Abstract: The seismic data usually is non-linear and non-stationary series, the time-frequency analysis technique has the capability to reflect the partial feature of signal in time domain and frequency domain as well. This paper studies four kinds of time-frequency analysis methods, including short time Fourier transform, Wavelet transform, also <i<S</i< transform and Wigner-Ville distribution. Theoretically writer used Ricker wavelet matching to simulate seismic data, in order to contrast with the real seismic records. On base of the results, it can be concluded that the generalized <i<S</i< transform presents better time-frequency localization precision and cross-terms suppressing. In order to better verify this theory, writer conducts several experiments in two different field, then analysis and elaborate both the advantages and disadvantages for these four time-frequency methods, in the meantime, the results also verified the superiority of generalized <i<S</i< transform.