NCEP Modeling Synergy Meeting Highlights: April 2, 2007


This meeting was led by Bill Bua and attended by John Ward, Dave Plummer, Geoff DiMego, Ken Campana, Joe Sienkiewicz, Ed Danaher and Joey Carr. The SPC’s Steve Weiss, David Bright, and. UCAR/COMET’s Stephen Jascourt attended by remote conference.


1. CCS


NCO went operational with new massively parallel machines at end of January.  All three (R&D, development, production) are running more-or-less okay, except for security patches. Quarterly upgrade plan for patches for security and bug fixes has now been implemented to reduce impact on model development computer.  The next quarterly upgrade is scheduled for the end of May, and will be done first to R&D, then operational, and finally development. Development/production switching will be less of a problem in the future.


Big implementations in the pipeline include the 1 May 2007 GFS implementation (GSI, hybrid coordinate).  There was a problem with potential MOS degradation that Joey Carr said TDL was working through, but that they will not stop us from going forward.  TDL, however, will need more lead time to test MOS products in the future.  


NCO will conduct its first annual review on April 18-19 to discuss expected computer resources in FY08 and 09, compared to expected NCEP needs for those resources.  The move to the new building will also be discussed, among other issues. Teleconferencing capability will exist, as only SEC will appear in person. At this review, we will begin AOP planning for FY 08.  NCO needs input on what resources will be needed by all centers.




2a. Global Modeling Branch:


Ken Campana reported.  Implementation 1 May 2007. Major changes include GSI, new data for analysis, change in vertical coordinate from sigma to sigma-pressure. The post-processed files include 30 more records in the flux and pressure GRIB files. So far, low level tropical winds, Asian jet and associated cyclogenesis, and short range forecast have all been improved. Stratosphere forecasts are a little better as well.  It was noted that the short range improvement disappears in the medium range.  Website for verification being done by Glenn White can be found at:


Parallel unified post continues to be developed, to go into parallel after the 1 May implementation. Some parameters are done differently in the NAM than in GFS, so one goal of the unified post will be to make these calculations the same way across all models.  For example, RH from NAM not adjusted for ice, unlike the GFS RH. Other examples include tropopause height, vorticity (currently, GFS calculates from spectral coefficients), and so on.  Brent Gordon is gathering differences, talking about how we’ll reconcile differences, and will get comments/objections, if any, to unifying these calculations.


2b. Mesoscale Modeling Branch:


Geoff: The hi-res window will be upgraded, involving replacement of existing WRF code for ARW, increasing horizontal resolution from 5.8 to 5.2 km, and WRF-NMM increase in horizontal resolution from 5.2 to 4 km.  Also WRF-NMM will have new code somewhere between NAM and repository code. The current three domains will be combined into 2 larger domains.


A new NAM implementation will begin testing with frozen code in June. The current NMM will be replaced with a faster version (code speedup for change of order in looping through horizontal and vertical coordinates).  Other changes include physics “tweaks” and expansion of domain for AK and Aviation Weather Center.  Horizontal resolution will stay at 12 km and levels at 60. 


SREF changes will include bias correction and upgrading the WRF code which is presently an unsupported older version, and increased resolution of all members closer to that of the SREF-Eta (32km). 


DNG: (Downscaled Numerical Guidance): smart init tool applied to grids sent over SBN take advantage of native grids – work is progressing well


RTMA: RTMA will be expanded beyond the CONUS to AK. Because there is no RUC for 1st guess in AK, we will use smart init to downscale the NAM in AK to 5-km.  Later, RTMA will expand into HI, Guam (in 2008), and Puerto Rico.  Quality control of radar data was not in charter, but will be pursued for RUC upgrade.


Fire weather: Runs won’t take more than 2 nodes.  Once Eta32 (running solely for Eta MOS) is discontinued, fire weather runs will be able to take the place of Eta32, and would run “over top” of the operational NAM.  Unfortunately, there is no schedule for stopping Eta32 yet, due to feedback from the field wanting continuation of Eta MOS.


Hurricane WRF will be running. Previously, it was planned to run alongside the GFDL model for intercomparison this year, but now there is talk about it replacing the GFDL model this year.


2c. Global Ensemble Prediction System: 


See above.  Resolution increase was to go in Q4 of Fiscal Year 2007.  The upcoming 1 May model change will be reflected in ensembles as of that date (GSI, hybrid sigma-pressure).


2d. Short Range Ensemble Prediction System:


See above.


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


No one was available representing the MMAB. The HYCOM ocean model is running in parallel.


3. Feedback from operational centers


Ocean Prediction Center:  Regarding the HYCOM ocean model, they are currently looking at data, and noticing that there are still problems with Gulf Stream following the bathymetry too closely.  MMAB is working on putting out another parallel that corrects this bias.  OPC is also evaluating the parallel GFS, and noting some changes in extratropical Pacific storms that may be related to the improved east Asian jet forecasts extending out into the western Pacific.


Storm Prediction Center had a question about the hi-res window upgrade timetable.  Geoff DiMego responded that it may be ready for summer implementation. The special runs from Matt Pyle that are currently available are using similar code, but not in quite the same configuration as the contemplated hi-res implementation.


4. The next meeting will be held Monday, April 30, 2007 at noon in EMC Rm. 209, with remote conference capability.