What’s New

v0.8.10 (unreleased)

• Fix methods in utils to work with xarray >= 2024.02.0. By Pascal Bourgault.

v0.8.9 (Feb 06, 2023)

• Convert integer (e.g. 1) units to string (e.g. "1") for pint. By Justus Magin.

v0.8.8 (Jan 19, 2023)

• Add conversion between CF geometries and Shapely objects for polygons. By Julia Signell.

• Support 32bit wide bit masks. By Michael St Laurent.

v0.8.7 (Dec 19, 2023)

• Add conversion between CF geometries and Shapely objects for lines. By Julia Signell.

v0.8.5 (Oct 24, 2023)

• Fix for get_bounds_dim_name. By Deepak Cherian.

v0.8.2 (July 05, 2023)

• Better reprs for flag masks. By Deepak Cherian.

v0.8.2 (June 23, 2023)

• Support for flag masks. By Clément Haëck and Deepak Cherian.

v0.8.1 (May 9, 2023)

• Stop bundling the standard name table and use pooch (new optional dependency) to download when needed. By Deepak Cherian.

• Major performance improvements. By Deepak Cherian.

• Fix bug where code was looking for T axis in SGRID info. By Kristen Thyng.

• Support X, Y for rotated pole grids. By Lars Buntemeyer.

v0.8.0 (Feb 8, 2023)

• Support interpreting SGRID Conventions to identify X, Y, Z axes. By Deepak Cherian.

• Add a rich repr. (GH#409). Use rich.print(ds.cf) or %load_ext rich in a Jupyter session to view a much richer representation of the .cf accessor. By Deepak Cherian.

• Support interpreting the grid_mapping attribute: (GH#391). See Dataset.cf.grid_mapping_names() and Dataset.cf["grid_mapping"], DataArray.cf["grid_mapping"]. By Lars Buntemeyer and Deepak Cherian.

v0.7.9 (Jan 31, 2023)

• Fix packaging of v0.7.8. That release was yanked off PyPI.

v0.7.8 (Jan 31, 2023)

• Optionally use the regex package to continue supporting global flags in regular expressions that are not at start of pattern (GH#408). By Kristen Thyng

• Added link to docs for a new example “COSIMA ocean-sea ice model demo” (GH#397). By Aidan Heerdegen

v0.7.7 (Jan 14, 2023)

• Fix to geometry.points_to_cf to support shapely 2.0. (GH#386). By Pascal Bourgault

v0.7.6 (Dec 07, 2022)

• Fix to cf.add_bounds to support all types of curved grids (GH#376). By Pascal Bourgault

• Allow custom criteria to match the variable name of DataArray objects (GH#379). By Mathias Hauser

• Support new Xarray indexes API when creating MultiIndex for “compression by gathering” datasets. (GH#381). By Deepak Cherian

v0.7.5 (Nov 15, 2022)

• cf.add_bounds can estimate 2D bounds using an approximate linear interpolation (GH#370). By Pascal Bourgault.

• Improve detection of bounds order by rellaxing check a bit (GH#361). By Lars Buntemeyer.

• Performance improvements. (GH#358). By Luke Davis

• Fix coordinate/axis detection (GH#359). By Martin Schupfner

v0.7.4 (July 14, 2022)

• Major performance improvement for __getitem_ (GH#349). By Deepak Cherian.

• Raise warning instead of error with malformed cell_measures attributes. Warnings now print names of variables with malformed attributes (GH#350). By Deepak Cherian.

v0.7.3 (June 30, 2022)

• Dataset.cf.guess_coord_axis() now skips known axes/coordinates and only returns a single guess per variable. Additional attributes such as units must be added to known axes/coordinates using Dataset.cf.add_canonical_attributes(). By Mattia Almansi.

• Increased support for cf_role variables. Added Dataset.cf.cf_roles By Deepak Cherian.

v0.7.2 (April 5, 2022)

• added encoder and decoder for writing pandas MultiIndex-es to file using “compression by gathering”. See Encoding and decoding for more. By Deepak Cherian.

• added another type of vertical coordinate to decode: ocean_sigma_coordinate. By Kristen Thyng.

v0.7.0 (January 24, 2022)

• Many improvements to autoguessing for plotting. By Deepak Cherian

• Fix detection of datetime-like variables. By Romain Caneill

• Integrate more unit aliases from xclim (Thanks!). By Deepak Cherian

• Significantly expanded and improved documention. By Deepak Cherian.

• integrate xclim’s CF-compliant unit formatter. GH#278. By Justus Magin.

• Fix dropping of bad variable names. GH#291. By Tom Vo.

v0.6.3 (December 16, 2021)

• Packaging improvements. By Filipe Fernandes.

• cf_xarray now works with Datasets created using _to_temp_dataset. By Pascal Bourgault.

v0.6.2 (December 7, 2021)

• Various bug fixes.

• New cf_xarray.geometry submodule with functions shapely_to_cf`() and cf_to_shapely() to convert between CF-compliant geometry datasets and DataArrays storing shapely geometries (new optional dependency). By Pascal Bourgault.

v0.6.1 (August 16, 2021)

• Support detecting pint-backed Variables with units-based criteria. By Deepak Cherian.

• Support reshaping nD bounds arrays to (n-1)D vertex arrays. By Deepak Cherian.

• Support rich comparisons with DataArray.cf and DataArray.cf.isin() for flag variables. By Deepak Cherian and Julius Busecke

v0.6.0 (June 29, 2021)

• Support indexing by cf_role attribute. By Deepak Cherian.

