Forecast Information

[manapaap]

Hi, I'm just getting started with using Herbie to download NWP data, so this might not be a well-formed question.

Is there any means of querying the time step and spatial resolution of a model through Herbie? For instance, if I download GFS data, is there any means of knowing that it is on a 1 hour time step and that its spatial resolution drops from 28 km to 70 km after 7 days? Also, is there any way of querying fxx for < 1 hour for models such as HRRR which may have that data?

[blaylockbk]

Hi @manapaap,

I hope you find Herbie helpful. Thanks for the questions.

Is there any means of querying the time step and spatial resolution of a model through Herbie? For instance, if I download GFS data, is there any means of knowing that it is on a 1 hour time step and that its spatial resolution drops from 28 km to 70 km after 7 days?

Querying information about a model's resolution and timestep and forecast length are purposely not included in Herbie; maintaining a library of that info would be difficult to maintain as models change. However, I have included references for each of the models to help you discover those details for yourself. For example, Herbie's GFS model template file has several links

Herbie/herbie/models/gfs.py

Lines 8 to 13 in f48ba8c

	self.DETAILS = {
	"nomads product description": "https://www.nco.ncep.noaa.gov/pmb/products/gfs/#GDAS",
	"google cloud platform": "https://console.cloud.google.com/marketplace/product/noaa-public/gfs?q=search&referrer=search&project=python-232920",
	"azure document": "https://github.com/microsoft/AIforEarthDatasets#noaa-global-forecast-system-gfs",
	"aws document": "https://registry.opendata.aws/noaa-gfs-bdp-pds",
	}

Each model's output is structured a little differently. In my experience working with new model data, it takes some time to figure out how the model output is organized; but keep at it and you will eventually understand the model output structure better.

Also, is there any way of querying fxx for < 1 hour for models such as HRRR which may have that data?

Yes, refer to #67. In short, the subhourly HRRR data is packed in hourly files. When you request forecast hour 1 fxx=1, you will find it has data at 15 min, 30 min, 45 min, and 60 min forecast. For example, in the index file https://noaa-hrrr-bdp-pds.s3.amazonaws.com/hrrr.20230718/conus/hrrr.t03z.wrfsubhf01.grib2.idx you will find it contains these GRIB messages for 2-m temperature

14:9469518:d=2023071803:TMP:2 m above ground:15 min fcst:
63:50631298:d=2023071803:TMP:2 m above ground:30 min fcst:
112:91096232:d=2023071803:TMP:2 m above ground:45 min fcst:
161:131312922:d=2023071803:TMP:2 m above ground:60 min fcst: