NCEP Modeling Synergy Meeting
Highlights: June 4, 2007
This meeting was led by Bill Bua and attended by Mark
Iredell, Mary Hart, Joey Carr, Eric Rogers, Mike Brennan, and Keith Brill. UCAR/COMET’s Stephen Jascourt and SPC’s David Bright attended by
remote conference.
1. CCS
Joey Carr briefed the group on the status of the NCO implementation list for the remainder of FY07. Details can be found at this website. Hurricane WRF (HWRF) is now scheduled to be implemented June 19, and a probabilistic Storm Surge model implementation is scheduled for June 12. Real-Time Mesoscale Analysis for the CONUS is scheduled for 26 June, while AK’s version is scheduled for 18 September. PR/HI/Guam domains for RTMA are scheduled for FY08. The Air Quality model 5-km (for the 5-km NDFD grid) domain increase to cover CONUS is scheduled for 18 September.
An upgrade of the so-called Hi-Res window runs has been delayed to the 4th quarter of FY07; it was originally scheduled for 26 June. This includes upgrades to the current version of the WRF-NMM and WRF-ARW, an increase in horizontal resolution from 5.4 to 4km in the NMM and 5.8 to 5.1km in the ARW, and the change from the three CONUS domains run once per day to two larger east and west CONUS domains run twice per day.
The Real-Time Ocean Forecast System (RTOFS) was upgraded on 5 June 2007. COFS (which is replaced by RTOFS) will be continued running for now, but will be discontinued in September. A new WaveWatch3 model will be implemented on 10 July 2007 for the Alaskan, Western North Atlantic, and Eastern North Pacific domains (but not the global domain yet). The Global Ocean Data Assimilation System (GODAS) upgrade, which includes deep water data and will run with a 1-day lag instead of an 8-day lag, will slip to Q1 of FY08, and will be merged with the Climate Data Assimilation System (CDAS) upgrade.
WRF-NMM
will have an emergency implementation in the next few weeks to deal with a bias
problem with dewpoints in Western Region. The regularly scheduled WRF-NMM implementation
(including a new balance constraint and new data types for the data
assimilation system, a possible
increase in domain size, and new guidance grids via smart initialization) is
now planned for 11 September 2007. Downscaled NDFD grids based on the
The GFS unified postprocessor is scheduled for 25 September, and code has been delivered to NCO. Code needs to be sped up a bit to run multiple processes at a time as with the WRF-NMM post. Evaluation of the new post will start 9/3, though those with access to “/com” can look at the GRIB data informally before that.
Ensembles: The SREF upgrade will slip to the 1st quarter of FY08 (including update of codes to most recent model versions, bias correction, increasing resolution from 40-32km for WRF-NMM, and 45-32km or equivalent for WRF-ARW and RSM. NAEFS new products (depending on GFS unified post upgrade) are now scheduled to be put in operations for 25 September 2007.
Formal evaluations of model parallel tests will now be required to specifically
discuss whether the forecast changes anticipated under the model change
justification actually were seen.
Computer
problems with Dew: Disk problems have
been occurring sporadically with long duration between events. Mist is also at
risk but fortunately has not experienced this problem yet. IBM and the disk
manufacturer are working on finding the cause and a solution. Question was
asked: Can files get lost? If writing
while the failure is occurring, yes; for already existing and stored files,
no. Next switch from mist to dew for ops
to upgrade mist is now scheduled for 18 June 2007, unless more problems
manifest on dew.
2. NOTES
FROM EMC
2a. Global Modeling Branch:
An implementation was done on 29 May 2007
dealing with data assimilation ingest (new satellite data, GOES 1x1
field-of-view products, METOP Eurosat for first time, and more stringent
quality control of satellite winds and other data). The Branch is now doing major model physics
testing, which got backed up because of the importance of getting a clear
signal of any change from the new GSI implementation. There is now running a whole set of winter
and summer tests to decide what to put into parallel for the next major
implementation. It is expected that a
major implementation will go in during Q2FY08.
2b.
Mesoscale Modeling Branch:
Nothing beyond what was covered in CCS.
2c. Global Ensemble Prediction System:
Canadians are updating their part of the
NAEFS to have same number of members per cycle as the NCEP MREF; the data is
not yet available for evaluation. Note was made that the Canadian ensemble will
continue to be run only 2x per day. Bias
correction of operational GFS as well as ensemble members is being done. There are experiments being done, for the
first two days that blend the ensemble forecast and the operational GFS to take
advantage of its skill (bias correction method).
2d.
Nothing further to report
2e. Marine Modeling and Analysis Branch
(MMAB):
No representative present.
3. Feedback from operational centers
None.
4. The next meeting will
be held Monday, 30 July 2007 at noon in EMC Rm. 209, with remote conference
capability.