v0.47.0
Contributors to this version: Juliette Lavoie (@juliettelavoie), Pascal Bourgault (@aulemahal), Trevor James Smith (@Zeitsperre), David Huard (@huard), Éric Dupuis (@coxipi).
Announcements
- To circumvent issues stemming from changes to the frequency code convention in
pandas
v2.2, we have pinnedxarray
(< 2023.11.0) andpandas
(< 2.2) for this release. This change will be reverted inxclim
v0.48.0 to support the newer versions. (xarray>= 2023.11.0
andpandas>= 2.2
). xclim
v0.47.0 will be the last release supporting Python3.8.
New features and enhancements
- New functions
xclim.ensembles.robustness_fractions
andxclim.ensembles.robustness_categories
. The former will replacexclim.ensembles.change_significance
which is now deprecated and will be removed inxclim
v0.49.0. (PR/1514). - Added indicator ID to searched terms in the indicator search documentation page. (GH/1525, PR/1528).
Bug fixes
- Fixed a bug with
n_escore=-1
inxclim.sdba.adjustment.NpdfTransform
. (GH/1515, PR/1516). - In the documentation, fixed the tooltips in the indicator search results. (GH/1524, PR/1527).
- If chunked inputs are passed to indicators
mean_radiant_temperature
andpotential_evapotranspiration
, sub-calculations of the solar angle will also use the same chunks, instead of a single one of the same size as the data. (GH/1536, PR/1542). - Fix wrong attributes in
xclim.indices.standardized_precipitation_index
,xclim.indices.standardized_precipitation_evapotranspiration_index
. (GH/1537, PR/1538).
Internal changes
- Pinned
cf-xarray
below v0.8.5 in Python3.8 installation to further extend legacy support. (PR/1519). pip check
in conda builds in GitHub workflows have been temporarily set to always pass. (PR/1531).- Configure RtD search rankings to emphasize notebooks and indicators over indices and raw source code. (PR/1526).
- Addressed around 100 very basic
mypy
typing errors and call signature errors. (PR/1532). - Use the intermediate step
_cumsum_reset_on_zero
instead ofrle
which is sufficient in_boundary_run
. (GH/1405, PR/1530).