An Integrated Ensemble/Variational Hybrid Data Assimilation System

Ensemble/variational hybrid data assimilation systems are currently under a lot of scrutiny. They are very attractive because they show the potential to leverage the robustness and efficiency of variational systems with the flow dependency and uncertainty estimation from Ensemble Kalman Filters (EnKFs). The drawback is that it requires to develop, interface and maintain two separate data assimilation systems. We propose a new approach for updating the ensemble perturbations within the variational system and without the need for an extra EnKF system. We will explain the implementation and show results for the WRF model in full ensemble data assimilation mode. The evolution of the analysis and the ensemble spread will be studied and compared to a state-of-the-art EnKF system. This new approach is much simpler to implement and can be used in hybrid mode.