During recent years, great efforts have been made in the design and testing of a hurricane initialization scheme using the variational data assimilation (WRF-Var) approach developed for the Advanced Research WRF (ARW) model. The WRF hurricane initialization (WRF-HI) is intended to be computationally efficient and robust, yet powerful enough to define a hurricane structure in balance with WRF model initial conditions. In addition to the conventional observations, synthetic hurricane observations and remote-sensing data – from both satellite and radar, can be combined to create optimal hurricane initial conditions via the WRF-Var minimization procedure.

There are three main components to the WRF-HI system: 1.) vortex relocation within the initial guess fields, 2.) specification of bogus observations, and 3) assimilation of synthetic observations along with conventional and available remotely sensed data. The WRF-Var system (both 3/4D-Var and their hybrid with EnKF) can be utilized in order to define the optimal initial hurricane state.

In this presentation, we will show experimental results for Katrina (2005). Both 3D-Var and 4D-Var experiments are conducted for this case. The 3D-Var experiments are conducted twice daily for the lifecycle of Katrina (2005) beginning 12 UTC August 25 and ending 12 UTC August 30. The WRF-HI scheme is also tested for Katrina (2005), focusing on the multi-scale features of a small-scale eye, the mesoscale vortex, and the large scale environmental flow. The effectiveness of the vortex relocation and bogussing, and the impact of available satellite and Doppler radar data (obtained from coast and airborne flight) for the initialization of both the hurricane vortex and the large scale environment and their ultimate impacts on the model forecasts for track and intensity are evaluated in this study.