Problems with GFS.v15.1.1

Having problems running the WRF Preprocessing Sysmte (WPS)?

Problems with GFS.v15.1.1

Postby thorsten » Thu Jun 13, 2019 4:39 pm

Dear All,

since yesterday June 12th at 12Z NCEP implemented the new version GFS.v15.1.1. Since then our operational forecast crashed. We run WPSV3.9.1 and metgrid.exe stops with:

Code: Select all
 ERROR: Error in ext_pkg_write_field
Abort(51980320) on node 0 (rank 0 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 51980320) - process 0


We download GFS from https://nomads.ncep.noaa.gov/pub/data/nccf/com/gfs/prod/ and we use the 0.5º gfs.t12z.pgrb2.0p50.f0* archives. Metgrid works fine with GFS for 00Z from12 of june. So this error has to be due to changes in the version of GFS.

The first thing we noticed (apart from the new URL) was the change in the number of registries. We checked also the grib2 codes Discp, Catgy, Param of the Vtable and they are the same in the new version of GFS. Don't know if this is true for the level type code. The document NCEP published yesterday does not inform about any change in format or grib2 codes.

I imagine there will be other persons who experienced this problem today and in my case it is quite urgent to find a solution.

I appreciate any response from whom has the same problem and managed to fix it or somebody from NOAA who could reveal us the changes they made.

many thanks,

Thorsten
thorsten
 
Posts: 12
Joined: Wed Apr 01, 2009 6:23 pm

Re: Problems with GFS.v15.1.1

Postby dcvz » Mon Jun 17, 2019 2:48 am

See the discussion in this thread: viewtopic.php?f=6&t=11314

An email message was sent to the wrf-news mailing list on May 24, 2019 regarding the impending change to the GFS. Folks can subscribe to that list here: https://mailman.ucar.edu/mailman/listinfo/wrf-news

The fix for the FV3 GFS was distributed with the WPS 4.0 release in June 2018.
dcvz
 
Posts: 187
Joined: Tue Apr 15, 2008 12:02 am

Re: Problems with GFS.v15.1.1

Postby mark.willms » Mon Jun 17, 2019 1:56 pm

We have had the same issue.

A hack-ish solution I found was to remove the 0.4mb Temperature from the grib file which was causing the issues for ungrib.exe.

wgrib2 {FILE} -not 0.4 -grib_out {NEW_FILE_NAME}

You can also use the GFS parallel server as a stopgap until you can either implement WPS 4.0 or use the above method. It serves the old-style GFS data.

If you come across a better method please post it here.
mark.willms
 
Posts: 3
Joined: Wed Jun 12, 2019 4:03 pm

Re: Problems with GFS.v15.1.1

Postby dcvz » Tue Jun 18, 2019 7:51 pm

Yes, removing the 0.4 mb level data is what the V4 ungrib code is doing. For some reason, the FV3 output includes 3 pressure levels with only T and Z. Metgrid requires all the state variables (u,v,t,rh,z) on the input pressure levels. Ungrib interpolates the missing fields to keep metgrid from crashing. However, it can't extrapolate for the 0.4 mb level. Passing the incomplete level to metgrid causes the failure. Removing the 0.4 mb data from the grib file using wgrib2 or grib filter is a good solution. No one should be using wrf to model such high altitudes, so there is no harm in deleting that level.
dcvz
 
Posts: 187
Joined: Tue Apr 15, 2008 12:02 am

Re: Problems with GFS.v15.1.1

Postby Manju » Wed Jun 19, 2019 3:47 am

thorsten wrote:Dear All,

since yesterday June 12th at 12Z NCEP implemented the new version GFS.v15.1.1. Since then our operational forecast crashed. We run WPSV3.9.1 and metgrid.exe stops with:

Code: Select all
 ERROR: Error in ext_pkg_write_field
Abort(51980320) on node 0 (rank 0 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 51980320) - process 0


We download GFS from https://nomads.ncep.noaa.gov/pub/data/nccf/com/gfs/prod/ and we use the 0.5º gfs.t12z.pgrb2.0p50.f0* archives. Metgrid works fine with GFS for 00Z from12 of june. So this error has to be due to changes in the version of GFS.

The first thing we noticed (apart from the new URL) was the change in the number of registries. We checked also the grib2 codes Discp, Catgy, Param of the Vtable and they are the same in the new version of GFS. Don't know if this is true for the level type code. The document NCEP published yesterday does not inform about any change in format or grib2 codes.

I imagine there will be other persons who experienced this problem today and in my case it is quite urgent to find a solution.

I appreciate any response from whom has the same problem and managed to fix it or somebody from NOAA who could reveal us the changes they made.

many thanks,

Thorsten


I had the same problem.WPS 4.0 completely solves the error.
Manju
 
Posts: 4
Joined: Mon Jun 17, 2019 1:18 am


Return to Runtime Problems

Who is online

Users browsing this forum: No registered users and 2 guests