Assimilation of GBVTD-derived Winds from Single-Doppler Radar for the Short-term Forecasting of Super Typhoon Saomai (0608) at Landfall

In recent years, a single-Doppler wind retrieval method, called the Ground Based Velocity-Track Display technique (GBVTD), has been successfully applied to retrieve two-dimensional primary circulations (including the axisymmetric tangential wind and radial wind, wavenumbers 1 through 3 of tangential wind, and the along-beam mean wind) of landfalling typhoons/hurricanes at different altitudes. This study explores the impact of assimilating GBVTD-retrieved winds on the cloud-resolving analysis and forecasting of Super Typhoon Saomai (2006), the strongest typhoon ever recorded over the offshore region of China, and that made landfall over Mainland China. The GBVTD-derived winds from the coastal operational radar at Wenzhou, China were assimilated over a 2-hour period before Saomai landfall, using the ARPS 3DVAR/cloud analysis through 30-min assimilation cycles. Twelve-hour predictions were made. The experiments demonstrate that the assimilation of GBVTD-derived winds resulted in better structure, intensity and precipitation forecasts than that from JMA (Japan Meteorological Agency) mesoscale reanalysis or that with direct assimilation of redial velocity (Vr) data. The improvements to the track and intensity prediction both last about 12 hours. Additional experiments to evaluate the impacts of GBVTD retrieval error specification and axisymmetric/asymmetric wind components on the forecast revealed that the vortex center estimation uncertainty, the main source of uncertainty in the GBVTD technique, has slight impact on the structure and intensity forecast. The assimilation of axis-symmetric wind components is primarily responsible for the improvement, while the asymmetric wind component has small impact. The retrieved along-beam mean winds are found to improve the track forecast in the first few hours. In addition, the incorporation of reflectivity data through cloud analysis can bring about additional improvement to intensity and precipitation forecast, but degrades the track forecast slightly. The ability of the GBVTD method in retrieving closed vortex circulations is the primary reason for it to outperform direct assimilation of Vr data, which tend to have incomplete azimuthal coverage. Our conclusions with the Saomai case are supported by subjective evaluation and quantitative precipitation verification. Similar experiments for the hurricane Ike (2008) case are underway, and will be presented at the conference.