NCEP Synergy Meeting Highlights: April 27, 2009


This meeting was led by David Novak and attended by Chris Caruso Magee (NCO); John Ward (EMC); Eric Rogers (EMC); Yuejian Zhu (EMC); Mary Hart (EMC); Bill Bua (COMET/NCEP),Stephen Jascourt (COMET/NCEP); Mike Brennan (TPC); Jim Hoke and Keith Brill (HPC);Joe Sienkiewicz (OPC); Steve Silberberg (AWC); David Bright (SPC); Kathy Gilbert (MDL); Bernard Meisner (Southern Region); Pete Browning (Central Region); David Myrick (Western Region); and Carven Scott (Alaska Region).


1. NCO


Chris Caruso Magee reported that Cirrus is running well. There is one issue with validation of the GFDL output that is being investigated.


The Power-6 stratus machine is being installed in Gaithersburg MD, with preparatory infrastructure work proceeding on schedule. The computer moratorium is scheduled, per the IBM contract, to end no later than October of 2009, and will most likely end in July of 2009. Once the moratorium ends, the SREF implementation will occur, followed by the GEFS and HWRF implementations.


Any implementation questions should be directed to Chris Caruso Magee and Dan Starosta.




2a. Global Climate and Weather Modeling Branch (GCWMB)


John Ward reported that the next GFS implementation contains a few bug fixes and post-processing changes. Post-processing changes to the GDAS are being made to match the GDAS post to that of the GFS. Currently, a “null” test to produce identical results to operations cannot be run with the current parallel GFS. GCWMB wants to be able to get identical results as a “sanity check” prior to running parallel tests.  The problem has been isolated to the aerosol routine in the code, and the global branch continues to work on the issue. A TIN will be posted soon. 


Testing is being done on a new semi-Lagrangian model which will be run at T878 (~0.4 degrees, or ~20 km horizontal grid spacing). Tentative plans are for a March 3, 2010 implementation.


2b. Mesoscale Modeling Branch (MMB)


WRF-NMM Eric Rogers reported that the WRF-NMM has a parallel on the Power-6 cirrus computer (in Fairmont, WV). Testing for minor bug fixes is occurring, including testing an improvement in the NAM GOES look-alike product.



MMB is recasting MMB models into the National Environmental Modeling System (NEMS) framework. Of particular interest is that the current WRF-NMM run in the NAM slot will be replaced with the NEMS-NMMb (“b” is the grid type) run, scheduled sometime in FY2010.


2c. Short Range Ensemble Forecast System (SREF)


SREF is running in parallel on the Power-6 cirrus computer. A bug fix to the NDAS was made in early April. This fix corrects an interpolation error in the NDAS skin temperature for the SREF Eta members. The SREF implementation is first on the list once the moratorium is lifted.


2d. Global Ensemble Prediction System


Yuejian Zhu is testing a parallel GEFS run which increases member resolution from T126 to T190, adds a stochastic perturbation scheme, and uses 8th order horizontal diffusion for all resolutions. Tests show improved forecast skill and reliability. Additionally, 27 new variables will be included in the “pgrba” files, including 10, 50, and 100 mb level data, surface and top of atmosphere flux data, and soil data.


This GEFS upgrade is second in the queue when the computer moratorium is lifted (see above section 1).


2e. Marine Modeling and Analysis Branch


No report.





Kathy Gilbert asked NCO to change all “NMM” references to “NAM” on Cirrus. NCO will do so.


NCEP Centers and NWS Regions


SPC noted that dbnet is working well on Cirrus.


HPC reported a high-bias for precipitation in the High-Res Window NMM run, especially precipitation related to convection. The bias appeared less pronounced in the ARW runs. EMC confirmed that the NMM high-bias appears in objective verification statistics (especially for higher precipitation thresholds in the warm season) and that the bias is less in the ARW.  It was suggested that a combination of factors may be contributing to the bias, including the convective permitting (but not resolving) grid scale, and that additional tuning of physics schemes may be necessary.


No issues were brought up by the regions participating in the call.



John Ward noted that EMC is investigating the use of a skill score for overall model performance, following the “UKMet Index” concept. For example, the score would be a combination of other skill scores (500 mb height anomaly correlation, RMSE 250 mb winds, RMSE MSLP, etc) with various weightings. The score would be normalized against persistence. EMC has begun investigating applying such a method to the GFS.


John Ward also mentioned that EMC is investigating the use of a “decision index” to assist in making implementation decisions. Since there are varying degrees of improvement for various parameters for any given implementation, it can be difficult to objectively determine overall impact. This index would be composed of skill scores from several elements (weighted), and normalized against the control run. John plans to send out a strawman proposal soon.


5. The next Synergy Meeting will be held at 2:30 pm EDT on Tuesday, 26 May 2009 in room 209, with remote teleconferencing capability.