SECOND
ENSEMBLE USER WORKSHOP
May 18-20, 2004, NCEP (room 707), Camp Springs, MD
Return
to EMC ensemble home page
The workshop is to provide a forum for user discussions about the following aspects of ensemble forecasting: configuration, data, post processing, products and training.
Mark Antolik | OST |
David Bright | SPC/SSB |
Keith Brill | HPC |
Paul Dallavalle | OST |
Brent Gordon | NCO |
Michael Graf | OCWWS/ASB |
Mike Halpert | CPC |
Kathy Hughes | OST |
Paul Janish | Entergy-Koch Trading (Houston) |
Rick Knabb | TPC |
Arun Kumar | CPC |
Suzanne Lenihan | OST/ Science Plans Branch |
Andrew Loughe | FSL/AD |
Peter Manousos | HPC |
Daniel Melendez | NOAA/NWS |
Dave Michaud | NCO |
Jon Mittelstadt | WFO/Pendleton |
Fred Mosher | AWC |
Linnae Neyman | NWS/Alaska Region |
Minh Nguy | Delinia Corp (Houston) |
Ed O'Lenic | CPC |
Dave Plummer | NCO |
Pedro Restrepo | OHD |
Glenn Rolph | NOAA/ARL |
Steve Schotz | NCO |
Joseph Sienkiewicz | OPC |
Barbara Stunder | NOAA/ARL |
Steve Tracton | ONR |
Dave Unger | CPC |
EMC PARTICIPANTS
Hsuan | Chen |
Jun | Du |
Lacey | Holland |
Ding-Chen | Hou |
Geoff | Manikin |
Jeff | McQueen |
Zoltan | Toth |
Mozheng | Wei |
Dick | Wobus |
BinBin | Zhou |
Yuejian | Zhu |
1) Ensemble Configuration: |
Resolution |
Number of members: are many needed or can we statistically interpolate/extrapolate? |
Multi-model approach |
Stochastic model pertubations |
Intial pertubations |
Multi-center ensemble approach |
2) Ensemble Data: (current & future storage/bandwidth) |
Low-resolution ensembles: is there a unique problem for ensembles? |
Trade-offs between resolution (& only selected variables) and number of members |
Who needs summary products; who needs output from all members? |
Could NWP have succeeded with strictly limited data access (only a few fields)? |
Balance between accepting constraints and striving for optimal future capability |
* Add low resolution anomaly forecasts to high-res climate? |
* Formats: Pgrib, Thinned Pgrib, Grib2, Different Map Projections, Gempak |
* Users: NCEP Service Centers, AWIPS, ftp sites (public & commercial users) |
3) Post Processing: |
Bias types: |
* Expected value (first moment) |
* Spread (second moment) |
* Temporal variations in spread |
Method types: |
* Regression |
* Kalman filter |
* Neural net |
* Baysean |
Sample size: |
* Function of (bias) signal vs. (random initial error) noise ratio |
** Adaptive, short training period for short lead times, longer, climate type sample for longer lead times? |
Output: |
* Model variables/grid |
* Downscaled/new variables (NDFD and/or sites) |
Where to apply: centrally or locally? |
4) Products & Training: |
Prepared products vs. Interactive access to ensemble data: who needs what? |
* NCEP Service Centers, WFOs, Public Servers (ftp, web)? |
Which prepared products most often used? |
Interactive tool set: |
* What’s needed? Use AWIPS recommendation as starting point |
* Who develops it? |
* On what platform (AWIPS, NAWIPS, web, other)? |
* Can same set of applications be used on different platforms? |
Data access vs. training: |
* Will limits to data access also limit spread of new ensemble methods? |
Formal training: |
* COMET |
* NCEP Service Centers |
* EMC Predictability meeting: should it be a telecon? |
* Public - media |
The meeting room will be equipped with one overhead projector and a computer supporting electronic presentations (Web, Microsoft, etc).
DC Metro Area | Airports & Hotels |
Maps for Metro-system Users: | National Airport Metro Station |
Old Town Alexandria & Metro | |
Old Town Alexandria Hotels | |
Hotel websites | Old Town Alexandria Area |
Colony South Hotel |