Implementation and evaluation of the Multigrid Beta Filter (MGBF) into JEDI

The Multigrid Beta filter (MGBF), a new technique for synthesis of covariances, is being implemented in the System-Agnostic Background Error Representation (SABER)/Joint Effort for Data Assimilation(JEDI) for the variational data assimilation over regional domains. MGBF in SABER/JEDI has been tested and evaluated in the variational analysis for the regional versions of the FV3 and MPAS models. While the implementation of MGBF in JEDI is still under active development, this paper describes the implementation strategies to integrate the MGBF component into SABER/JEDI, which is supposed to perform as a generic background error modeling system. It discussed refactorization of the MGBF package following Object Oriental Programming principles. The evaluation of JEDI variational analysis using a MGBF compared with other SABER components will be reported,using a batch of MGBF tests of various setups including: calibrated MGBF coefficients and its potential automation, choices of the wind related control variables, optimized selections of the number of MGBF grid generations and so on. A 3dHybrid EnVar using MGBF is being tested in the regional data assimilation system of Rapid Refresh Forecast System (RRFS) under development. Finally, further developments of MGBF in SABER/JEDI are discussed.