Research On Seismic Blind De-convolution Method Based On BSS

Currently, seismic exploration has been in the fine exploration stage, trying to seek the complex and cryptic reservoir, so not only need to obtain the strong reflectors which can reflect the information of major layers, but also need to get the weak reflectors which can reflect the information of cryptic reservoir. And usually, the seismic wavelet transmitted in earth is mixed-phase wavelet, and the reflectivity sequence of strata is not Gaussian white noise distribution. Traditional de-convolution methods do not meet these requirements, so a new de-convolution method is required to improve the accuracy of seismic de-convolution.In view of the facts that the wavelet excited on the surface is unknown, the earth filtering effects and the reflectivity sequence of strata is also unknown, therefore, in the circumstances of no priori assumptions, the process to acquire the seismic wavelet and reflection coefficient is basically a blind process, at this time, the de-convolution is named as seismic blind de-convolution. Seismic blind de-convolution uses the information of higher-order statistics on the seismic traces, and does not require(or weaken) the priori assumptions on the seismic wavelet and reflection coefficient, the seismic blind de-convolution can be used to extract the mixed-phase seismic wavelet and estimate the reflectivity sequence of strata.After a comprehensive summary of previous researchers'work, the research subject is deeply analyzed, and a blind de-convolution implement program is established by first extracting the seismic wavelet, and then using the bussgang blind de-convolution method to acquire the reflection coefficient. The higher-order statistics has the ability to contain the wavelet phase information and to suppress the interference from color noise, higher-order statistics of seismic records can be used to estimate the seismic wavelet by applying the signal reconstruction theory and method. Then the initial anti-wavelet operator which is needed by the blind de-convolution algorithm can be obtained. After the implementation of the algorithm, the data simulation experiments can be carried out, and then a comprehensive analysis of algorithm performance is needed according to the simulation results, so as to improve the algorithm. Finally the feasibility of the method can be further verified by processing the field seismic records.