Assimilation of GBVTD-derived Winds from Single-Doppler Radar for Short-term Forecasting of Super Typhoon Saomai (0608) at Landfall
【摘要】：A single-Doppler wind retrieval method called the Ground-Based Velocity-Track Display technique (GBVTD) has been developed in recent years to retrieve horizontal circulations of tropical cyclones when they are within the range of coastal Doppler radars. The technique is able to retrieve axisymmetric tangential and radial winds, asymmetric tangential winds for wavenumbers 1 through 3, and the along-beam mean winds within the tropical cyclones. It has been successfully applied to several typhoon cases for the tropical cyclone monitoring and warning. This study explores, for the first time, the assimilation of GBVTD-retrieved winds into a tropical cyclone prediction model, and examines its impact relative to that of directly assimilated radial velocity data. Super Typhoon Saomai (2006), the most intense landfalling typhoon ever recorded in China, is chosen as the test case, and data from the coastal operational radar at Wenzhou, China are used. The ARPS 3DVAR/cloud analysis system is used to assimilate either the radial velocity data directly or the GBVTD-retrieved winds, at 30-min intervals for 2 hours. A control forecast is also made using an initial condition from the Japan Meteorological Agency mesoscale reanalysis. The assimilation of the GBVTD-retrieved winds results in much improved structure and intensity analyses of Saomai compared to those in the JMA (Japan Meteorological Agency) mesoscale reanalysis as well as that assimilating radial velocity (Vr) data directly. The ability of the GBVTD method in obtaining complete vortex circulations in the inner-core region is the primary reason for its superior performance over direct assimilation of Vr data; for the latter, the azimuthal data coverage is often incomplete. With the improvement in the initial conditions, the subsequent forecasts of typhoon intensity, track and precipitation are also improved. The improvements to both track and intensity predictions persist over a 12-hour forecast period, which is mostly after landfall. Subjective and quantitative evaluations of the precipitation and circulation patterns show consistent results. A further sensitivity experiment shows that the asymmetric wind component in the GBVTD retrieval has a small impact on the prediction.