Abstract: Quality factor（Q）is an important parameter for hydrocarbon detection and reservoir characterization, which reflects the subsurface seismic wave attenuation. The accuracy of Q value estimation greatly depends on the extraction of the time?varying seismic wavelet spectrum and the Q value estimating method. S transform can be used to accurately extract the time?varying seismic wavelet spectrum, and the combination of statistical frequency attributes (CSFA) can be applied to estimate the Q value in order to get the stable Q value. By this method, the regional oscillation of centroid frequency curve from real data and Q?value instability can be well solved. Application of this technology to the eastern and western Junggar basin allows the target zones’ resolutions to be obviously improved, of which the bandwidth can be increased by 5~10 Hz