What happened:

Indexing on DataArrays sometimes produce inconsistent behaviors on DataArrays indexed with a DatetimeIndex. There are cases (see code below) where copied versions of the array obtained with DataArray.copy() loose the index's freq attribute when calling isel(), whereas copied versions obtained using DataArray.sortby() do not.

What you expected to happen:

I would expect the freq not to be dropped from the axis when indexing the DataArray. Or at least, if it is dropped, it should always be dropped, and not depend on how the DataArray has been obtained (i.e., it should not depend on whether it's a copy() of another array).

Minimal Complete Verifiable Example:

```python import xarray as xr import pandas as pd import numpy as np from pandas.tseries.frequencies import to_offset

freq = to_offset('MS') time_index = pd.DatetimeIndex([pd.Timestamp('20130101') + i * freq for i in range(10)])

xa_orig = xr.DataArray(np.random.randn(10), dims=('time'), coords={'time': time_index})

xa_copy = xa_orig.copy()

xa_sorted = xa_orig.sortby('time')

Set freq on all DataArrays:

xa_orig.get_index('time').freq = freq xa_copy.get_index('time').freq = freq xa_sorted.get_index('time').freq = freq

Print to confirm the freq is set correctly:

print(xa_orig.get_index('time').freq) print(xa_copy.get_index('time').freq) print(xa_sorted.get_index('time').freq) ```

Output (OK - as expected) ```

<MonthBegin> <MonthBegin> <MonthBegin> ```

Now, try some indexing using isel() on slices: python print(xa_orig.isel({'time': slice(0, 5, None)}).get_index('time').freq) print(xa_copy.isel({'time': slice(0, 5, None)}).get_index('time').freq) # freq is dropped! print(xa_sorted.isel({'time': slice(0, 5, None)}).get_index('time').freq)

Output (Not as expected: freq is dropped inconsistently, only on the copied DataArray) ```

<MonthBegin> None <MonthBegin> ```

Now, try the same using isel() on lists: python print(xa_orig.isel({'time': [0, 1, 2, 3, 4]}).get_index('time').freq) print(xa_copy.isel({'time': [0, 1, 2, 3, 4]}).get_index('time').freq) print(xa_sorted.isel({'time': [0, 1, 2, 3, 4]}).get_index('time').freq)

Output (Not as expected: freq always dropped) ```

None None None ```

Anything else we need to know?:

Environment:

What happened?

First, apology if this is not actually a bug - I'm not too sure of what the intended behaviour should be. But I find this counter-intuitive.

When indexing a DataArray that is indexed using a RangeIndex, the resulting index is an Int64Index: ```python my_da.get_index('time')

RangeIndex(start=0, stop=100, step=1, name='time')

a = my_da.sel({'time': pd.RangeIndex(0,2)}) a.get_index('time')

Int64Index([0, 1], dtype='int64', name='time') ```

Setting the index to the desired RangeIndex using assign_coords() then works. But I find it a bit problematic that sel() returns an Int64Index even when used with a RangeIndex. Also because Int64Index has been recently deprecated in Pandas 1.4.

What did you expect to happen?

I would have expected the resulting DataArray to be indexed with the same RangeIndex used in sel().

Minimal Complete Verifiable Example

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

my_da = xr.DataArray(np.random.rand(100,), dims=('time'), coords={'time': pd.RangeIndex(0, 100)})

print(my_da.get_index('time')) a = my_da.sel({'time': pd.RangeIndex(0,2)}) print(a.get_index('time')) ```

Relevant log output

python RangeIndex(start=0, stop=100, step=1, name='time') Int64Index([0, 1], dtype='int64', name='time')

Anything else we need to know?

No response

Environment

INSTALLED VERSIONS

commit: None python: 3.8.5 (default, Sep 4 2020, 02:22:02) [Clang 10.0.0 ] python-bits: 64 OS: Darwin OS-release: 20.6.0 machine: x86_64 processor: i386 byteorder: little LC_ALL: None LANG: None LOCALE: (None, 'UTF-8') libhdf5: None libnetcdf: None

xarray: 0.20.2 pandas: 1.4.0 numpy: 1.22.1 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: None cfgrib: None iris: None bottleneck: None dask: None distributed: None matplotlib: 3.5.1 cartopy: None seaborn: None numbagg: None fsspec: 2021.11.1 cupy: None pint: None sparse: None setuptools: 59.5.0 pip: 21.3.1 conda: None pytest: 6.2.5 IPython: 8.0.1 sphinx: 4.3.2

What happened:

Calling sortby() on a DataArray indexed with a pd.DatetimeIndex discards the underlying freq component of the pd.DatetimeIndex.

What you expected to happen:

The freq component should probably be kept.

Minimal Complete Verifiable Example:

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

ar = xr.DataArray(np.random.randn(4, 2), dims=('time', 'component'), coords={'time': pd.date_range('20130101', '20130104'), 'component': ['a', 'b']})

print(ar.get_index('time'))

ar_sorted = ar.sortby('time')

print(ar_sorted.get_index('time')) ```

Output for the above snippet (note how the freq becomes None): DatetimeIndex(['2013-01-01', '2013-01-02', '2013-01-03', '2013-01-04'], dtype='datetime64[ns]', name='time', freq='D') DatetimeIndex(['2013-01-01', '2013-01-02', '2013-01-03', '2013-01-04'], dtype='datetime64[ns]', name='time', freq=None)

Anything else we need to know?:

Environment: