NCEP Modeling Synergy Meeting Highlights: June 26, 2006


This meeting was led by Bill Bua and attended by Mark Iredell, Bill Bua, Geoff Dimego, Joe Carr, Keith Brill, Dave Plummer, John Ward, Mary Hart, Joe Sienkiewicz and David Michaud.  Additionally, SPC’s Steve Weiss, AWC’s Steven Silberburg, and UCAR/COMET’s Stephen Jascourt attended by remote conference.


1. CCS

John Ward.  The frost and snow computers have been decommissioned.  Installation has begun on the “mist” computer, and aqua is up and running and being used.  Mini-on-D system.  An implementation moratorium will begin on or shortly after 8 Aug. and will continue until mid-January 2007 while installation and testing of the new IBM MPI system proceeds.  Before the moratorium, we have the  4x day SREF, a mini-package for the RUC, the realtime mesoscale analysis (RTMA) done off the RUC, the hybrid GFS, and the Great Lakes application of the WaveWatch3.  From TDL, the gridded MOS is expected in July.


On the TOC moratorium:  an initial operating capability (IOC) test has been successfully completed.  Data was able to be passed to awips successfully for a full week.  Final acceptance, however, will not take place until all external input/output circuits to new system are added.  With the successful IOC, though, the TOC moratorium is pretty much over.  The new TOC will support AWIPS 7.2.  It was noted that NCEP also got data for 2 weeks from the TOC successfully on our dev machine.


John and Dave Michaud also reported that NCO kept up with changes to the TOC so that there would be no hold up when their moratorium was lifted. 




2a. Global Modeling Branch:


Mark Iredell reported that the code for a minor implementation parallel will be run by NCO starting tomorrow, with implementation currently scheduled for 25 July 2006.  Also available upon request is a series of summer reruns for TPC and the nested hurricane model, along with a cold season retrospective from 15 December 2005 through 15 January 2006.  TPC has signed off on the implementation based on the warm season reruns, including Dennis, Katrina and Rita.  Results showed that forecast track skill was the same, while forecast strength was better for hurricanes.


Changes to GFS will include:


1.     Forecast part of the model (as opposed to data assimilation) will be changed to hybrid sigma-pressure.  Transition starts at the 1st sigma level and ends at about 70-hPa, where coordinate is completely pressure.  With surface pressure at 1000-hPa, no difference in the location of hybrid sigma-pressure vs. sigma surfaces is noted, with largest differences in locations with high model terrain.  Some improvement is noted in the forecast evolution of the Asian jet as a result of this change. 

2.     A few other improvements were made that had been on the “back burner”: 

a.     Ozone physics,

b.     Glacial ice physics, and

c.     Antarctica land/ice elevation problems. 


The Medium-Range Ensemble Forecast system and data assimilation will still be in sigma for now.  HPC will have data available for the parallel GFS via server. 


GFS MOS:  TDL should have enough data to adjust the GFS MOS equations for this implementation. 


2b. Mesoscale Modeling Branch (MMB):


Geoff Dimego reported that the WRF-NMM was implemented on 20 June. A 32-km Eta is being run to feed MOS.  SREF will soon be run to 87 hours and 4x per day (03, 09, 15, and 21 UTC).  Implementation will hopefully take place 28 or 29 June (depending on critical weather in the Eastern U.S.). 


Real-Time Mesoscale Analysis (RTMA):  The RTMA is a surface 2D-VAR assimilation system using the Gridscale Statistical Interpolation code, and is being evaluated locally. It was to be available for each hour, 35 min after the hour.  A problem with mesonet data has cropped up, however, with FSL/MADIS, so data dumps needed later cutoff (30-35 min after the hour). This results in a delay of about 15 min.  The RTMA will be going out with AWIPS build 7.2.  The MADIS problem with delays results from the slowness of their computers.  They may be able to upgrade later in the year with additional funding.  Alternately, the analysis could get data from Western Region, or supplement MADIS with Mesowest data from a Western Region feed.


MMB hopes to restart fire weather runs next fiscal year.  The hi-res window run code will also be updated to reflect the current NCEP WRF-NMM and NCAR-ARW models.


2c. Global Ensemble Prediction System: 


Mark Iredell reported that the MREF group is still recovering from last month’s MREF implementation.  They hope to be going from 14 to 20 members in Feb 2007 (after the computer moratorium).


2d. Short Range Ensemble Prediction System:


See above.  Plans for next year include bias correction, and upgrade of model, etc. 


2e. Marine Modeling and Analysis Branch (MMAB): 


The Great Lakes WaveWatch3 model is about ready to be implemented.  An upgrade (assimilation improvements) to the Ocean Forecast System (OFS) is to run in parallel, but probably will not be implemented until after the moratorium.  Running of the Lake Erie National Ocean Service (NOS) model is being taken over by NCO.






Keith Brill asked about the Unified Post.  Mark Iredell responded that a new position has been advertised for doing that work.  Once hired, the hiree will take about a year or more after that to complete the unification of the postprocessed output from the GFS.


SPC requested operational GFS deterministic output files in the same time and space resolution as the GFS medium range ensemble members.  Presently, the high-resolution deterministic GFS time and space resolution decreases after F180 (12 hrly and 2.5 degree, respectively), while the recent upgrade to the MREF provides one degree and 6 hourly output through F360 for each of the 14 members.  These differences make it difficult to incorporate the high-res GFS deterministic forecast among the other ensemble members.

Dave Michaud agreed and suggested we include the request at the next EMC NTOP discussion.


4. The next meeting will be held Monday August 14 at noon in EMC Rm 209 with remote conference capability.