Mixed-phase seismic wavelet extraction of SIMO system based on vector prediction

Given the limitation of the mixed-phase wavelet extraction based on high order statistics, a new method of mixed-phase wavelet estimation of SIMO (Single Input Multiple Output) system based on vector prediction is presented. For the extraction of mixed-phase wavelet, this method uses the data from the adjacent two or more traces to construct the expression of inverse-wavelet by applying the vector prediction. In this process, the information of phase in the system is obtained by the second order cycle stationary statistics and CMP gather will be viewed as the output of SIMO system. Utilizing the extracted phase information, the CMP gather is processed by means of pure-phase filtering, which can substitute for the traditional phase correction processing. And the deconvolution is performed on the CMP gather with the deconvoing operator obtained from the above step, thus the quality of stack section can be improved by correcting the wavelets with respect to the different traces, making the amplitudes, frequencies and waveforms of these wavelets in different traces consistent. Finally, the method in this paper is appropriate for wavelet extraction and deconvolution with arbitrary phase. This is confirmed by model calculation and real data test. So this method has the potential of practical application.