Development of the Ensemble Kalman Filter in Version 2 of the Rapid Refresh Forecast System

As part of the goals of the Unified Forecast System (UFS), NOAA recently began developing the Rapid Refresh Forecast System (RRFS) as its next-generation, convection-allowing regional ensemble and DA forecast system. Ensemble data assimilation in RRFS is performed using a Gridpoint Statistical Interpolation (GSI)-based hourly-cycled ensemble Kalman filter (EnKF). For RRFSv2, NOAA plans to transition from GSI-based DA to the Joint Effort for Data Assimilation Integration (JEDI). The JEDI project aims to accelerate research and operational implementation through a unified framework for DA. This transition to JEDI for RRFSv2 is being jointly developed by the Environmental Modeling Center (EMC) and the Global System Laboratory (GSL).

The JEDI framework for ensemble data assimilation currently supports the Local Ensemble Transform Kalman Filter (LETKF) and its Gain formulation (GETKF), which uses modulated ensembles to emulate model-space vertical localization. This presentation will compare the performance of JEDI-based solvers for RRFSv2 against existing configurations of the GSI-based EnKF used for RRFSv1. We will discuss early progress, challenges encountered during the transition, and present ensemble analysis comparisons between GSI and JEDI.