Derived climate variables built with xarray.
xclim is a library of functions to compute climate indices from observations or model simulations. It is built using xarray and can benefit from the parallelization handling provided by dask. Its objective is to make it as simple as possible for users to compute indices from large climate datasets and for scientists to write new indices with very little boilerplate.
For example, the following would compute monthly mean temperature from daily mean temperature:
import xclim import xarray as xr ds = xr.open_dataset(filename) tg = xclim.ICCLIM.TG(ds.tas, freq='YS')
For applications where meta-data and missing values are important to get right, xclim provides a class for each index that validates inputs, checks for missing values, converts units and assigns metadata attributes to the output. This also provides a mechanism for users to customize the indices to their own specifications and preferences.
xclim currently provides over 50 indices related to mean, minimum and maximum daily temperature, daily precipitation, streamflow and sea ice concentration.
The official documentation is at https://xclim.readthedocs.io/
xclim is in active development and it’s being used in production by climate services specialists. If you’re interested in participating to the development, want to suggest features, new indices or report bugs, please leave us a message on the issue tracker. There is also a chat room on gitter ().
- Added support for operations on dimensionless variables (units = ‘1’)
- Moved xclim.locales to xclim.core.locales in a batch of internal changes aimed to removed most potential cyclic imports cases.
- Missing checks and input validation refactored with addition of custom missing class registration (xclim.core.checks.register_missing_method) and simple validation method decorator (xclim.core.checks.check).
- New xclim.set_options context to control the missing checks, input validation and locales.
- New xclim.sdba module for statistical downscaling and bias-adjustment of climate data.
- Added convert_calendar and interp_calendar to help in the conversion between calendars.
- Added at_least_n_valid function, indentifying null calculations based on minimum threshold.
- Added support for freq=None in missing calculations.
- Fixed outdated code examples in the docs and docstrings.
- Doctests are now run as part of the test suite.
- Added vectorize flag to subset_shape and create_mask_vectorize function based on shapely.vectorize as default backend for mask creation.
- Removed start_yr and end_yr flags from subsetting functions.
- Add multi gridpoints support in subset.subset_gridpoint.
- Better wrapped_partial for more meaningful inspection.
- Add indices for relative humidity, specific humidity and saturation vapor pressure with a few choices of method.
- Allow lazy units conversion.
- CRS definitions of projected DataSets are now written to file according to Climate and Forecast-convention standards.
- Add utilities to merge attributes and update history in xclim.core.formatting.
- Ensembles : Allow alignment of datasets with same frequency but different offsets.
- Bug fixes in run_length for run-with-dates methods when the date is not found in the run.
- Remove deepcopy from subset.subset_shape to improve memory usage.
- Add missing_wmo function, identifying null calculations based on criteria from WMO.
- Add missing_pct function, identifying null calculations based on percentage of missing values.
- Improvement in FWI: Vectorization of DC, DMC and FFMC with numba and small code refactoring for better maintainability.
- Added example notebook for creating a catalog of selected indices
- Added growing_season_end, last_spring_frost, dry_days, hot_spell_frequency, hot_spell_max_length, and maximum_consecutive_frost_free_days indices.
- Dropped use of fiona.crs class in lieu of the newer pyproj CRS handler for subset_shape operations.
- Complete internal reorganization of xclim.
- Internationalization of xclim : add locales submodule for localized metadata.
- Add feature to retrieve coordinate values instead of index in run_length.first_run. Add run_length.last_run.
- Fix bug in subset_gridpoint to work on lat/lon coords of any dimension when they are not a dimension of the data.
- Refactoring of the documentation.
- Added support for pint 0.10
- Add atmos.heat_wave_total_length (fixing a namespace issue)
- Fixes in utils.percentile_doy and indices.winter_rain_ratio for multidimensionnal datasets.
- Rewrote the subset.subset_shape function to allow for dask.delayed (lazy) computation.
- Added utility functions to compute time_bnds when resampling data encoded with CFTimeIndex (non-standard calendars).
- Fix in subset.subset_gridpoint for dask array coordinates.
- Modified subset_shape to support subsetting with GeoPandas datatypes directly.
- Fix in subset.wrap_lons_and_split_at_greenwich to preserve multi-region dataframes.
- Improve the memory use of indices.growing_season_length.
- Better handling of data with atypically named lat and lon dimensions.
- Added six Fire Weather indices.
- Documentation improvements: list of indicators, RTD theme, notebook example.
- Added sea_ice_extent and sea_ice_area indicators.
- Reverted #311, removing the _rolling util function. Added optimal keywords to rolling() calls.
- Fixed ensembles.create_ensemble errors for builds against xarray master branch.
- Reformatted code to make better use of Python3.6 conventions (f-strings and object signatures).
- Fixed randomly failing tests of checks.missing_any.
- Improvement of ensemble.ensemble_percentile and ensemble.create_ensemble.
- Added a distance function computing the geodesic distance to a point.
- Added a tolerance argument to subset_gridpoint raising an error if distance to closest point is larger than tolerance.
- Created land module for standardized access to streamflow indices.
- Enhancement to utils.Indicator to have more dynamic attributes using callables.
- Added indices heat_wave_total_length and tas / tg to average tasmin and tasmax into tas.
