apply_ufunc doesn't inherit encoding

[jpdehollain]

What happened?

Encoding in data arrays are lost after applying any ufunc. There should be an argument similar to keep_attrs in apply_ufunc for encoding.

What did you expect to happen?

The encoding property should be inherited (or resolved if more than one data arrays are involved) when applying a ufunc.

Minimal Complete Verifiable Example

import xarray as xr
import numpy as np

my_ufunc = lambda x: x + 1
xarr1 = xr.DataArray(np.array([1,2,3]))
xarr1.encoding = {'dummy': 'baz'}
xarr2 = xr.apply_ufunc(my_ufunc, xarr1)
print(xarr1.encoding, xarr2.encoding)

# Workaround:
xarr2.encoding = xarr1.encoding.copy()
print(xarr1.encoding, xarr2.encoding)

MVCE confirmation

• Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray.
• Complete example — the example is self-contained, including all data and the text of any traceback.
• Verifiable example — the example copy & pastes into an IPython prompt or Binder notebook, returning the result.
• New issue — a search of GitHub Issues suggests this is not a duplicate.
• Recent environment — the issue occurs with the latest version of xarray and its dependencies.

Relevant log output

Anything else we need to know?

No response

Environment

INSTALLED VERSIONS

commit: None
python: 3.13.3 | packaged by conda-forge | (main, Apr 14 2025, 20:31:24) [MSC v.1943 64 bit (AMD64)]
python-bits: 64
OS: Windows
OS-release: 11
machine: AMD64
processor: Intel64 Family 6 Model 170 Stepping 4, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: ('English_Australia', '1252')
libhdf5: 1.14.6
libnetcdf: 4.9.2

xarray: 0.1.dev5937+g070af11
pandas: 2.2.3
numpy: 2.2.5
scipy: 1.15.2
netCDF4: 1.7.2
pydap: None
h5netcdf: None
h5py: None
zarr: None
cftime: 1.6.4
nc_time_axis: None
iris: None
bottleneck: None
dask: None
distributed: None
matplotlib: None
cartopy: None
seaborn: None
numbagg: None
fsspec: None
cupy: None
pint: None
sparse: None
flox: None
numpy_groupies: None
setuptools: None
pip: 25.1.1
conda: None
pytest: None
mypy: None
IPython: 9.2.0
sphinx: None

[welcome]

Thanks for opening your first issue here at xarray! Be sure to follow the issue template!
If you have an idea for a solution, we would really welcome a Pull Request with proposed changes.
See the Contributing Guide for more.
It may take us a while to respond here, but we really value your contribution. Contributors like you help make xarray better.
Thank you!

[kmuehlbauer]

Thanks @jpdehollain for raising this. Please see #6323 for more details. Closing in the light of #5336 and #5082.

[jpdehollain]

Thanks @kmuehlbauer and apologies for not finding that issue before posting (I searched for inherit instead of propagate 😆)..

To add some context in case it aids the discussion, I have a dataset with data arrays that can contain either float or str (dtype=object) types. The arrays get updated one coordinate value at a time (which means that some data values get filled) and updates happen across sessions so I need to store the Dataset.

I use encoding on the str arrays only to set the fill value, because otherwise they get converted to an incompatible type when I load the Dataset from file.

Setting the encoding on the entire Dataset at the .to_netcdf feels inefficient because I only want it on the str type arrays, so in this particular case it would be inconvenient to drop the encoding property all together from the variables

[kmuehlbauer]

@jpdehollain Thanks for the additional context. Yes, this is a somewhat not satisfactory situation. I've issues like this in my workflows, too. Usually I'm wrapping apply_ufunc in another function which fixes/copies the encoding as needed. It might be a bit disturbing, but in the end you have full control.

[dcherian]

You could also just stick the information in attrs. I believe the encoding logic will look there too

[jpdehollain]

You could also just stick the information in attrs. I believe the encoding logic will look there too

Thanks for the suggestion @dcherian. I just tried that but it didn't work for me, e.g., if I create:

ds = xr.Dataset({'foostr': ('dim', np.array(['one',np.nan,np.nan], dtype=object))},
                 coords={'dim': [1,2,3]})
# ds['foostr'].encoding = {'_FillValue': 'nan'}
ds['foostr'].attrs = {'_FillValue': 'nan'}

and then save it and load it:

ds.to_netcdf('test.nc')
xr.load_dataset('test.nc')

The string array gets loaded with empty strings. If instead I swap for the commented line above, the dataset loads in the correct way.