Development of a regional hybrid variational/ensemble Kalman filter data assimilation system for tropical cyclones

A hybrid variational/ensemble Kalman filter based data assimilation system has been developed for global NWP at NCEP, and has been shown to significantly improve the skill of tropical cyclone track forecasts. It is based on the Gridpoint Statistical Interpolation (GSI) 3DVar system, and allows for the use of flow-dependent background-error covariances, estimated from a ensemble Kalman filter (EnKF). The hybrid GSI/EnKF system is being extended to work with the regional HWRF modeling system. Here we describe the results of initial tests of an integrated global/regional GSI/EnKF system, in which the global GFS ensemble provides lateral boundary conditions for a fixed regional HWRF domain that extends from the eastern Pacific across the entire Atlantic. A run of the HWRF-based regional GSI/EnKF system using global ensemble perturbations to estimate the background-error covariance in the variational system will serve as a baseline to assess the impact of cycling a separate HWRF ensemble within the regional domain.