- Fixed a bug with typed call signatures that caused downstream failures on library import.
- Added a _rolling util function to fix memory issues on large dask datasets.
- Added the subset_shape function to subset utilities for clipping region-masked datasets via polygons.
- Fixed a bug where certain dependencies caused ReadTheDocs builds to fail.
- Added many statically typed function signatures for better function documentation.
- Improved DeprecationWarnings and UserWarnings ensemble for xclim subsetting functions.
- Dropped support for Python3.5.
- Added type hinting to call signatures of many functions for more explicit type-checking.
- Added Kmeans clustering ensemble reduction algorithms.
- Added utilities for converting between wind velocity (sfcWind) and wind components (uas, vas) arrays.
- Added type hinting to call signatures of many functions for more explicit type-checking.
- Now supporting explicit builds for Windows OS via Travis CI.
- Fix failing test with Python 3.7.
- Fixed bug in subset.subset_bbox that could add unwanted coordinates/dims to some variables when applied to an entire dataset.
- Reformatted packaging configuration to pure Py3 wheel that ignore tests and test data.
- Now officially supporting Python3.8!
- Enhancement to precip_accumulation() to allow estimated amounts solid (or liquid) phase precipitation.
- Bugfix for frequency analysis choking on time series with NaNs only.
- Added indices to ICCLIM module.
- Added indices days_over_precip_thresh and fraction_over_precip_thresh.
- Migrated to a major.minor.patch-release semantic versioning system.
- Removed attributes in netCDF output from Indicators that are not in the CF-convention.
- Added fit indicator to fit the parameters of a distribution to a series.
- Added utilities with ensemble, run length, and subset algorithms to the documentation.
- Source code development standards now implement Python Black formatting.
- Pre-commit is now used to launch code formatting inspections for local development.
- Documentation now includes more detailed usage and an example workflow notebook.
- Development build configurations are now available via both Anaconda and pip install methods.
- Modified create_ensembles() to allow creation of ensemble dataset without a time dimension as well as from xr.Datasets.
- Modified create ensembles() to pad input data with nans when time dimensions are unequal.
- Updated subset_gridpoint() and subset_bbox() to use .sel method if ‘lon’ and ‘lat’ dims are present.
- Added Azure Pipelines to automatically build xclim in Microsoft Windows environments. – REMOVED
- Now employing PEP8 + Black compatible autoformatting.
- Added Windows and macOS images to Travis CI build ensemble.
- Added variable thresholds for tasmax and tasmin in daily_freezethaw_events.
- Updated subset.py to use date formatted strings (“%Y”, “%Y%m” etc.) in temporal subsetting.
- Clean-up of day-of-year resampling. Precipitation percentile threshold will work without a doy index.
- Addressed deprecations for xarray 0.13.0.
- Added a decorator function that verifies validity and reformats subset calls using start_date or end_date signatures.
- Fixed a bug where ‘lon’ or ‘lon_bounds’ would return false values if either signatures were set to 0.
- Dropped support for Python 2.
- Added support for period of the year subsetting in checks.missing_any.
- Now allow for passing positive longitude values when subsetting data with negative longitudes.
- Improved runlength calculations for small grid size arrays via ufunc_1dim flag.
This is a significant jump in the release. Many modifications have been made and will be added to the documentation in the coming days. Among the many changes:
- New indices have been added with documentation and call examples.
- Run_length based operations have been optimized.
- Support for CF non-standard calendars.
- Automated/improved unit conversion and management via pint library.
- Added ensemble utilities for creation and analysis of muti-model climate ensembles.
- Added subsetting utilities for spatio-temporal subsets of xarray data objects.
- Added streamflow indicators.
- Refactoring of the code : separation of indices.py into a directory with sub-files (simple, threshold and multivariate); ensembles and subset utilities separated into distinct modules (pulled from utils.py).
- Indicators are now split into packages named by realms. import xclim.atmos to load indicators related to atmospheric variables.
This was a staging release and is functionally identical to 0.7-beta.
- Support for resampling of data structured using non-standard CF-Time calendars.
- Added several ICCLIM and other indicators.
- Dropped support for Python 3.4.
- Now under Apache v2.0 license.
- Stable PyPI-based dependencies.
- Dask optimizations for better memory management.
- Introduced class-based indicator calculations with data integrity verification and CF-Compliant-like metadata writing functionality.
Class-based indicators are new methods that allow index calculation with error-checking and provide on-the-fly metadata checks for CF-Compliant (and CF-compliant-like) data that are passed to them. When written to NetCDF, outputs of these indicators will append appropriate metadata based on the indicator, threshold values, moving window length, and time period / resampling frequency examined.
- File attributes checks.
- Added daily downsampler function.
- Better documentation on ICCLIM indices.
- Added total precipitation indicator.
- Fully PEP8 compliant and available under MIT License.
- Added icclim module.
- Reworked documentation, docs theme.
- Added first indices.
- First release on PyPI.
Release history Release notifications | RSS feed
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size xclim-0.17.0-py3-none-any.whl (133.9 kB)||File type Wheel||Python version py3||Upload date||Hashes View|
|Filename, size xclim-0.17.0.tar.gz (35.4 MB)||File type Source||Python version None||Upload date||Hashes View|