Based on WRF model and the Hybrid ensemble3Dvar data assimilation system, this study compares radar radial velocity (Vr) assimilation with TTREC (TyphoonTracking radar echoes by correlation) retrieved wind data assimilation and their impacts on the analyses and forecasts of tropical cyclone (TC) “Chanthu (2010)”. In addition to the cycling strategy of Vr assimilation and TTREC wind data assimilation, a combined radar data assimilation strategy is proposed, which assimilates TTREC winds at the first cycle and assimilates Vr data during the rest cycles. Concerning the respective advantages of two types of radar wind, the new cycling strategy aims at exploring the suitable radar data assimilation strategy for the forecasting of landfalling TCs. Comparing three different cycling strategies, the cycled assimilation of Vr data requires more analysis cycles, which leads to delayed starting time for the deterministic forecast. By contrast, the assimilation of TTREC data produces better analyses and intensity forecasts than the assimilation of Vr data during the earlier assimilation, due to the larger data coverage and more complete wind fields for TC circulation. However, as the TTREC winds contain retrieved errors, the increase of analysis cycles leads to degraded TC forecast. When the two types of radar wind data are jointly used, the combined radar data assimilation strategy provides favorable forecast results no matter how long the analysis cycles are.