Abstract: S transform is a time-frequency transformation method used to analyze non-stationary signals, and it can depict the time-frequency characteristics of seismic signals very well. In this paper, S transform is used to remove noises in seismic surface wave data. The theoretical basis of S transform is introduced first, and then time-frequency filter and threshold filter are designed for processing ambient noise data and surface wave data of natural earthquakes, respectively. The results show that the short period part of group velocity dispersion curve of surface wave data is improved after S transform denoising, and the group velocity can be tracked continuously to about 6 s, but the long period part is missing. The effect of S transform is better than that ofphase match filter (PMF), and the combination of the two will get more ideal results.