NCEP Modeling Synergy Meeting Highlights: April 30, 2007


This meeting was led by Bill Bua and attended by Mark Iredell, John Ward, Geoff DiMego, Ken Campana, Ed Danaher, Daniel Pawlak (AF), Joe Sienkiewicz, Jim Hoke and Keith Brill. UCAR/COMET’s Stephen Jascourt attended by remote conference.


1. CCS


John Ward reported that NCEP has experienced smooth operations on the R&D, development, and operational machines. A new quarterly upgrades schedule has been implemented to make life easier on users.  The next upgrade is scheduled for late-May through June.


Briefing of NCEP director:  We’re now on a quarterly schedule of updates for all implementations currently in pipeline. The proposed GFS implementation was presented on 23 April and was approved for implementation tomorrow. Others implementations at various stages of readiness on the schedule were also discussed.  This new briefing schedule will help keep things on schedule.


NCO had its first ever review meeting April 18-19.  This effectively was the kickoff for the FY08 Annual Operating Plan development. It is expected to give NCO a better feel for user needs, and should help the Center understand the staffing resources that will be available during such things as the move to the new building. This year’s meeting was just for NCEP centers, but will be expanded to include AK and Pacific Regions. The NCO review will occur annually from now on.


Eighteen separate implementations are in the pipeline for NCO right now. Some discussion ensued on problems with notification, because of staff changes at centers requiring notifications of implementation. NCO will be getting out notifications earlier, and making sure staff changes are taken into account, including when test data will be available (should be 45 days prior to implementation). Meteorological Development Lab will be added to list of those notified.


It was noted that “outsiders” are also on the list (e.g. AccuWeather, WSI). One hundred fifty entities have subscribed overall, through “Family of Services”.



2a. Global Modeling Branch:


Mark Iredell reported on the new GFS implementation discussed above. There is some concern regarding sigma files and use of GFS for other models as boundary conditions. The Branch hopes that things will go smoothly, but not EVERYTHING that requires GFS with new files in real time has been run (Mark estimates that 50-80 downstream codes are affected by the implementation).


In the pipeline:  A unified post (for delivery to NCO in May).  The unified post will run in parallel for awhile eliminating snags.   The Canadian Meteorological Center (CMC) is increasing the number of members in their ensembles to 20, to match ours for the joint North American Ensemble Forecast System (NAEFS) with the intention of later producing combined products (mean, spread, probability). Presently, only separate products for each ensemble set are produced. An observational upgrade is scheduled to add codes for different observational platforms such as GOES 1x1 field of view data [didn’t say if imager or sounder or both], METOP data, etc .  On the table for the Climate Forecast System (CFS): add two new members to the CFS ensemble and move the data dump time up to one day lag from a seven day lag.


2b. Mesoscale Modeling Branch:


Geoff DiMego: Next implementation will be the Hi-Res window, expanding the domains to around 2/3 of CONUS for a twice daily east-central run at 00z and 12z and a west-central run at 06z. The Alaska run will be moved to 18z. The ARW member will run at 5.1 km grid spacing, while the NMM run will be at 4 km grid spacing. Using the faster indexing order that the NMM had before the WRF infrastructure was imposed improves NMM speed by around 10%, helping the resolution increase to fit within the computer resource allotment. Codes will be upgraded to the version 2.2 of WRF, and will be delivered this week.


RTMA: distributed on AWIPS but not in NAWIPS because interest was from the field offices, not the Centers. An upgrade long in progress is strongly desired by users. The RTMA is available hourly at 45 minutes past the analysis hour, though NCEP was aiming to deliver the product by 35 minutes past the hour. The delay is the result of slowness in receiving MesoWest data from MADIS, which could be improved by purchasing a new server for MADIS.  RTMA parameters available include analyses of temperature, winds, and moisture; and interpolation to the NDFD grid of the Stage II precipitation analysis and NESDIS-based Effective Cloud Amount. An estimation of analysis uncertainty for T, U, V, and Td is provided, which may be used in weighting errors in scoring verification of NDFD forecast grids. As more variables are added to NDFD, they will also be added to RTMA; for example, significant wave height is on the list. New locations to receive data will be AK this year, then HI, PR, Guam. The NCEP Director wanted to know whether any Air Force data is being used in the RTMA. No one was sure about this (checked with Dan P.). 


NAM Upgrade:  Expansion of domain to N, W, and E (18% increase in coverage all told). Coding has been made faster to make up for increase in domain size. The many small changes include new observation types (e.g. MODIS winds, AIRS data), model physics “tweaks”, increased divergence damping during the NAM data assimilation and the GSI new strong constraint for better balance fields. Some of these are inspired by moisture problems out west last summer right after implementation.


SREF upgrade:  will probably not be implemented until Q1 2008.  Changes will include bias correction of ensemble data (but probably not for precipitation), code improvements, resolution increase of non-Eta members to as close to Eta horizontal resolution as possible.


Fire Weather runs:  Need to go to OSIP and get funding for reinstatement of fire weather runs (proposed are small-domain 3-km nested NAM WRF-NMM runs right around a given fire domain).


Air Quality (AQ) upgrade (ozone and particulate matter output) to full CONUS. There already exists a national “smoke” product, based on the HYSPLIT dispersion model, which is run where there NESDIS data indicates fires exist.  This version of HYSPLIT can do multiple sources simultaneously. However, the fire data do not get into the CMAQ model used to generate the particulate matter forecasts. Thus, the particulate matter product and smoke product may paint different pictures.


2c. Global Ensemble Prediction System:   See above


2d. Short Range Ensemble Prediction System:  See above


2e. Marine Modeling and Analysis Branch (MMAB): 


No report at this time. 


3. Feedback from operational centers


Joe Sienkiewicz reported that the Ocean Prediction Center will be participating in the RTOFS evaluation for the upcoming implementation


4. The next meeting will be held Monday, 4 June 2007 at noon in EMC Rm. 209, with remote conference capability.