EMC/HPC/MPC/NCO/CPC synergy meeting highlights 09/24/01
This meeting was led by Peter Manousos and followed a generic agenda denoted by the subtitles below. Attendees included Geoff Dimego, Hua-Lu Pan, Jim Hoke, Steve Tracton, John Ward, Pete Caplan, Zoltan Toth, Jim Hoke, Kevin McCarthy, Dave Reynolds, Ed Danaher, Steve Jascourt, Keith Brill, Bill Bua, James Partain, and Steve Lord and Jeff Boroughs.
1. IBM SP
John Ward reported that the upgrade of the SP occurs in a multi-step process - upgrade side B, monitor for problems, if ok move operations to side B to then upgrade side A, and finally move operations back to side A. There have been numerous hiccups with the first step (upgrading side B) over the recent days. However, side B seems to be stable as of last Friday. Therefore, operations will be moved to side B this week and side A will then be upgraded this weekend. The Climate Run's first internal operational test and evaluation started Sept 6 and will first produce products Oct 6 (this is still on schedule even with the recent upgrade hiccups). Of the 60 gigabytes/month that will be created, only 13 will initially be posted to the OSO server. Further evaluation will determine how much additional data will get posted in subsequent months. The communications line from OSO to Bowie has been upgraded from 10 megabits/second to 155 megabits/second.
2. Notes from EMC
a. Global Modeling Group: Hua-Lu reported that the paradigm of a collapsed AVN/MRF model to a single global model run 4x/day was discussed with the NCEP Director. A formal plan will now be drafted to implement this Global Modeling System. There is discussion of producing hourly output from the Global Modeling System on constant height levels out to 24-36 hours geared toward AWC in addition to the normal product suite. The Global Modeling Branch and Climate Modeling Branches will merge into the Global Weather and Climate Modeling Branch (GCWMB). Climate Modeling efforts are not part of the GCWMB.
b. Mesoscale Modeling Group: Geoff Dimego reiterated the e-mail announcement on the Eta 12km model slated to become operational on November 27. Parallel output is available on the Mesoscale Modeling Group web site (http://www.emc.ncep.noaa.gov/mmb/mmbpll/etapll12/). Code has been delivered to NCO for parallel production on the SP. Other than resolution increase, the only other difference to the current parallel is the inclusion of NOAA-16 radiances. Two additional changes slated for testing and implementation are the inclusion of radial velocities from the WSR 88Ds and a change to the grid scale precipitation algorithm. This algorithm will include distinction of water species and therefore fall velocities (including hydrometeors). This implies the species will be subject to horizontal and vertical advection (as opposed to the current scheme which was optimized for 80km resolution, advects only cloud water, and unloads all the precip over one grid point in one time step). There is the outside chance that the 3D-VAR will be modified to include anisotropic error covariances for moisture analysis. This would allow moisture data to be spread more "intelligently" - along isentropic surfaces, rather than in the circular pattern now used. EMC still awaits FSL to deliver frozen RUC 20 codes to test and optimize on the production side of the SP. Targeted implementation for the RUC 20 is Jan. 22, 2001. Frozen codes are not yet available as FSL is optimizing both general efficiency of code for use on SP and also the use of a migrating bird algorithm to quality-control the WSR 88D radial velocities.
c. Global ensembles : Zoltan Toth reported that Dick Wobus will be running a parallel version of the Global Ensembles once per day with 10 members and at T126 resolution out to 7.5 days. This data will be available soon now that the upgrades to the SP will be complete. A letter from Ralph Petersen was disseminated to all NCEP Center Directors asking to identify a representative to help formulate requirements for the global ensemble products. This representative will also help identify which products from the ECMWF ensemble and high resolution control would be most useful (now that NCEP has permission to use these fields in operational practice). No NCEP Center Director has responded as of yet. Canadian ensembles are also available and a JIF memo will be prepared to post process this data set like the ECMWF ensembles. Eventually the ECMWF, Canadian, and NCEP global ensembles will be combined to provide NCEP Centers with a multi-center ensemble prediction system product suite. This will be available in addition to the normal NCEP global ensemble prediction system product suite. Zoltan is working with Yuejian Zhu on a PQSF (probabilistic quantitative snowfall forecast) product to help support the HPC Winter Weather Experiment. Also Tim Marchok, with advice from Hua-Lu Pan and Zoltan are working to modify Tim's tropical cyclone tracking algorithm with a NASA algorithm. The goal is to produce a algorithm that not only tracks tropical cyclones, but also tracks developing and existing extra-tropical systems. This would support MPC and HPC operations at the least.
d. Short range ensembles (SREF): Steve Tracton reported that the SREFs are stable and Jun Du has incorporated the Baldwin-Schichtel precip type algorithm (currently used in the operational AVN and Eta runs) for both sets of members in the SREF. There is a small bug with this algorithm in the RSM members of the SREF, but this should be fixed and result in a JIF to make this operational. Verification stats already compiled will be used to optimize the scaling factor for the breeding process in the SREFs. No word has come yet from FSL, NSSL, or CAPS regarding incorporating RUC, Eta - Kain/Fritch, and ARPS members respectively into the current SREF configuration.
3. Input to EMC from Operational Centers - HPC Dave Reynolds inquired out on the bull's eyes in the Eta QPF. Geoff Dimego responded that this is a result of the current precip algorithm being utilized at a resolution for which it was not intended. The current algorithm is designed for 80km and has essentially not been changed since the subsequent resolution reductions to 48-, 32-, and 22-km, HOWEVER, a new algorithm is slated to be implemented in the Eta-12km bundle (see 2b above). MPC inquired about the apparent superior initialization of existing strong tropical systems in the AVN/MRF as compared to the Eta. EMC will investigate and feels that it is possible that the Eta algorithm to "bogus in" storms may purposefully still be turned off. This was done since a better forecast actually resulted in prior tropical seasons if this algorithm was not activated. Reruns will be conducted to verify if this is still the case.
4. Next Meeting Proposed Monday October 29th at noon in room 209.