interpolate_na doesn't support extrapolation

[zxdawn]

What happened:
interpolate_na doesn't support extrapolation

What you expected to happen:
Support extrapolation.

Minimal Complete Verifiable Example:

import xarray as xr
import numpy as np

x = xr.DataArray(
    [[0, 1, np.nan, np.nan, 2, np.nan, np.nan]],
    dims=['y', 'x'],
    coords={"x": xr.Variable("x", [0, 1, 1.1, 1.8, 2, 4, 5]),
            'y': xr.Variable("y", [0])},
)

x = x.interpolate_na(dim="x", method="linear", use_coordinate="x")
print(x)

The output is:

array([[0. , 1. , 1.1, 1.8, 2. , nan, nan]])

It should be this array after extrapolation:

array([[0. , 1. , 1.1, 1.8, 2. , 4, 5]])

Environment:

Output of xr.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.8.5 | packaged by conda-forge | (default, Jul 24 2020, 01:25:15)
[GCC 7.5.0]
python-bits: 64
OS: Linux
OS-release: 5.4.0-42-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8
libhdf5: 1.10.6
libnetcdf: 4.7.4

xarray: 0.16.0
pandas: 1.0.5
numpy: 1.19.1
scipy: 1.5.2
netCDF4: 1.5.4
pydap: None
h5netcdf: None
h5py: 2.10.0
Nio: None
zarr: 2.4.0
cftime: 1.2.1
nc_time_axis: None
PseudoNetCDF: None
rasterio: 1.1.5
cfgrib: None
iris: None
bottleneck: None
dask: 2.21.0
distributed: 2.21.0
matplotlib: 3.3.0
cartopy: None
seaborn: None
numbagg: None
pint: None
setuptools: 49.2.0.post20200712
pip: 20.1.1
conda: None
pytest: None
IPython: None
sphinx: None

[dcherian]

Pass fill_value="extrapolate" and it will do so. We should add this to the docstring.

[zxdawn]

Thanks, it works well.

[keewis]

how would we document that? fill_value="extrapolate" is passed to scipy.interpolate.interp1d which is covered by **kwargs. I'd add examples, but I'm not sure if that's visible enough.

[dcherian]

Either Examples or Notes? It's such a common question, it'd be nice to have it in the docstring somewhere.