P-43

A hybrid-4DVAR system for the JMA nonhydrostatic regional model

Le Duc (JAMSTEC), Takuya Kawabata (MRI), Kosuke Ito (U. Ryukyus), and Kazuo Saito (MRI)

A hybrid variational-ensemble data assimilation system has been developed for the Japan Meteorological Agency's (JMA) non-hydrostatic regional model NHM. The system consists of two components: the variational one 4DVAR and the ensemble one 4D-LETKF. The variational part was adopted from the NHM-4DVAR system developed at Meteorological Research Institute of Japan. The forward and adjoint models were derived from a simplified version of NHM. The ensemble part was based on the NHM-LETKF system developed at JMA. There is a two-way interaction between two sub-systems. NHM-4DVAR estimates the first moment of the posterior distribution using partial information of the second moment of the prior distribution provided by the ensemble part. NHM-LETKF approximates the second moment of the posterior distribution, though only in an error subspace, and then combines this with the first moment estimated by NHM-4DVAR to propagate the system state and its uncertainty in time. In addition the modeling of the climatological background error covariance in NHM-4DVAR has been revised to include the cross-correlations between different variables. These cross correlations have also been taken into account in the construction of localization matrices. Experiments with real observations will be shown.