| Seismic processing seeks to extract sub-surface information from recorded seismic data. Since the recorded data contains waves that have experienced the Earth and other waves that have not, it is necessary to separate these two categories of waves. In addition, seismic waves that have experienced reflections in the Earth depend on both the Earth properties and the character of the wave that propagates from the source. The separation of the latter waves to extract the portion that only depends on the Earth properties (wavelet deconvolution) is a required step in many processing procedures.; Furthermore, inverse scattering sub-series methods for removing multiples and for depth imaging and inversion primaries without knowledge of the sub-surface place a high-bar on the efficiency on the completeness and definition of the seismic experiment. The latter requisite includes the estimation of the source signature.; A new approach to addressing that problem is presented here that combines, extends and alters two previous wave theoretical approaches to find a good approximation to the wavelet. The proposed alternation and resulting useful algorithm can be interpreted as a generalized inversion. The input to the algorithm is the pressure field on the cable. The procedure first predicts the normal derivative of the pressure along the cable, and then estimates the wavelet. Empirical tests indicate that the particular alteration produces an accurate and stable estimated wavelet, and demonstrate the added value of this approach for attenuating free surface multiples when using the wavelet. |
