Abstract: The Model for Typhoon Track Prediction (MTTP) at National Meteorological Center (NMC), which has provided operational guidance for forecasters since June 1996, has always been imposed a scale-selective sophisticated filtering to remove the incorrect shallow vortex from the first guess fields and has implemented axisymmetric bogus vortex into the initial condition. However, with the upgrade of global spectral model system at NMC (from T106 to T213), the global forecast model now has a horizontal resolution that allows the representation of the tropical cyclone (TC) circulation, although the resolution is not adequate for resolving the inner core structure. Correspondingly, some valuable information about TC physical characteristics can be provided by the first guess fields from global gridded analyses. When damping out the entire shallow vortex of first guess fields, the background circulation around TC center has already been smoothed away by sophisticated filtering scheme, especially the asymmetric wind components, and it has adverse effect on the modified background field. Nowadays, the operational TC initialization scheme is constructed from a bogus vortex based on a few parameters analyzed by forecasters (such as TC central pressure, central position and radius of 15 m/s wind) and has an axisymmetric property. Because TC vortex has an obvious character of asymmetric structure, the implementation of axisymmetric property can contribute to partial track forecast biases more or less. In order to investigate the impact of background asymmetric circulation on the TC track forecast, an experimental scheme to introduce asymmetry into TC bogus vortex is carried out as follows:Extract asymmetric component of geopotential height from first guess fields by taking the difference between the total geopotential height and its azimuthal average about TC center; derive gradient wind from an azimuthal average of the total geopotential height; extract asymmetric component of wind by taking the difference between gradient wind and total wind of first guess fields; merge asymmetric components into axisymmetric bogus vortex and produce TC initial condition of MTTP model. It is expected that the scheme can provide a significant enhancement of initial TC asymmetric circulation and structure. Using the new scheme, a total of 12 forecast experiments for typhoon Sinlaku in 2002 are performed and the track forecast skills are evaluated in comparison with the operational ones. From the study it is found that much more accurate TC bogus data are produced by the new scheme compared with operational one. The verification of the experiments also shows that mean track forecast biases can be reduced effectively by introducing asymmetric components into axisymmetric bogus vortex.