I am downloading grib2 files from the NWS NOMADS GFS service details here

The files are provided on a 4x per day cycle and the date format is YYYYMMDDHH. I believe that the processing is actually done 4 times per day.

I am a little confused about the naming convention and what time to actually go and get the files.

An example url using the filter service is: http://nomads.ncep.noaa.gov/cgi-bin/filter_gfs.pl?file=gfs.t00z.pgrbf00.grib2&lev_surface=on&var_PRES=on&leftlon=0&rightlon=360&toplat=90&bottomlat=-90&dir=%%2Fgfs.YYYYMMDDHH

My understanding is that he HH can be 00, 06, 12, 18.

  • How does this file naming system work?
  • Does any one have a script that downloads this data on a scheduled basis and figures out what the file names are?
  • does this file have forecast weather available in advance?

1 Answer 1


If you want GFS, you can just download them using a directory based approach (a more RESTful design) instead of the filter approach (which is potentially subject to change).

For example, GFS 1x1 for today (2013-10-22) at 1800 can be downloaded from: http://nomads.ncep.noaa.gov/pub/data/nccf/com/gfs/prod/gfs.2013102218/master/

You can choose which areas you want. As an example: http://nomads.ncep.noaa.gov/pub/data/nccf/com/gfs/prod/gfs.2013102218/master/gfs.t18z.mastergrb2f00 (note that the 18 occurs twice).

The winds in GFS are forecast (even if historical) - its Global Forecast System. Note that there are assimilation activities as well, but those are packaged separately. Typically it'll be updated every six hours, and valid for 48 hours, 72 hours or sometimes longer (4-14 days) ahead.

More details at http://nomads.ncep.noaa.gov/txt_descriptions/GFS_high_resolution_doc.shtml and the links off that page

  • I know this info is probably available somewhere, but what is the frequency of creation of the files, once per day, every 4 hours? I am a little confused over the naming and frequency. Oct 23, 2013 at 5:04
  • Those files are quite large. I only want wind, temp and pressure at the surface for now, which is why I adopted the filtered approach to reduce the data transfer size and extraction times using wgrib2 on my servers. What is your thoughts, whole file and extract or filtered approach? Oct 23, 2013 at 5:09
  • Updated to show issue and validity. The extra questions are just that - things that belong in another question.
    – BradHards
    Oct 23, 2013 at 6:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.