This dataset is comprised of the output from kerchunk's scan_grib across the entire AWS-hosted HRRR forecast files, including pressure levels, surface, and sub-hourly, but not native levels.
Each grib message is its own json file, and each init time is its own zip containing the whole extracted json files for all the forecast times for that init time. Once the kerchunk extraction is
complete, the plan is to combine them so that they can be opened in a single call to Xarray as one very large… See the full description on the dataset page:
https://huggingface.co/datasets/jacobbieker/hrrr-kerchunk.
