Temporal polyfit does not use the same coordinate as polyval

[aulemahal]

What happened?

When fitting a variable along a temporal dimension, results from computing the fitted "trend" with xr.polyval were very different from the original variable. It seems that the function is not evaluated on the same coordinate the fit was made on.

What did you expect to happen?

The fit should be made on the same coordinate as the evaluation.

Minimal Complete Verifiable Example

import xarray as xr
import numpy as np
import matplotlib.pyplot as plt

da = xr.DataArray(np.arange(100), dims=('time',), coords={'time': xr.date_range('2001-01-01', periods=100, freq='YS')})

fit = da.polyfit('time', deg=3)

val = xr.polyval(da.time, fit.polyfit_coefficients)

print(da[0], val[0])
# 0,  31.00174342
# The data is linear, the fit should be exact.

plt.plot(da.time, da, label='Original')
plt.plot(da.time, val, label='Fit')

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

No response

Anything else we need to know?

xr.polyval transforms the temporal coordinate here :

xarray/xarray/core/computation.py

Line 2070 in dbb98b4

coord = _ensure_numeric(coord)

And da.polyfit does that here :

xarray/xarray/core/dataset.py

Lines 9069 to 9077 in dbb98b4

	x: Any = self.coords[dim].variable
	x = _floatize_x((x,), (x,))[0][0]
	try:
	x = x.values.astype(np.float64)
	except TypeError as e:
	raise TypeError(
	f"Dim {dim!r} must be castable to float64, got {type(x).__name__}."
	) from e

The results are similar, however, in polyfit the new x starts at 0, while this offsetting is not done in polyval.

This bug was introduced in #9369 I believe. I added a review there woops 😓. Would it be possible to use the same function in both polyfit and polyval ?

I can make a PR. My intuition would be to use xr.core.computation._ensure_numeric in da.polyfit.

Environment

INSTALLED VERSIONS

commit: None
python: 3.12.7 | packaged by conda-forge | (main, Oct 4 2024, 16:05:46) [GCC 13.3.0]
python-bits: 64
OS: Linux
OS-release: 6.11.5-200.fc40.x86_64
machine: x86_64
processor:
byteorder: little
LC_ALL: None
LANG: fr_CA.UTF-8
LOCALE: ('fr_CA', 'UTF-8')
libhdf5: 1.14.3
libnetcdf: 4.9.2

xarray: 2024.10.0
pandas: 2.2.3
numpy: 2.0.2
scipy: 1.14.1
netCDF4: 1.7.1
pydap: None
h5netcdf: 1.4.0
h5py: 3.12.1
zarr: None
cftime: 1.6.4
nc_time_axis: 1.4.1
iris: None
bottleneck: 1.4.2
dask: 2024.10.0
distributed: 2024.10.0
matplotlib: 3.9.2
cartopy: None
seaborn: None
numbagg: None
fsspec: 2024.10.0
cupy: None
pint: 0.24.3
sparse: 0.16.0a10.dev3+ga73b20d
flox: 0.9.12
numpy_groupies: 0.11.2
setuptools: 75.1.0
pip: 24.2
conda: None
pytest: 8.3.3
mypy: 1.13.0
IPython: 8.29.0
sphinx: 8.1.3