Towards Improving Hurricane Vortex Initialization Through Assimilation of Radar and All-sky Satellite Observations

Accurate representation of hurricane vortex is essential for numerical prediction of hurricane track and intensity. Due to insufficient in-situ hurricane inner-core observations and difficulties to assimilate satellite observations over the cloudy and precipitation regions, vortex initialization remains a challenging problem in operational hurricane prediction even with the advanced numerical modeling and data assimilation systems. This presentation aims at addressing some of recent problems with hurricane vortex initialization and also discussing the promises and challenges in assimilating radar and satellite observations for the vortex initialization.

First, the initial vortex spin-down problem in current NCEP operational HWRF forecasts is examined with the GSI-based hybrid ensemble-3DVAR data assimilation system. Results from several methods that are developed to reduce the initial vortex spin-down are summarized. These methods include the use of high-resolution ensemble background error covariance at HWRF native domain resolutions in the GSI-based ensemble-3DVAR hybrid system, assimilation of TC Vital data, and applying for digital filter. Then, a new method for vortex initialization is proposed and tested with direct assimilation of synthetic observations with tail Doppler radar (TDR) radial velocities. In recognition of difficulties to completely overcome vortex spin-up and spin-down problems, the final part of this talk discusses the possibility to improve the vortex initialization through assimilation of radar and all-sky satellite observations at hurricane inner-core. Preliminary data assimilation efforts with assimilation of radar reflectivity from TDR and Level 1C satellite radiances from the NASA Global Precipitation Measurements (GPM) Mission are presented. Challenges and future developments are discussed.