2.5 WRF-ARW research to operations update: the Rapid-Refresh (RAP) version 4, High-Resolution Rapid Refresh (HRRR) version 3 and convection-allowing ensemble prediction

Alexander, Curtis, Steve Weygandt, Stan Benjamin, David Dowell, National Oceanic and Atmospheric Administration (NOAA), Ming Hu, Tanya Smirnova, Joe Olson, Jaymes Kenyon, NOAA and Cooperative Institute for Research in Environmental Sciences (CIRES), Georg Grell, NOAA, Eric James, NOAA and CIRES, Haidao Lin, NOAA and Cooperative Institute for Research in the Atmosphere (CIRA), Terra Ladwig, NOAA and CIRES, John Brown, Trevor Alcott, NOAA, and Isidora Jankov, NOAA and CIRA

This presentation will contain a description of both the planned operational upgrades to Rapid Refresh (RAP) version 4 and High-Resolution Rapid Refresh (HRRR) version 3 in February 2018, and progress in development of HRRR-like convection-allowing ensemble analysis, forecast and post-processing. Forecast skill improvements related to specific enhancements in RAP/HRRR WRF-ARW numerics, physical parameterizations and associated data assimilation in the upgrade package with be presented including continued refinement of subgrid-scale cloud representation, adoption of the hybrid vertical model coordinate and more effective assimilation of new and existing observations for improved retention of clouds and a reduced convective bias, especially in the first six forecast hours.
The development of convection-allowing ensemble forecast capability using a HRRR-based WRF-ARW configuration will be highlighted with an emphasis on hourly-cycling 3-km ensemble Kalman filter data assimilation of radar, satellite and conventional observations. The capability has applications for both improved deterministic HRRR forecasts and HRRR ensemble (HRRRE) prediction using stochastic soil and boundary layer parameter perturbations, lateral boundary perturbations and inflation during data assimilation to promote spread and represent both initial condition and forecast uncertainties. An ensemble post-processing system is applied to produce all-season weather hazard probabilities using either time-lagged HRRR or the HRRRE as input.