• Implemented Dataset.cf.add_canonical_attributes() and DataArray.cf.add_canonical_attributes() to add CF canonical attributes. By Mattia Almansi.

• Begin adding support for units with a unit registry for pint arrays. GH#197. By Jon Thielen and Justus Magin.

• Dataset.cf.rename_like() also updates the bounds and cell_measures attributes. By Mattia Almansi.

• Support of custom vocabularies/criteria: user can input criteria for identifying variables by their name and attributes to be able to refer to them by custom names like ds.cf["ssh"]. GH#234. By Kristen Thyng and Deepak Cherian.

v0.5.2 (May 11, 2021)

• Add some explicit support for CMIP6 output. By Deepak Cherian.

• Replace the dims argument of Dataset.cf.add_bounds() with keys, allowing to use CF keys. By Mattia Almansi.

• Added DataArray.cf.formula_terms and Dataset.cf.formula_terms. By Deepak Cherian.

• Added Dataset.cf.bounds to return a dictionary mapping valid keys to the variable names of their bounds. By Mattia Almansi.

• DataArray.cf.differentiate() and Dataset.cf.differentiate() can optionally correct sign of the derivative by interpreting the "positive" attribute. By Deepak Cherian.

v0.5.1 (Feb 24, 2021)

Minor bugfix release, thanks to Pascal Bourgault.

v0.5.0 (Feb 24, 2021)

• Replace cf.describe() with Dataset.cf.__repr__(). By Mattia Almansi.

• Automatically set x or y for DataArray.cf.plot. By Deepak Cherian.

• Added scripts to document coordinate and axes criteria with tables. By Mattia Almansi.

• Support for .drop_vars(), .drop_sel(), .drop_dims(), .set_coords(), .reset_coords(). By Mattia Almansi.

• Support for using standard_name in more functions. (GH#128) By Deepak Cherian

• Allow DataArray.cf.__getitem__() with standard names. By Deepak Cherian

• Rewrite the values of Dataset.coords and Dataset.data_vars with objects returned by Dataset.cf.__getitem__(). This allows extraction of DataArrays when there are clashes between DataArray names and “special” CF names like T. (GH#129, GH#130). By Deepak Cherian

• Retrieve bounds dimension name with Dataset.cf.get_bounds_dim_name(). By Pascal Bourgault.

• Fix iteration and arithmetic with GroupBy objects. By Deepak Cherian.

v0.4.0 (Jan 22, 2021)

• Support for arbitrary cell measures indexing. By Mattia Almansi.

• Avoid using grid_latitude and grid_longitude for detecting latitude and longitude variables. By Pascal Bourgault.

v0.3.1 (Nov 25, 2020)

• Support Dataset.cf.cell_measures. By Deepak Cherian.

• Added Dataset.cf.axes to return a dictionary mapping available Axis standard names to variable names of an xarray object, Dataset.cf.coordinates for Coordinates, Dataset.cf.cell_measures for Cell Measures, and Dataset.cf.standard_names for all variables. Kristen Thyng and Mattia Almansi.

• Changed Dataset.cf.get_valid_keys() to Dataset.cf.keys(). Kristen Thyng.

• Added Dataset.cf.decode_vertical_coords() for decoding of parameterized vertical coordinate variables. (GH#34, GH#103). Deepak Cherian.

• Added top-level bounds_to_vertices() and vertices_to_bounds() as well as Dataset.cf.bounds_to_vertices() to convert from coordinate bounds in a CF format (shape (nx, 2)) to a vertices format (shape (nx+1)). (GH#108). Pascal Bourgault.

v0.3.0 (Sep 27, 2020)

This release brings changes necessary to make cf_xarray more useful with the ROMS model in particular. Thanks to Kristen Thyng for opening many issues.

• vertical and Z are not synonyms any more. In particular, the attribute positive: up now will only match vertical and not Z. Deepak Cherian.

• Fixed tests that would only pass if ran in a specific order. Julia Kent.

v0.2.1 (Aug 06, 2020)

• Support for the bounds attribute. (GH#68, GH#32). Deepak Cherian.

• Add Dataset.cf.guess_coord_axis() to automagically guess axis and coord names, and add appropriate attributes. (GH#67, GH#46). Deepak Cherian.

v0.2.0 (Jul 28, 2020)

• cf_xarray is now available on conda-forge. Thanks to Anderson Banihirwe and Filipe Fernandes

• Remap datetime accessor syntax for groupby. E.g. .cf.groupby("T.month") → .cf.groupby("ocean_time.month"). (GH#64, GH#6). Julia Kent.

• Added Dataset.cf.rename_like() to rename matching variables. Only coordinate variables i.e. those that match the criteria for ("latitude", "longitude", "vertical", "time") are renamed for now. (GH#55) Deepak Cherian.

• Added Dataset.cf.add_bounds() to add guessed bounds for 1D coordinates. (GH#53) Deepak Cherian.

v0.1.5

• Begin documenting things for contributors in Contributing.

• Parse ancillary_variables attribute. These variables are converted to coordinate variables.

• Support Dataset.reset_index()

• Wrap .sizes and .chunks. (GH#42) Deepak Cherian.

  >>> ds.cf.sizes
  {'X': 53, 'Y': 25, 'T': 2920, 'longitude': 53, 'latitude': 25, 'time': 2920}
  

v0.1.4

• Support indexing by standard_name

• Set default xincrease and yincrease by interpreting the positive attribute.

v0.1.3

• Support expanding key to multiple dimension names.