« UCAR Mentorship and... | Main | Unidata is looking... »

09 October 2020

Hi,

MetPy 1.0 Release Candidate 2 has been released. The biggest change is that now XArray DataArrays can be returned from most calculations when they are passed as inputs (this may break some code not expecting this behavior). This is the final planned release candidate before the full 1.0 release (planned in about a month), so any testing and feedback is appreciated, especially in the DataArray support. We should also have a 1.0 upgrade guide available soon in the documentation.

This release can be installed with:

pip install --pre metpy==1.0rc2

conda install -c conda-forge/label/metpy_rc metpy=1.0

Additional changes in RC2 include:

dx/dy and dim_x/dim_y arguments divergence and vorticity are now only passable as keyword arguments
Can no longer set values in a DataArray using the unit_array accessor attribute
lat_lon_grid_deltas now takes a pyproj.Geod instance directly rather than keyword arguments for constructing one.
Support for inputs where Pint Quantity instances are stored within DataArray instances
Fixes errors when passing masked arrays to various CAPE calculations
Support for Albers Equal Area projection
Many fixes updates for NEXRAD format support, including direct reading of file objects from the NEXRAD S3 archive (from e.g. boto3) and using pathlib.Path instances, as well as some updates for NEXRAD ORPG Build 19.0 changes.
Improve default date handling in parse_metar_file
PyProj >= 2.3 is now a required dependency for calculations, and CartoPy is only needed for related plotting capabilities.
PlotObs.level now defaults to None, or no subsetting.
Using standard library modules importlib.resources and importlib.metadata (and their backports) rather than pkg_resources from setuptools. This fixes errors getting MetPy's version from unconventional installs.

For full release notes see the GitHub release page. Let us know if you run into any problems, either at MetPy issue tracker or using support-python@unidata.ucar.edu. You can also ask questions using the "MetPy" tag on Stack Overflow.

Posted by Ryan May [ Comments [1] ]