Get GFS data for WRF

A great place to get the GFS initial and boundary data which we need to run a WRF forecast is here: The NCAR Research Data Archive has 0.25 degree GFS forecast data archived since January 2015.

The first step is to register, so that you can access the archive:

Once you have registered, the archive we want is the NCEP GFS 0.25 Degree Global Forecast Grids Historical Archive, ds084.1: Click the “DATA ACCESS” tab and you will see a range of options to get the data. Click the “Get a Subset” link.

Select the “Initialization (Reference) Time” option, and set both date boxes to the same date-time: the start time for the forecast.

Next, select the following parameters:

  • Geopotential height
  • Ice cover
  • Land cover (0=sea, 1=land)
  • MSLP (Eta model reduction)
  • Pressure
  • Pressure reduced to MSL
  • Relative humidity
  • Soil temperature
  • Temperature
  • u-component of wind
  • v-component of wind
  • Volumetric soil moisture content
  • Water equivalent of accumulated snow depth

Then click “CONTINUE”. It will check your selections, and allow you to get a subset.

  • Under the “Gridded Product” menu, select the Analysis, and all required ?-hour Forecast files you need.
  • Under “Spatial Selection”, choose “Data Within a Bounding Box”. You can select “Draw Box” to draw your area on the map, then refine North, South, East and West in the text box. Make sure that the area you define is quite a bit larger than the WRF domain you plan to run on, to ensure that the grid points needed for interpolation are all included.
  • You may need to “CLICK HERE” at the bottom of the page to update your request, then click “Submit Request”.

Once your request has been submitted, you need to wait until you get an email to say that your data are ready. Once you get the email, sign in to the RDA server at the link provided, and then click on the link after “you will find your data at”. Select the box beside your download file. At the top of the page, you will see a button called “PYTHON DOWNLOAD SCRIPT”. Click on that, and your browser should download a file called something like “”

You probably don’t want to download the data onto your computer, it needs to be on the computer where you will run your WRF forecast. So, use scp to copy to the server where you will run your forecast (e.g. Log on to that server, and move the download script to the directory where you want to download the data (e.g. DATA/GFS/YYYYMMDDHH, where you replace YYYYMMDDHH with the appropriate date, like 2023030800).

Once the download script is in the correct DATA directory, run it from that directory:

python ./ password

Where you replace “password” with your NCAR RDA user password. This should download your data. Use the “tar -xf” command to extract the gfs data from your downloaded file. It’s a good idea to delete “rm” the .tar file now.