Storm-Scale Ensemble Data Assimilation and Forecast Development for the High-Resolution Rapid Refresh (HRRR) and Future Applications in the Unified Forecast System

The next and final update to the deterministic Rapid Refresh, version 5 (RAPv5), and High-Resolution Rapid Refresh, version 4 (HRRRv4), is currently scheduled for an operational implementation at NOAA/NCEP in mid-2020. Numerous physics, dynamics and data assimilation changes are being included as part of this upgrade.  This presentation will discuss the scope of this implementation including  an emphasis on the development of an hourly-cycled regional 36-member storm-scale ensemble analysis system with demonstrated improvements to deterministic forecasts at the convective scale.  The design of this HRRR data assimilation system (HRRRDAS) will be motivated through choices to include multiple-scales of perturbations from both a global and this regional convection-allowing ensemble, scale-dependent use of different observation types in an Ensemble Kalman Filter and use of inflation to maintain ensemble spread.  While not included in the HRRRv4 operational implementation, an experimental storm-scale ensemble forecast system leveraging a subset of the 36 members will also be described.  Design of the HRRR ensemble (HRRRE) forecasts will be discussed including use of stochastic parameter perturbations across the single model physics suite along with comparisons to other storm-scale multi-model/physics ensemble designs highlighting both the benefits of this single-model/physics ensemble approach along with challenges in maintaining appropriate spread at longer forecast lengths.  This final configuration of the RAP/HRRR model systems will serve as an operational baseline for the transition to a regional FV3-based convection allowing application in a Unified Forecast System (UFS), known as the Rapid Refresh Forecast System (RRFS), and this transition will be described.