What happened: When calling DataArray.to_dataset(dim=...) for splitting the array along one dimension, its attributes are copied to the resulting Dataset but not to the variables.

What you expected to happen: Attributes are copied to the variables.

Minimal Complete Verifiable Example:

```python import numpy as np import xarray as xr

this setting does not change the behaviour

xr.set_options(keep_attrs=True)

data = np.arange(234, dtype="uint16").reshape((2,3,4)) coords = [("band", ["1", "2"]), ("y", np.arange(3)), ("x", np.arange(4))] da = xr.DataArray(data, coords=coords, attrs=dict(myattr=9999)) da ```

```python

the attribute is always copied to the Dataset regardless of the promote_attrs value

ds_split = da.to_dataset("band", promote_attrs=False) ds_split ```

Anything else we need to know?: If I avoid the splitting, the attribute is only in the Variable and not in the Dataset. python ds_3d = da.to_dataset(name="ds_3d") ds_3d

This reproduces the expected behaviour: python ds_sel = xr.Dataset({band: da.sel(band=band, drop=True) for band in da.band.values}) ds_sel

Environment:

Output of <tt>xr.show_versions()</tt>

INSTALLED VERSIONS ------------------ commit: None python: 3.8.12 | packaged by conda-forge | (default, Oct 12 2021, 21:59:51) [GCC 9.4.0] python-bits: 64 OS: Linux OS-release: 4.15.0-111-generic machine: x86_64 processor: x86_64 byteorder: little LC_ALL: None LANG: en_US.UTF-8 LOCALE: ('en_US', 'UTF-8') libhdf5: None libnetcdf: None xarray: 0.20.1 pandas: 1.3.4 numpy: 1.21.4 scipy: 1.7.3 netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None nc_time_axis: None PseudoNetCDF: None rasterio: 1.2.10 cfgrib: 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 setuptools: 59.4.0 pip: 21.3.1 conda: None pytest: None IPython: 7.30.0 sphinx: None