Version 0.13 Updates

Version 0.13.8

Fixes

• Fixed issue when kwargs passed to the Xarray engine as a dict were modified in-place (#671).

Version 0.13.7

Fixes

• Fixed issue when Xarray created from GRIB with chunks failed in computations (#669).

Version 0.13.6

This release reverts the changes introduced in version 0.13.5, which caused a regression (#663). The issue will be addressed in a future release.

Version 0.13.5

Fixes

• Fixed issue when NetCDF data with 1D latitudes/longitudes appearing as variables was not parsed into a fieldlist when read with from_source() (#658).

Warning

This release introduced a regression (#663) and was reverted in version 0.13.6. The issue will be addressed in a future release.

Version 0.13.4

Fixes

• Fixed issue when iterating through the keys and values in some BUFR messages failed with KeyError (#653)

Version 0.13.3

Fixes

• Fixed issue when could not convert ZIP containing a single NetCDF file to Xarray (#650)

Version 0.13.2

Fixes

• Fixed issue when could not read GRIB data from FDB streams in multiprocessing (#647)

Version 0.13.1

Fixes

• Fixed issue when to_target() could not assign an encoder to the “.grib1” and “.grib2” suffixes (#643)

Version 0.13.0

Deprecations

• The “settings” has been renamed to config

• settings.auto_save_settings is now config.autosave

• Data object save() is deprecated

• Data object write() is deprecated

• new_grib_output() is deprecated

• GribOutput is deprecated

• new_grib_coder() is deprecated

• GribCoder is deprecated

Configuration

• The “settings” has been renamed to config. The API did not change with the exception of settings.auto_save_settings, which now is config.autosave. The “settings” object is still available for backward compatibility but will be removed in a future release. Users are encouraged to migrate the code to use config instead. (#586)

• The configuration file changed to ~/.config/earthkit/data/config.yaml. When it is not available, the old configuration file at “~/.config/earthkit/settings.yaml” is loaded and saved into the new path. This is done until “settings” is removed.

• As new feature, the configuration file can be specified via the EARTHKIT_DATA_CONFIG_FILE environmental variable. The environmental variable takes precedence over the default configuration file (it is only read at startup).

See here for more details.

New writer API

• Introduced a new Writer API to facilitate the creation of custom data writers. It is based on targets that can represent a file, a database, a remote server etc and are able write data by using a suitable encoder (#596).

• The existing writing mechanisms are still kept ensuring backward compatibility, but marked deprecated and will be removed in a future release. For all deprecated methods/objects, see as follows:

  • Data object save() is deprecated

  • Data object write() is deprecated

  • new_grib_output() is deprecated

  • GribOutput is deprecated

  • new_grib_coder() is deprecated

  • GribCoder is deprecated

See the notebook examples:

• Writing to a file target

• Writing GRIB to a file target

• Writing GRIB to file-pattern target

• Writing GRIB to an FDB target

• Converting GRIB to GeoTIFF

• GRIB encoder

New features

• Refactored wekeo and wekeocds to use hda version 2 (#593). The minimum hda version is now 2.22.

• Added support for patterns with dates using timedelta as strftimedlta() for the file-pattern source (#606)

• Enabled using string formatter for output file patterns in new_grib_output and GribOutput (#603)

• Enabled creating list-of-dicts fieldlists without latitudes/longitudes (#636)

• Added cpu() to the torch backend (#578)