`DataArray.interp` only works for numeric dtypes

[w-k-jones]

What is your issue?

Dataset.interp was updated to allow non-numeric dtypes in #5008

However, DataArray.interp still has an outdated check for numeric dtypes, meaning that other dtypes (e.g. datetimes) result in a TypeError:

xarray/xarray/core/dataarray.py

Line 2357 in f1a2c08

if self.dtype.kind not in "uifc":

[kmuehlbauer]

@w-k-jones Thanks for raising this. Are you able to send in a PR with the fix?

[w-k-jones]

Hi @kmuehlbauer , yes, should be a quick fix, I'll just need to find a bit of time! Is it suitable just toe remove the check from DataArray.interp and rely on the error handling in Dataset.interp? Also are any additional tests required, or are the current tests suitable for checking this all works correctly? I might add a quick test that DataArray.interp works with datetime types so that it isn't inadvertently changed in future.

[w-k-jones]

I've had a bit more of a look into this, and found some issues with the way Dataset.interp handles non-numeric values (including silently dropping datetime types and some issues with strings #8456) that are better to avoid when specifically interpolating a single variable. For now I'd suggest that the check for numeric dtypes in DataArray.interp remains in place until these issues are fixed to avoid unexpected return values

[gcaria]

Sounds like the silently dropping of datetime types reported by @w-k-jones deserves an issue of its own, thoughts @kmuehlbauer ?

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

times = np.arange(5)
data_values = pd.to_datetime(["2025-01-01", "2025-01-03", "2025-01-02", "2025-01-05", "2025-01-04"])
data = xr.DataArray(data_values, coords=[times], dims=["x"])
data = data.to_dataset(name='foo')
print(data)
new_x = np.linspace(0, 4, 9)
data_interp = data.interp(x=new_x)
print(data_interp)
----------------------------------------------------------------------

<xarray.Dataset> Size: 80B
Dimensions:  (x: 5)
Coordinates:
  * x        (x) int64 40B 0 1 2 3 4
Data variables:
    foo      (x) datetime64[ns] 40B 2025-01-01 2025-01-03 ... 2025-01-04

<xarray.Dataset> Size: 72B
Dimensions:  (x: 9)
Coordinates:
  * x        (x) float64 72B 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Data variables:
    *empty*