f Time-frequency analysis - a new tool for waves separation

Non stationary signal processing methods are usualy applied to characterize dispersive signals. The authors are implied in studiyng surface-waves which energetic time frequency representation exhibits several patterns. In a first time, non related Fourier time-frequency methods (Sliding Lagunas method) is used to characterize different waves. In a second time, a filtering is applied to the time-frequency representation in order to extract the individual waves of the representation. The application of these methods to non stationary signals as seismologic or seismic signals allows us to value physical parameters related to the dispersive waves and to the propagation laws. The validation of this processing is shown from earthquake data study.