Skip to content

Latest commit

 

History

History
188 lines (150 loc) · 8.13 KB

CHANGES.md

File metadata and controls

188 lines (150 loc) · 8.13 KB

Changes in 0.11.5 (in development)

  • Improved support of dataset with unsigned integers
  • Improved determination of chunks (and thereby handling of, e.g., PERMAFROST datasets)

Changes in 0.11.4

  • Maintenance release to handle changes from external package

Changes in 0.11.3

  • Fixed retrieval of data sets with unsigned data types (this affected SNOW datasets)
  • Increased list of supported kerchunk datasets

Changes in 0.11.2

  • Fixed retrieval of time steps of BIOMASS datasets

Changes in 0.11.1

  • Improved support of vector data cubes
  • Improved support of FIRE ECV datasets

Changes in 0.11

  • Migrated code from ESA CCI Toolbox. This encompasses the following changes:
    • Added store aliases esa-cdc and esa-climate-data-centre, esa-cci and esa-cci-zarr
    • Added new data store esa-cci-kc (and corresponding xcube data store esa-cci-kc) that allows performant accessing of selected datasets of the ESA Climate Data Centre using a Zarr view of the original NetCDF files. This approach is made possible by using the kerchunk package. Also added new Notebook that demonstrates usage of the new data store.
    • Added opener "datafame:geojson:esa-cdc" to open data as data frames
    • Added opener "vectordatacube::esa-cdc" to open data as vector data cubes
    • Updated list of dataset states
  • Set up pipeline to publish package on pypi

Changes in 0.10.2

  • Fixed support for climatology datasets

Changes in 0.10.1

  • Fixed support for SST climatology dataset

Changes in 0.10

  • Support Climatology Datasets
  • Ensure compatiblity with Python versions > 3.10. This concerns the way new event loops are created within threads. A deprecated event loop policy has been removed. Solves issues #61 and #64

Changes in 0.9.9

  • Removed cache
  • Pinned pydap version
  • Improved determination of CCI data chunks

Changes in 0.9.8

  • Consider environment parameter READ_CEDA_CATALOGUE to trigger whether metadata from the ceda catalogue shall be read. Enabled by default.
  • Updated list of supported datasets. No odp-provided datasets are excluded anymore.

Changes in 0.9.7

  • Zarr Datastore reads data ids from json file if provided
  • Updated CCI Store to set zarr store (from here on, xcube 0.12.1 is required) #62

Changes in 0.9.6

  • Updated Example Notebooks
  • Extended support of datasets. Previously, datasets which had a chunking of the time dimension larger than one were not handled properly. This affected OZONE in a monthly resolution, SEALEVEL in a monthly resolution, and several ICESHEETS datasets. #12
  • Chunk sizes may now also be decreased, in order to achieve optimal chunkings. For some datasets (e.g., BIOMASS), this increases the performance and helps avoiding memory issues. #48

Changes in 0.9.5

  • Allow reading of datasets that do not specify chunk sizes (e.g., FIRE) Cate #1033.

Changes in 0.9.4

  • Ensure compatibility with xarray >= 0.20.2 and python >= 3.10
  • Ensure compatibility with expected odp update

Changes in 0.9.3

  • Always show time bounds as coordinate, not as data variable
  • Prevent IndexError when requesting data with a time range

Changes in 0.9.2

  • Fixed issue where opening datasets in a daily resolution without a delimiting time range would cause memory errors #56.

Changes in 0.9.1

  • Fixed issue that datasets with spatial bounds could not be opened.
  • Show version as __version__
  • The interface of the method search_data of the CciOdpDataStore has been changed. Search parameters ecv, frequency institute, processing_level, product_string, product_version, data_type, sensor, and platform may now be passed in a dictionary parameter named cci_attrs. This makes it possible again to use the parameter data_type #54.

Changes in 0.9.0

  • Version 0.9 now requires xcube 0.9 because of incompatible API changes in the xcube data store framework.
  • CciOdpCubeOpener has been removed.
  • CciOdpDatasetOpener and CciOdpDataStore now have a new constructor parameter normalize, that may be used to apply normalization steps to the CCI datasets.
  • Set coordinates correctly. All coordinates are present in data descriptions and opened datasets, no coordinates appear as data variables #42.
  • CRS are supported correctly. CRS variables are present in datasets in case the CRS is different from 'WGS84', the CRS information is provided by the data descriptor #50.

Changes in 0.8.1

  • Fixed an issue that caused that occasionally values returned by open_data consisted of random numbers where a fill value would have been expected. (#47)
  • DataDescriptors contain coords
  • Internal virtual file system is built lazily, so opening datasets has become faster.
  • Store parameter method get_search_params_schema has been revised to correctly support all parameter values.
  • Support more datasets from ODP.
  • Fixed support of user_agent parameter
  • Added CCI Zarr Store as convenience store to access cci zarr datasets

Changes in 0.8.0

  • Added user_agent as additional optional store parameter.
  • Provided xcube data store framework interface compatibility with breaking changes in xcube 0.8.0 (see xcube-dev/xcube#420).

Changes in 0.7.0

  • Removed constant-valued parameters from opener schema
  • Renamed store parameters opensearch_url and opensearch_description_url to endpoint_url and endpoint_description_url, respectively.
  • Chunkstore considers bounding box when accessing data. Less data is accessed and normalized. (#33)
  • Fixed time range detection for datasets with daily time frequency.
  • Fixed problem with the encoding of a dataset's coordinate variables that occurs when using xcube_cci with xcube 0.6.0. (#27)
  • Removed CLI

Changes in 0.6.0.

  • Support type specifiers #18. The CCI Store supports type specifiers dataset and dataset[cube]
  • Descriptions of variables and dimensions are different for the same dataset, depending on what type specifier is set.
  • There are now two DataOpeners: The CciOdpDatasetOpener and the CciOdpCubeOpener. Both openers are used by the CciOdpDataStore, with the CciOdpDatasetOpener being the default. The CciOdpDatasetOpener will open any data sets from the CCI Open Data Portal without changing their dimensions. The CciOdpCubeOpener will normalize datasets to have dimensions lat, lon, time (and possibly others). Subsetting is only supported for data cubes. As not all datasets can be normalized to cubes, the CciOdpCubeOpener supports a subset of the datasets that can be accessed with the CciOdpDatasetOpener.
  • Establish common data store conventions (#10
  • xcube-cci can now get the time ranges for satellite-orbit-frequency datasets available via opensearch
  • Introduced new optional parameters to CciStore:
    • enable_warnings
    • num_retries
    • _retry_backoff_max
    • _retry_backoff_base
  • Updated setup.py #16
  • Added opener parameters time_range and spatial_res
  • String-encoded variables are added to datasets as metadata
  • Updated example notebooks

Changes in 0.5.0.

Initial version. This version has been designed to work with the xcube store framework that has been introduced with xcube v0.5.0. It affords

  • a CciOdpDataOpener Implementaton for opening datasets from the ESA CCI Open Data Portal. The Opener has open parameters variable_names, time_period, bbox, and crs.
  • a CciStore Implementation that uses and extends the aforementioned opener and allows for searching the ESA CCI Open Data Portal