = arr1.copy()
    ...: del arr1_.coords["x"]
    ...: del arr1_.coords["y"]
    ...: arr1_.coords, arr1_.indexes
Out[13]:
(Coordinates:
     *empty*, y: Int64Index([0, 1], dtype='int64', name='y')
 x: Int64Index([0, 1], dtype='int64', name='x'))

Since #3481 the indexes are passed along in binary operations (such as multiplication) and will thus cause the coordinates reappear. Note that coordinates without indexes (non-dimension coordinates) will be removed properly.

If we want to support this way of dropping coordinates we should update the indexes, explicitly or via drop_vars, otherwise we might want to make *Coordinates objects read-only.

For now the easiest way to properly remove coordinates is by using drop_vars.

[mraspaud]

@keewis thanks for the quick reply. I wasn't aware the builtin del wasn't supposed to work here, I'll try with drop_vars instead.

[keewis]

I don't know whether or not this was supposed to work. Since there is a explicit __delitem__, I'd say it was at some point and my question above is more along the lines of "should this still work?".

[mraspaud]

Thanks for the clarification. I can confirm that drop_vars works as expected. As a user, I would vote for having __delitem__ call drop_vars, at least for now to keep backwards compatibility ?

[dcherian]

IMO we should fix to remove associated indexes and issue a patch release.