380 likes | 392 Views
This update provides information on the high-resolution regional prediction system in the Northwest, including changes in domain, structural improvements, hardware enhancements, and plans for future improvements.
E N D
Update on the Northwest Regional Modeling System 2012 Cliff Mass and David Ovens University of Washington
NW High Resolution Regional Prediction • Supported by NW Modeling Consortium-a collection of local, state, and Federal Agencies AND the private sector (Iberdrola Energy, King-5) • Currently runs with 36, 12, 4, and 1.3 km grid spacing using the WRF (Weather Research and Forecasting) ARW model twice a day at 0000 and 1200 UTC (5 AM and 5 PM PDT) • One of the highest resolution numerical weather prediction efforts in the U.S.
High-Resolution WRF Forecasts • Forecast Periods: • 36-12-4 km out to 84h • 36-12 out to 180 h • 1.33 km out to 48h • Initialization and boundary conditions from the National Weather Service GFS model • Graphics available on the web and grids available to contributing consortium members
Still running the old MM5 model initialized with the NWS NAM model for a quick-look prediction twice a day 36-12 km only
12km 2-m temperature
In general, verifications look favorable for the 1.3 km—but there are real issues in verifying
Behind the Scenes: Hardware • Added more Intel Newhalem 8 and 12-cpu servers • Using 136 processors for the UW high-resolution runs. • Excellent parallelization—many processors working together. • Connected with high-bandwidth Infiniband networking (40 Gb per second). • Mirrored RAID disk arrays for dependability
But what about vertical resolution? • Do we need more vertical resolution—more vertical levels– in the new 4/3 km domain? • Testing is now going on…. • Will more vertical levels will help with a significant problem we see– an inability to maintain shallow cold air layers.
Adding ten more levels below 200 meters (12 UTC 20 January 38 levels 48 levels
48 38
48 38
The Long Term Future of the UW Modeling Effort • Based on a large (64 member) ensemble of forecasts at 36 and 4 km grid spacing. WRF model and DART Ensemble Kalman Filter (EnKF) System • Every three hours assimilate a wide range of observations to create 64 different analyses. • Then we forecast forward for 3 hours and then assimilate new observations. • Thus, we have a continuous cycle of probabilistic analyses.
EnKF Ensemble Forecasting System • We can run ensemble of forecasts forward to give us probabilistic forecasts for any period we want. Now doing 24h ahead, four times a day. • Planning to go to a 1-hr cycle and to use more observations (e.g., more surface pressure obs). • More next year.
KING TV PPM Energy