Comparing attrs that contain a dict of NumPy arrays raises ValueError in identical / assert_identical #11777

What happened?

When an attribute value is a dict that contains a NumPy array with more than one element, comparing two objects that carry this attribute fails:

• DataArray.identical and xr.testing.assert_identical raise "ValueError: The truth value of an array with more than one element is ambiguous", even when the attributes are equal (and also when they differ, instead of returning False / raising AssertionError).
• xr.merge with combine_attrs="drop_conflicts" drops the attribute although both values are equal, and combine_attrs="no_conflicts" raises a MergeError whose message shows two identical values.

Top-level array attributes and lists of arrays already work, because xarray.core.utils.equivalent special-cases ndarray and list. A dict falls through to ==, and dict equality calls bool() on the element-wise array comparison.

What did you expect to happen?

Nested dicts to be compared value by value, the same way top-level attrs are compared by dict_equiv: equal values give True / are kept on merge, differing values give False.

Minimal Complete Verifiable Example

import numpy as np
import xarray as xr

a = xr.DataArray([1, 2], dims="x", attrs={"calibration": {"coeffs": np.array([1.0, 2.0])}})
b = xr.DataArray([1, 2], dims="x", attrs={"calibration": {"coeffs": np.array([1.0, 2.0])}})

a.identical(b)  # ValueError
xr.merge([xr.Dataset(attrs=a.attrs), xr.Dataset(attrs=b.attrs)], combine_attrs="drop_conflicts").attrs  # {}
xr.merge([xr.Dataset(attrs=a.attrs), xr.Dataset(attrs=b.attrs)], combine_attrs="no_conflicts")  # MergeError

Steps to reproduce

No response

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

a.identical(b): ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
merge drop_conflicts attrs: {}
merge no_conflicts attrs: MergeError: combine_attrs='no_conflicts', but some values are not the same. Merging {'calibration': {'coeffs': array([1., 2.])}} with {'calibration': {'coeffs': array([1., 2.])}}

Anything else we need to know?

Possible fix: in xarray.core.utils.equivalent, when both values are dicts, return dict_equiv(first, second), which already uses equivalent for the values. Only plain dicts, not Mapping, so Datasets are not affected. I have a small patch with a regression test and a whats-new entry and would be happy to open a PR.

Disclosure: found, reproduced and the draft fix tested locally with the help of an AI coding agent (Claude Code); reviewed before filing.

Environment