Development and Evaluation of the Rapid Refresh Forecast System Ensemble Components

Continued development and evaluation of the Rapid Refresh Forecast System (RRFS) is underway to advance convective allowing model guidance in NOAA operations. RRFS version 1 is part of the NOAA numerical weather prediction (NWP) modeling evolution into the Finite-Volume Cubed-Sphere (FV3)-based Unified Forecasting System (UFS). The RRFS Data Assimilation System (RDAS) includes an ensemble data assimilation component, that consists of 30-member ensemble hourly data assimilation cycling. RDAS cycling includes ensemble cloud analysis (mainly cloud clearing), EnKF analysis of the conventional observations and the direct radar reflectivity data assimilation. RDAS also includes a deterministic control run that uses 3DEnVar assimilation and leverages the RDAS ensemble for background error covariances. The control is used to recenter the ensemble analysis, thus RDAS has 2-way coupling between the EnKF cycled ensemble members (being re-centered at the control analysis) and a control run (with hybrid GSI analysis using the RRFS ensemble 1-hour forecast).

RRFS is an ensemble forecast system. The ensemble free forecast is subset from the RDAS EnKF members and runs at select cycles daily with stochastic physics (SPP, SPPT). In addition to the real-time RRFS runs, multiple retrospective experiments are conducted to evaluate and tune the RDAS configuration, including both the deterministic and ensemble data assimilation. This presentation will share preliminary evaluation of the regional ensemble data assimilation system and further research and development of the system, including the multi-scale data assimilation and new observation types.