Using the technique of smooth filtering and cylindrical filtering, the initial vortex circulation and large-scale environmental field were separated from the background field. Then the separated initial vortex circulation was translated and reinserted in the location where it was observed. This led to the determination of a method of initial vortex relocation. For seven tropical cyclones at 23 points of measurement time in the years of 2006 and 2007, two schemes, either directly adding a tropical cyclone bogus model in the background or adding it after the relocation of the initial vortex in the background field, were employed. Simulation experiments were compared. The results showed that the mean errors of the simulated tropical cyclone tracks at 24 and 48 hours were both smaller with the scheme of adding tropical cyclone bogus model after the relocation of the initial vortex in the background field. The relocation method of the initial vortex decreases the error caused by the deviation of the initial tropical cyclone location in tropical cyclone models. The relocation method is conducive to improving the track forecast of tropical cyclone models and has a good perspective for operational application.