pywgrib2_xr is a Python package to manipulate GRIB2 files. The background engine is wgrib2. wgrib2 is usually known as a standalone executable. Recent releases contain C and Fortran library API. Version 3.0 comes with low-level Python package pywgrib2_s (s for simple). pywgrib2_xr (xr for xarray) decodes GRIB2 files to xarray datasets. 9It is influenced by cfgrib, another Python package to read GRIB (version 1 and 2) files, while trying to address cfgrib issues with some data files originating at NCEP.
The process of reading GRIB2 files is done in two stages. The first one is creation of a template. The template is built by scanning a full set of data files (i.e. without missing data) for a single arbitrary reference time, subject to selection criteria. Template encapsulates dataset structure. The idea comes from AWIPS1 NetCDF template files used to instantiate storage of gridded data.
In the second stage, GRIB2 files are decoded, each message fills a slice in the dataset built from the template. the function open_dataset simply concatenates datasets along the first dimension reftime. This greatly simplifies handling of missing data files.
pywgrib2_xr low-level interface provides the same capabilities as wgrib2 executable, including built-in remapping library iplib.
The documentation is on GitHub pages https://yt87.github.io/pywgrib2_xr
pywgrib2 is is released under BSD Zero Clause Licence https://choosealicense.com/licenses/0bsd