EMC/HPC/MPC/NCO/CPC Synergy Meeting Highlights 03/01/04

This meeting was led by Peter Manousos and was attended by Jim Hoke, Hua Lu Pan, Larry Burroughs, Keith Brill, Ed Danaher, Dave Plummer, Zoltan Toth, Bill Bua, and Joe Sienkiewicz.

1. IBM SP and CCS

NCO PMB was unable to attend due to a meeting at NWSH.

2. Notes from EMC

a. Global Modeling Group: No significant changes from last month other than the GFS output is available in BUFR format for same sites as the Eta via FTP 4 times daily - ftp://ftpprd.ncep.noaa.gov/pub/emc/gmb/armbufr/meteogram (3 hourly output out to 180 hours).

b. Mesoscale Modeling Group: Nothing reported.

c. Global Ensembles: Zoltan Toth reported that the Global Ensemble parallel system will be implemented operationally 12Z March 9th (4x daily runs with 1x1 deg resolution out to 7.5 days, 12 members at 00Z and 11 members 06,12, and 18Z cycles). Zoltan also reported that he is working with NCO to bring in the Canadian ensemble output to the NCEP service centers in 3 phases:

1 - Addition of new variables to NCEP’s ensemble output (per request of NCEP and CMC).

2 - Make CMC ensemble output operationally available (in grib format).

3 - Generation of Ensemble mean, spread, and PQPF for NCEP output by September 2004.

An additional phase will combine the NCEP and CMC ensembles. Zoltan also added that ECMWF 1x1 deg resolution 50 member ensemble output is available in grib format to NCEP. A NCEP Office of the Director -led meeting will be held mid to late spring of this year between EMC Ensemble Prediction System experts and NCEP users to determine which directions to focus with regards to ensemble output. A ensemble based bias corrected PQPF will also need to be evaluated this spring by the NCEP service centers.


d. Short Range Ensembles: Nothing reported.


e. Marine Modeling and Analysis Branch (MMAB): No significant changes from last month other than March 2nd is the planned date to implement the global wave/visibility/icing fields out to 180 hours. A system is being designed to obtain objective stats for this output.

3. Input to EMC and other information from Operational Centers

HPC inquired of who to contact at NCO to ensure the ensemble grib output can be viewed in NAWIPS.

OPC asked EMC about the clustering algorithm output status. EMC responded that a more sophisticated algorithm than that utilized pre Cray-fire has not yet fully developed. A meeting will be initiated by HPC/OPC to address this issue. HPC also inquired of EMC on the clarification of the distribution limitation of ECMWF data (i.e. is it possible to get this data to the WFOs in addition to NCEP). Zoltan will be attending a meeting at ECMWF center next week and will raise this issue. A great deal of discussion also centered on the method of parallel evaluation. Ideally the optimal model development/evaluation/implementation system would be flexible enough to allow objective statistical measures to serve as the means of determining if or not an implementation should occur for bundles imparting only minor changes to a modeling system. Conversely, evaluation by NCEP and WFOs would be required for bundles imparting larger changes. Further, output from parallel systems requiring NCEP evaluation should be available in the same format as the operational output.

4. Next Meeting Proposed Monday March 29th, 2004 at noon in room 209.