EMC/HPC/MPC/NCO/CPC synergy meeting highlights 07/29/02
This meeting was led by Stephen Jascourt and followed a generic agenda denoted by the subtitles below. Attendees included John Ward, Geoff Dimego, Hua-Lu Pan, Larry Burroughs, Steve Tracton, Jun Du, Bin-Bin Zhou, Jim Hoke, Ed Danaher, Keith Brill, Pete Caplan, and Mark Shirey.
1. IBM SP
John Ward reported that the segmentation fault problems in the T254 global model tests have been resolved and now implementation is on hold waiting for NWS Director Jack Kelly’s decision on whether to proceed with the 20 minute delay compared to the present T170 implementation or wait until the new supercomputer is ready. Apparently, the objections to the model completion time are coming from the private sector [note: the completion out to the same forecast hour is actually earlier with the 20 minute delay than it was with the lower resolution T126 model on the Cray computer 3 years ago]. Jack Kelly will be briefed on August 5. In other news, the code conversion to the new supercomputer is progressing on schedule.
2. Notes from EMC
a. Global Modeling Group: Hua-Lu Pan reported that minor changes to the assimilation of satellite data are presently being tested at T170 for a September implementation. Forecast impact is expected to be minimal.
b. Mesoscale Modeling Group: Geoff Dimego reported three major items:
1) The Nonhydrostatic Mesoscale Model (NMM) with the sigma-pressure hybrid vertical coordinate replaced the Eta model in the nested high-resolution windows runs starting July 17. This is the beginning of the transition toward the Weather Research and Forecasting (WRF) model, a development to which the field will need to pay close attention.
2) 12-km fields on the CONUS 12-km grid #218 and an Alaska grid are finally getting into AWIPS now, though only a dozen fields are presently available on this grid.
3) The Fall bundle will, as noted last month, include assimilation of NEXRAD radial velocities (VAD wind profiles already are in), upgrades to satellite radiance processing, and cycling of microphysics fields and updates to the radiation making it consistent with the microphysics that were introduced last November. However, the GOES cloud top assimilation is postponed until early 2003, primarily because the development machine has been used for testing the global T254 model and thus unavailable to sufficiently test the cloud top assimilation before the moratorium on model changes when the new supercomputer is being tested this fall.
Additionally, it was noted that the RUC 3d-var will not be able to be ready to run before the model change moratorium, so it will probably get implemented early next year on the new computer.
c. Global Ensembles: Hua-Lu Pan reported that two changes will occur in September: 1) the initial perturbations have been too large in the southern hemisphere, so they are being rescaled, 2) the breeding cycle will be reduced to 6 hours, consistent with running the ensemble system every 6 hours now. The hurricane relocation for each ensemble member has not produced good results, so its implementation is on hold.
d. Short Range Ensembles: Steve Tracton reported that the Kain/Frisch members of the SREF are still officially in parallel mode but their operational implementation is imminent. Better yet, the SREF suite is expected to become operational NWS-wide early this fall. NWS Headquarters is presently preparing requirements documents to enable this. Even so, nobody knows when SREF fields will make it into AWIPS. Preparations are underway for FSL to run RUC members of the SREF using Eta initial and boundary conditions, and there are plans for CAPS to run ARPS members soon, though they have not decided whether their members will have initial perturbations or variations in model physics. Meanwhile, the SREF web site has been redesigned to be much more user-friendly. The trial new version is linked on the SREF home page, which is at http://wwwt.emc.ncep.noaa.gov/mmb/SREF/SREF.html
e. Marine Modeling and Analysis Branch (MMAB) - Larry Burroughs reported that in June, the hurricane wave model began ingesting winds hourly rather than 6-hourly from the GFDL hurricane model and implemented a flexible radius of influence, resulting in smoother wave fields. Upcoming changes include a request to run the wave models four times daily and to increase output frequency to three hours for vessel icing and open ocean fog. Additionally, the regional wave models will be getting distributed into AWIPS through patches in the current AWIPS build rather than having to wait for the next build.
3. Input to EMC from Operational Centers
The Eta model has a precipitable water dry bias that increases in magnitude during the forecast. Geoff DiMego responded that the problem is known and already being investigated; the causes are not yet known.
4. Next Meeting Proposed Monday August 26th, 2002 at noon in room 209.