GMTB Workflow Documentation
Configuring GFS

Running GFS requires the following files:

  • psub script
  • configuration file
  • rlist file

The psub script is the script that submits parallel jobs for your experiment. The configuration file contains names of variables and their corresponding values to be used in your experiment and the rlist file contains the list of data used in your experiment.

Currently, versions of these files are not available in the repository. We copied our files from Kate Howard's directory on theia: /scratch4/NCEPDEV/global/save/Kate.Howard/para_gfs/prnems670_SAVE and edited and renamed them for our experiment.

Original File Name Modified File Name
psub psub
NEMS_para_config_T670_T254_unified para_config_T574_NEMS_sasctrl
prnems6701.gsi.rlist sasctrl1.gsi.rlist

Configuration File

Below are the changes we made to the configuration file for our experiment:

PSLOT=sasctrl # Experiment ID
GMTB_HOME=/scratch4/BMC/gmtb/gmtb-tierIII
EXPDIR=$GMTB_HOME/parm/$PSLOT
RLIST=$EXPDIR/${PSLOT}1.gsi.rlist
ACCOUNT=gmtb
group_name=gmtb
GROUP=gmtb
TOPDIR=${GMTB_HOME}
NOSCRUB=${GMTB_HOME}/${PSLOT}
GSD_BASE_SVN=/scratch4/BMC/gmtb/svn
STMP=${GMTB_HOME}/stmp3
PTMP=${GMTB_HOME}/stmp4
ROTDIR=${GMTB_HOME}/${PSLOT}/${CDATE}
BASEDIR=$GSD_BASE_SVN/gfs/$tags/gfs_workflow.$workflow_ver/para
BASE_BUFRSND=$GSD_BASE_SVN/gfs/$tags/gfs.$gfs_ver
BASE_ENKF=$GSD_BASE_SVN/gfs/$tags/gdas.$gdas_ver
BASE_EMCSFC=$GSD_BASE_SVN/gfs/$tags/global_shared.$global_shared_ver
BASE_GSM=$GSD_BASE_SVN/gfs/$tags/global_shared.$global_shared_ver
BASE_GSI=$GSD_BASE_SVN/gfs/$tags/global_shared.$global_shared_ver
BASE_POST=$GSD_BASE_SVN/gfs/$tags/global_shared.$global_shared_ver
BASE_TROPCY=$GSD_BASE_SVN/storm_relocation/tropcy_qc_reloc.$tropcy_ver
BASE_NEMS=$GSD_BASE_SVN/nems/gf_test_new/NEMS
ARCDIR=$GMTB_HOME/archive/$PSLOT
ATARDIR=/BMC/gmtb/2year/${PSLOT}/${CDATE}
RUNLOG=$EXPDIR/${PSLOT}.runlog
gfs_cyc=1 # GFS cycles (00, 06, 12 and 18Z), defaults to 1 (00Z) cycle
fseg=1 # number of AM forecast segments (maximum=3) default=1
npe_node_po=24 # number of pes per node for post step (default 16)
fmax1=240 # maximum forecast hour for 1st segment - default 192hrs
fout1=06 # GFS sig,sfc,flx output frequency for 1st segment, default=3hr
ch1=05 ; cm1=30 ; cs1=00 # Forecast time - 1st segment (JKH: hours, min, sec)
PROCESS_TROPCY=YES
export use_nuopc=.true.
ARCHCOPY=NO # copy ARCR, ARCO files in rlist to on-line archive
export VRFYPRCP=NO # Precip threat scores
export VRFYFITS=NO
export ACPROFit=NO
export VRFYRAD=NO # Radiance data extract
export VRFYOZN=NO # Ozone data extract
export VRFYGENESIS=YES # Genesis tracks
downset=1
VSDB_STEP1=NO # compute stats in vsdb format, default=NO
VRFYG2OBS=NO #grid2obs verification
SEND2WEB=NO # whether or not to send maps to webhost
ARCA00GFS='$ATARDIR/${CDATE}_${PSLOT}.tar'

Rlist File

Below are the changes we made to change which files were archived to the mass store.

Comment out these lines

*/*/arch/ARCA = pgbanl.$CDUMP.$CDATE ### GRIB2 analysis file
*/*/arch/ARCA = gfnf00.$CDUMP.$CDATE ### sigma anlysis file
*/*/arch/ARCA = sfnf00.$CDUMP.$CDATE ### surface analysis file
*/*/arch/ARCA = pgbf*.$CDUMP.$CDATE ### 1.0 degree GRIB2 files

Add this line

*/*/arch/ARCA = pgrbq*.$CDUMP.$CDATE.grib2 ### 0.25 degree GRIB2 files