EMC/HPC/MPC/NCO/CPC synergy meeting highlights 07/01/02
This meeting was led by Peter Manousos and followed a generic agenda denoted by the subtitles below. Attendees included Geoff Dimego, Hua-Lu Pan, Steve Tracton, Jun Du, Bin Bin, Stephen Jascourt, Dave Michaud, Ed Danaher, Joe Carr, James Partain, Keith Brill, and Bill Bua.
1. IBM SP
John Ward reported that components of the new system have been delivered and will be installed July 8, 2002 and tested as a single node system. The production side of the new system will be delivered and installed around October 1, 2002. This implies that all code on the current system will be frozen on this date through when the new system is declared operational (Jan. 2003). Dave Michaud stated that this further implies that all JIFs must be submitted to NCO by early September 2002. The development side of the new system might be delivered in October 2002. This implies that a separate phase of test and evaluation of the development machine might be complete in Jan. or Feb. 2003. John also reported that the T254 implementation will be tested on the B side of the current system for 4 days starting July 11th 2002. If the T254 Global Model runs no more than 20 minutes later than the presently operational T170 version, the upgrade of the Global Model will be implemented operationally around August 13th, 2002. If not however, the T254 implementation will have to wait until the new system has been declared operational. The T254 is projected to just barely fit into the computer schedule, eliminating any cushion for catching up when there are any model delays. This means that if the Eta model is held up even five minutes for late raob data, the catch-up cushion is lost and all later model runs will be delayed. To avoid this domino effect and for logistical reasons to minimize the number of different product delays, when things are running behind schedule, catch-up cushion will be made by running the Eta on the development side of the current supercomputer, which will delay only Eta-related products and for only one model cycle. John also reported that the MRF look alike files will be discontinued in the Jan./Feb. 2003 time frame. Finally, NCF upgraded the uplink system so that the Global Model output is getting to the WFOs 45 minutes to an hour sooner than previously. Dave Michaud reported that the SREF Kain/Frisch members will be implemented July 9th, 2002.
2. Notes from EMC
a. Global Modeling Group: Hua-Lu reported that the relocation algorithm for tropical storms in the ensemble members will be implemented AFTER the T254 has been implemented operationally. The T254 runs will continue to be output on the same one-degree grid used for output from the currently operational T170 model. A parallel run of the Global Model is being run with minor changes to the analysis system. Planned improvements to the radiation parameterization continue to be worked on but will not be ready before the model change moratorium in October.
b. Mesoscale Modeling Group: Geoff Dimego reported that with the potentially tight fit resulting from the implementation of the Global Model upgrade, the Eta will be automatically triggered rather than allowing the NCO Senior Duty Meteorologist to delay the it by 10-15 minutes to catch any late radiosonde observations or communications delays of Mexican and Canadian radio data. This policy will be in effect during the period between the operational implementation of the Global Model upgrade and the operational implementation of the new super computer. It means that the 12z and 00z Eta runs may be missing more raob data than usual but these data will be getting into the assimilation cycle for the “offtime” runs 6 hours later. A bundle slated for implementation in September 2002 will include changes to the 3D Var (to better use satellite radiances and incorporate radial velocities of the WSR 88D), assimilation of cloud top pressure from GOES, and allowing microphysics variables to be cycled and interact more appropriately with other parts of the model including output fields.
c. Global Ensembles: No items to report.
d. Short Range Ensembles: Steve Tracton, Jun-Du, and Bin Bin reported that the Kain/Frisch members of the SREF will be implemented July 9th, 2002. The SREF web site is being upgraded and will feature a better user interface and include precipitation verification statistics.
e. Marine Modeling and Analysis Branch (MMAB) - No items to report.
3. Input to EMC from Operational Centers
No significant items to report.
4. Next Meeting Proposed Monday July 28th, 2002 at noon in room 209.