Calibration of Parameters of the Multigrid Beta Filter (MGBF) and its Performance within the Three-Dimensional Real-Time Mesoscale Analysis (3D RTMA) Project

The Multigrid Beta filter (MGBF) is a new technique developed at EMC for synthesis of covariances. It removes the scaling problems of its predecessor, the Recursive Filter, and offers a series of promising features, such as the treatment of cross-covariances within a variational approach, and inclusion of the temporal component of fully four-dimensional covariances. The MGBF is incorporated in the Rapid Refresh (RRFS) version of Grid-Point Statistical Interpolation (GSI), and for now is being applied to modeling of the background error covariance, though applications to localization of the ensemble and the characterization of analysis errors are also under development. This paper discusses different approaches that are considered for calibration of the MGBF technique with a focus on a lookup table method, based on inputs of correlation lengths derived from error statistics generated using the NMC method, and presents the first results derived by application of this method within the Three-Dimensional Real-Time Mesoscale Analysis (3D RTMA) project.