Hybridization of static and ensemble B in JEDI EnVar


I would like to know how the ensemble covariances are blended into the variational framework in JEDI EnVar application. Is it achieved by explicitly replacing the static B with a weighted sum of static and ensemble B (Hamill and Snyder 2000), or through augmented control variables (Lorenc 2003) as in GSI?


Hi @hbo9955, sorry for this late reply. The EnVar in JEDI-OOPS uses Buehner (2005) implementation, which Wang et al. (2007) showed to be equivalent to Lorenc (2003).