Abstract: A method of seismic data processing is described in which a particular wavelet is selected from a plurality of wavelets as being most characteristic of a received seismic signal. A subtracted signal can be determined by subtracting a weighted signal of the particular wavelet from the received signal. From the subtracted signal, an additional particular wavelet can be chosen. The process of subtracting a signal and determining an additional particular one of the plurality of wavelets can be repeated until a criterion is met. The method can be repeated at several depths. The resultant spectral analysis can be used to determine, for example, an absorption coefficient.