Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

GFS grib2 data download 0.5 Degree

Where can I find NCEP GFS 0.5 Degree Global Forecast Grids (Eg file: gfs.0p50.2020120300.f000.grib2) Historical Archive (say 2019 to till date) data download ?​

Last edited:
Thank you for a timely reply, though it did not solve my case.

In NCAR RDA site when I searched for GFS data. I found 10 search results none matching my run criteria. I have found GFS 0p25 data (ds084.1) not 0p50 or 1p00 data of 2020 /2021/2022.
I have found all the dataset for last 10 days (if it helps someone reading this thread) in this link: Index of /pub/data/nccf/com/gfs/prod

I am basically looking forward for products like precipitation amount, winds (CISC), Geopotential Height (GPH) and other basic met variables for my limited study. Also I have computational limitation in running GFS 0p25 for my domains of Interest.

Is there any down conversion technique to change resolution of gfs.tCCz.pgrb2.0p25.fFFF files to gfs.tCCz.pgrb2.1p00.fFFF.

As a beginner in the community and a WRF enthusiast I request for a solution to my problem or rather a link where I can find the data (will fix it quick).
Is it fine if you download the GFS quarter degree data, then read the data at every other grids. In this case it is way like a 0.5 degree data (although not exactly the same as the original 0.5 degree data).

Another option is that, NOAA has archived some 0.5 degree GFS data years ago. Please ask them whether they still ahve these data available for public use.
I have tried using 0p25 data . However, steps like ungrib.exe, real.exe and wrf.exe are giving either memory overflow errors or the virtual environment is going into guru meditation. I am using gfs data as IC and BC of my regional models.

Can u throw some light on how exactly can I read the data at every other grid as u mentioned?

I have solved the case of Guru meditation crash in my VM while ungrib.exe/ metgrid.exe.

It was basically due to the RAM allocation of VM. I had allocated entire RAM (16/16GB) memory to the VM environment thinking a faster model run. However during high memory usage, the host forced the VM to shut down (Guru Meditation). Now I allocated 12/16 GB to the VM and now, ungrib and metgrid started working smoothly with 0p25 data.

Climbed few steps in the ladder. Thank you so much @William.Hatheway & @Ming Chen for the help.
Now I am stuck at a new issue with real.exe (screenshot)

It is suggesting to scale dt in input file I guess. But I have tried all possible lower dt values but the error prevails. Just as an addon I have already run entire WRF configuration successfully using 0p50 and 1p00 data of 2023 Dec(latest).

problem started when I used 0p25 data of older dates (2020/2021/2022).

Can you look into rsl.out.000, namelist.input and namelist.wps files and see if there are some error or configuration mistake that I can't find.

pasting rsl out file content here (just in-case someone searches the same error)

Domain 1: Current date being processed: 2021-01-04_00:00:00.0000, which is loop # 1 out of 25
configflags%julyr, %julday, %gmt: 2021 4 0.00000000
d01 2021-01-04_00:00:00 Yes, this special data is acceptable to use: OUTPUT FROM METGRID V4.3
d01 2021-01-04_00:00:00 Input data is acceptable to use:
metgrid input_wrf.F first_date_input = 2021-01-04_00:00:00
metgrid input_wrf.F first_date_nml = 2021-01-04_00:00:00
d01 2021-01-04_00:00:00 Timing for input 0 s.
Max map factor in domain 1 = 1.09. Scale the dt in the model accordingly.
Missing surface temp, replaced with closest level, use_surface set to false.
Missing surface RH, replaced with closest level, use_surface set to false.
Missing surface u wind, replaced with closest level, use_surface set to false.
Missing surface v wind, replaced with closest level, use_surface set to false.
ERROR in psfc: flag_psfc = 0, flag_soilhgt = 0, flag_slp = 0, sfcp_to_sfcp = F
-------------- FATAL CALLED ---------------
not enough info for a p sfc computation


  • Screenshot from 2023-12-09 10-05-23.png
    Screenshot from 2023-12-09 10-05-23.png
    235.8 KB · Views: 4
  • namelist.wps
    1 KB · Views: 1
  • namelist.input
    3.8 KB · Views: 1
The data in your met_em files are not complete, i.e., the required variables of PSFC, PMSL, SOILHGT are missing.

Please clarify what data you ungrib and what Vtable you used for this data? If you run with GFS data, please download data from CISL RDA: NCAR RDA Dataset ds084.1