Skip to content

Releases: xCDAT/xcdat

v0.10.1

01 Oct 21:41
7ae6b72

Choose a tag to compare

v0.10.1 (01 October 2025)

This patch release introduces bug fixes, documentation updates, and DevOps changes, including constraining cf_xarray >=0.10.7, preserving Xarray legacy behaviors for open_mfdataset() with compat and join arguments to prevent unexpected breaking changes for users, dropping Python 3.10 support, and adding PyAOS to the endorsements list.

Bug Fixes

  • Constrain cf_xarray >=0.10.7 by @tomvothecoder in #804. This constraint is necessary because cf_xarray=0.10.7 introduces a fix to bounds_to_vertices(), enabling support for non-monotonic bounds and propagating correct regridding results in xESMF for such cases (refer to xarray-contrib/cf-xarray#579).

Documentation

DevOps

  • Preserve Xarray legacy compat and join behaviors by @tomvothecoder in #800. This change was made to maintain consistent functionality across environments after Xarray updates, preventing unexpected breaking changes for downstream users.
  • Drop Python 3.10 support by @tomvothecoder in #794

Full Changelog: v0.10.0...v0.10.1

v0.10.0

27 Aug 20:12
8335789

Choose a tag to compare

v0.10.0 (27 Aug 2025)

This release introduces a new weight threshold option for temporal operations, improves regridding alignment, and fixes multiple issues with xgcm.Grid and xESMF integration. It also includes a DevOps fix to maintain consistency in handling xgcm.Grid metadata.

Enhancements

Bug Fixes

Full Changelog: v0.9.1...v0.10.0

v0.9.1

08 Jul 18:59
0d7e112

Choose a tag to compare

v0.9.1 (08 July 2025)

This patch release includes enhancements to bounds handling, improved error messages, and documentation updates.

Enhancements

Bug Fixes

Documentation

Full Changelog: v0.9.0...v0.9.1

v0.9.0

02 Jun 18:53
6bf8885

Choose a tag to compare

v0.9.0 (02 June 2025)

This release introduces key functional and performance enhancements across regridding, averaging, and coordinate handling workflows. Users can now output regridding weights and infer vertical target data, improving flexibility in vertical and horizontal transformations. Time frequency inference logic is more robust through median delta computation, and new options for skipna handling and spatial weight thresholds support more accurate and configurable statistical operations. Additional improvements to coordinate handling enable support for curvilinear grids. The release also includes critical bug fixes, expanded documentation, and updated compatibility for modern Python environments by dropping Python 3.9 and adding compatibility with Python 3.13.

Enhancements

  • Adds option to output regridding weights and create mask for regridding by @jasonb5 in #752
  • Adds ability to infer target data for vertical regridding by @jasonb5 in #756
  • Use the median of the delta instead of min for time freq inference by @chengzhuzhang in #768
  • Add weight threshold option for spatial averaging by @pochedls in #672
  • Enable skipna for spatial and temporal mean operations by @lee1043 in #655
  • Enhance coordinate handling and add curvilinear dataset support by @tomvothecoder in #736

Bug Fixes

  • Eliminate performance bottleneck in temporal.group_average() by @will-s-hart in #767
  • Fix incorrect dimension used for temporal weights generation by @tomvothecoder in #749

Documentation

DevOps

New Contributors

Full Changelog: v0.8.0...v0.9.0

v0.8.0

15 Feb 00:06
b260fcb

Choose a tag to compare

v0.8.0 (14 February 2025)

This minor release introduces a new feature for temporal averaging with custom seasons spanning the calendar year. It also includes the ability to detect and drop incomplete seasons using the drop_incomplete_season config, which will eventually replace drop_incomplete_djf (previously limited to DJF seasons). Additionally, a bug in the regrid2 regridder has been fixed, ensuring coordinates are now preserved correctly.

