[Ncep.list.emc.fv3gfs_tickets] FV3GFS Ticket #11: Cycling workflow.

[FV3GFS Trac Ticket]

#11: Cycling workflow.
-------------------------+-------------------------------------------------
  Reporter:              |      Owner:  fanglin.yang@…
  rahul.mahajan@…        |     Status:  reopened
      Type:              |  Milestone:
  enhancement            |    Version:
  Priority:  major       |   Keywords:
 Component:  component1  |  workflow,scripting,cycling,utilities
Resolution:              |
-------------------------+-------------------------------------------------

Comment (by rahul.mahajan@…):

 Replying to [comment:79 fanglin.yang@…]:
 Thank you Fanglin.
 This discussion has now been implemented in the branch EXP-cyc. We will
 revisit #2 when it is ready.

 Rahul

 > Based the discussion at a small meeting attended by Moorthi, George,
 Jeff, Rahul, Xu, Daryl and Fanglin, two actionable items were identified.
 >
 > 1) Short-time: the following parameter settings will n be used for
 cycled experiments
 >
 > If [ $cdump = gdas ]; then
 > FTSFS=0; FAISS=0;     FAISL=0
 > FSMCL(2:4)=60
 > FSNOL=99999
 > If [ $cyc = 18 ]; then FSNOL=-2; fi
 > FNTSFA = dump     rgtsst
 > FNACNA = dump   seaince
 > FNSNOA = dump   snow
 > fi
 > If [ $cdump = gfs ]; then
 > FTSFS=90;
 > FAISS=99999;     FAISL=99999
 > FSMCL(2:4)=99999
 > FSNOL=99999           ?????
 > FNTSFA = “   “
 > FNACNA = “  “
 > FNSNOA = “  “
 > fi
 >
 > Essentially, GDAS will run gcycle at kdt=1 like the current offline
 global_cycle to update surface analyses.  GFS cycle will then pick up the
 analyses from its guess, the 6-hr gdas forecast. Note that snow in the GFS
 cycle may not get updated promptly in such a simplified approach.  In
 current operational GFS cycle updates snow as soon as new snow analysis
 becomes available.
 >
 > 2) Long-Term (in a few weeks/months):  George Gayno will update the
 offline global_cycle to read in FV3GFS surface restart file (including
 NSST) on model native grid,  interpolate  observations (snow, sst, ice
 etc) and NSST analyses after running GSI (on Gaussian grid) to model
 native grid, update and write out new sfcanl on model native grid for
 "warm" start. The model will no longer need to read in "sfc increments".
 >
 > Fanglin
 >
 > Replying to [comment:76 rahul.mahajan@…]:
 > > Replying to [comment:75 fanglin.yang@…]:
 > > Fanglin,
 > >
 > > Thank you for your active participation in identifying and resolving
 the issues with cycling the surface variables. I will update the
 {{{FHCYC}}} variable value as per your suggestion.
 > >
 > > Thanks,
 > > Rahul
 > >
 > > > Rahul, we (Xu, George, Helin and I myself ) are actively looking
 into the issue.  In the NEMS GFS world,  gcycle is called offline as part
 of the GDAS/GFS ANAL step to update sfcanl to all avaliable observations
 (including SST, snow, ice, soil moisture and vegetation greeness etc).
 Since in the FV3GFS world gcycle is no longer called, we need to make sure
 those updates are executed as part of the forecast at the first time step.
 Initialy investigation indicates that both the model code and scripts need
 to be updated.
 > > >
 > > > I also noticed you have FHCYC=6 in your cycled experiment.  Please
 change it to 24.
 > > >
 > > > Fanglin

--
Ticket URL: <https://svnemc.ncep.noaa.gov/trac/fv3gfs/ticket/11#comment:80>
fv3gfs <https://svnemc.ncep.noaa.gov/trac/fv3gfs>
NGGPS FV3GFS Development