Initializing a Hurricane Vortex with an Ensemble Kalman Filter

Short-term hurricane track and intensity forecast can be largely affected by the errors in the initial vortex. Initializing a realistic hurricane vortex in a mesoscale forecast model still remains a challenge. A new ensemble Kalman filter (EnKF) vortex initialization scheme is proposed.

With this new scheme, hurricane positions and other characteristics which are continuously provided by the satellite and radar imagery can be easily assimilated in the ensemble forecast system. Numerical experiments using a simple 2D model and the Weather and Research Forecast (WRF) model demonstrate that assimilating hurricane position observations can prevent the ensemble track divergence. The subsequent ensemble or deterministic forecasts initialized from the EnKF analysis are improved.