Enhancements

Bug Fixes

Documentation

DevOps

Full Changelog: v0.7.3...v0.8.0

v0.7.3

06 Nov 20:51
1cfd369

Choose a tag to compare

v0.7.3 (06 November 2024)

This patch release updates the NumPy constraint to numpy >=2.0.0,<3.0.0 to ensure compatibility with NumPy 2.0 (which introduces breaking changes). It also fixes a bug in the get_bounds() method where bounds could not be found on supported non-CF axes (e.g., "latitude", "longitude", etc.) even with the "bounds" attribute set on the axes.

Bug Fixes

  • Update get_bounds() to support mappable non-CF axes using "bounds" attr by @tomvothecoder in #708

Documentation

DevOps

Full Changelog: v0.7.2...v0.7.3

v0.7.2

02 Oct 18:52
f9d1566

Choose a tag to compare

v0.7.2 (02 October 2024)

This patch release introduces significant performance improvements to the temporal grouping averaging APIs (group_average, climatology, and departures) and adds support for piControl and other simulations that have time coordinates starting at year 1 (e.g., "0001-01-01") when dropping incomplete seasons.

Enhancements

  • [Refactor] Improve the performance of temporal group averaging by @tomvothecoder in #689

Bug Fixes

  • Update temporal.py to properly handle piControl and other simulations that start at year 1 when dropping incomplete seasons by @lee1043 in #696

Documentation

DevOps

Full Changelog: v0.7.1...v0.7.2

v0.7.1

24 Jun 20:56
1d57e25

Choose a tag to compare

v0.7.1 (24 June 2024)

This patch release fixes a bug in the Regrid2 API where a static order of dimensions are incorrectly expected. It updates add_missing_bounds() to convert np.timedelta64 values to pandas.Timedelta objects to support Xarray's datetime component accessor.

This release also includes numerous updates to the documentation, including adding a general guide to parallel computing with Dask notebook (link). It also ensures all existing notebooks and documentation are up to date with the latest and relevant information.

Bug Fixes

  • Fixes regrid2 mapping output to input ordering by @jasonb5 in #653
  • Update add_missing_bounds() to convert np.timedelta64 to pd.Timedelta to support Xarray's datetime component accessor by @lee1043 in #660

Documentation

Full Changelog: v0.7.0...v0.7.1

v0.7.0

10 Apr 20:40
799b81f

Choose a tag to compare

v0.7.0 (10 April 2024)

This minor release includes enhancements to the performance of the Regrid2 API and fixes Regrid2 to align the behavior of how missing values are handled with CDAT. There are various other bug fixes and documentation updates listed below.

Enhancements

  • Improving regrid2 performance by @jasonb5 in #533
  • Update Regrid2 missing and fill value behaviors to align with CDAT and add unmapped_to_nan arg for output data by @jasonb5 in #613

Bug Fixes

  • Fix Regrid2 to convert bounds as Dask Arrays to NumPy Arrays for compatibility with NumPy based code by @tomvothecoder and @lee1043 in #634
  • Fix climo notebook missing T bounds and add notebook env setup in all example notebooks by @tomvothecoder in #623
  • Update unweighted temporal averages to not require bounds by @tomvothecoder in #579

Documentation

Deprecations

  • Remove deprecated features and APIs by @tomvothecoder in #628, including:
    • horizontal_xesmf() and horizontal_regrid2()
    • **kwargs from create_grid()
    • add_bounds accepting boolean arg in open_dataset() and open_mfdataset()
    • Remove CDML/XML support from open_dataset() and open_mfdataset() since CDAT is EOL since Dec/2023

Full Changelog: v0.6.1...0.7.0

v0.6.1

29 Nov 17:46
104a5d5

Choose a tag to compare

v0.6.1 (29 November 2023)

This patch version adds a default value to the axes argument in ds.bounds.add_missing_bounds() (axes=["X", "Y", "T"]). The axes argument was added in v0.6.0 and did not have a default value, which inadvertently introduced a breaking change to the API.

xesmf is now a required dependency because its core library, ESMF, supports Windows as of Feb/2023. More information can be found here.

Bug Fixes

DevOps

Documentation

  • Update doc: Add link to the ESFG seminar xCDAT introduction video by @lee1043 in #571
  • Fix v0.6.0 changelog headers for proper nesting by @tomvothecoder in #559

Full Changelog:

v0.6.0...v0.6.1