Abstract: The accuracy of thunderstorm tracking vector is the key to determine the effect of precipitation nowcasting. Regional tracking represented by TREC (Tracking Radar Echoes by Correlation) and object-based tracking represented by TITAN (Thunderstorm Identification, Tracking, Analysis, and Nowcasting) are two typical methods for tracking thunderstorm motion vectors. TREC is based on tracking grid radar echo data, which can better reflect the overall movement trend of both stratiform and convective precipitation. TITAN can identify and analyze two or three dimensional attributes of thunderstorms, automatically track the moving speed and direction of a thunderstorm, and form centroid-tracking vectors that can better describe the moving speed and direction of small-scale thunderstorm. In this paper, vectors of TREC and TITAN are blended to generate a new extrapolated moving vectors, which can retain the advantages of the overall information of large-scale thunderstorm movement while give full play to TITAN's advantages in describing detailed information of small scale thunderstorm motion. The blending experiment shows that the new technology can obviously improve the accuracy of precipitation vector fields by improving the forecast accuracy of precipitation area and intensity, and has positive effects on improving the skill of precipitation nowcasting in Beijing area.