Enable fetching of archived HRRR from GCP

[nwagenbrenner]

We can subset now with the GDAL /vsigs/ virtual file system driver.

https://gdal.org/user/virtual_file_systems.html

[nwagenbrenner]

@masonwillman Take a look at the nomads class for a reference on how our wx model fetching is implemented. These archived HRRR files are at https://console.cloud.google.com/storage/browser/high-resolution-rapid-refresh. You can see attached for a python script that downloads a full file from the server.

fetchArchivedHrrr.zip

The process we implement in WindNinja will need to clip out a small subset of the full dataset on disk based on our DEM extents using the GDAL /vsigs/ virtual file system driver. This is an example that should work from the command line (you'll need to get a free GCP API key to use and associated client email to populate in the command below in place of the XXXX's).

gdal_translate -b 697 -projwin -113.55 45.00 -112.55 44.00 -projwin_srs EPSG:4326 -of GTiff /vsigs/high-resolution-rapid-refresh/hrrr.20210525/conus/hrrr.t01z.wrfprsf00.grib2 --config GS_OAUTH2_PRIVATE_KEY_FILE XXXX --config GS_OAUTH2_CLIENT_EMAIL XXX out.tif

[nwagenbrenner]

HRRR is also available on AWS:

https://registry.opendata.aws/noaa-hrrr-pds/

[RuiZhang-kwf8]

Fetching for one band is possible now from both AWS and GCP. Currently working on identifying the correct abbreviation for certain parts of the band. Have some questions about this.