EMC/HPC/MPC synergy meeting highlights 03/26/01

This meeting was led by Stephen Jascourt and followed a generic agenda denoted by the subtitles below. Attendees included Geoff DiMego, John Ward, Hua Lu Pan, Bill Bua, Keith Brill, Ed Danaher, Kevin McCarthy, Wes Junker, Dave Feit, Dave Reynolds, and Steve Lord.

1. IBM SP

The "B" side is still in "acceptance" mode and will become officially available on March 29 (already it is being used, including for the Eta parallel runs).

Gempak is available on both sides, so parallel model tests run on either side can be made available on NAWIPS. There was a problem with 24/7 maintenance of realtime datasets on the "B" side, requiring realtime parallels to pull data from the "A" side. This will be rectified so that the "B" side will be capable of serving as an operational backup facility in case "A" experiences a failure.

2. Notes from EMC

a. Mesoscale Modeling Group: Geoff DiMego reported on several topics:

i) Changes to be implemented 3/29/01: Eta 84 hour extension, fix of diagnostic 10-meter winds, and some code that outputs lateral boundary conditions for use by nested higher-resolution runs ("threats").

ii) Spring implementation: New 3D-VAR is in Eta-x parallel as of 3/23/01 and is being extended to 84 hours for comparison with the about-to-be-operational Eta extension, but will likely undergo at least one major change shortly. Land surface package changes will mostly affect the ground conditions and thus the diagnostic 2-meter temperatures, though patchy snow may have a larger effect (allowing more realistic warming above freezing in areas of snow cover). Precipitation assimilation is almost ready for parallel testing. The Stage IV data used to adjust model precipitation rates has a severe uncorrectable low bias with snow, so locations where the model precipitation type would be snow will not be adjusted.

The combined bundle will be put into a real-time parallel and August 2000 will be used for a warm-season retrospective run. Geoff is looking for test cases to rerun! Please let him know if you have any notable cases! These changes are estimated to be ready for implementation in around two months. The radiation problems will not be fixed for this implementation.Tests will also be done to ensure that the changes do not cause trouble at 12 km (for the regional and threats nest and the future 12-km implementation).

iii) The higher resolution nests should be running operationally and available in NAWIPS in June. These are run over regional domains once per day at the same time every day, in the GFDL hurricane slot. AWIPS bandwidth limitations precludes their going out over SBN, but the field can be pull grids off the server here. Additionally, Geoff envisions a "threats" run over a relocatable smaller domain to be selected by the NCEP centers for dealing with the problem of the day. This too would be available on the web and NAWIPS, but no need for field distribution since primary use would be for NCEP.

iv) Looking ahead, when the operational Eta goes to 12 km, the regional and threats nests will have even higher resolution and will become non-hydrostatic (model already has non-hydrostatic option). The new microphysics (cloud and grid-scale precip. parameterization) probably coming for an autumn 2001 implementation will have an option for more explicit microphysics categories (right now only cloud water or cloud ice are carried explicitly), so that for instance, falling snow could be advected. This option might be used for the nests but probably would not be used on the 12 km grid.

b. Global Modeling Group: Hua-Lu Pan reported that the MRFX (explicit clouds and convective momentum mixing - this is the version with greatly reduced spurious tropical cyclones) should be winding down its evaluation period, with reports due in from the NCEP centers by March 31, though the deadline will likely be extended for a week until April 7. It will go to CAFTI in May. TPC has expressed concern over poor track performance for weak storms, raising the unfortunate prospect that they may veto implementation. After that, the focus turns to a T254 parallel. Also there is interest in testing the new 2D-VAR SST and some improvements in the handling of initial and forecast snow, but computing resources to run tests on these are not available.

c. Short range ensembles (SREF).

John Ward and Geoff DiMego reported that SREF will have a limited core product suite in NAWIPS in May and a full product suite by late June for the real-time test. Discussions are occurring with Dave Helms for getting SREF products into AWIPS, though presently it is not clear when or if this will occur. The 21 UTC and 09 UTC SREF will be run during the hour prior to start of the 00 UTC and 12 UTC operational Eta, producing products available at the same time as the Eta products. After an extended period of real-time test and evaluation, SREF will go through CAFTI before it can be officially implemented.

3. Input to EMC from Operational Centers

Nothing was reported - any issues might have been covered by above discussions.

4. Next Meeting Proposed Monday April 30, 2001 at noon in room 209.