Study On The Multiple Elimination In Seismic Data Processing

How to efficiently eliminate multiples in seismic data processing is a difficult problem which has not been well solved so far. In seismic data processing, the remnant muttiples considered to be primary wave or a part of the primary wave will reduce the ratio of signal to noise. At the present time, there are many methods to eliminate multiples, for example, predictive deconvolution, frequency-wavenumber filtering, the Radon transform, the deconvolution inτ-p domain etc. But these methods have their own advantages and disadvantages. If these methods are used inappropriately, they, at a certain extent, will inflect effective signal.In this paper, we firstly study the various methods of multiples attenuation. and then overview some popular methods of multiples attenuation in practice, including predictive deconvolution, frequency-wavenumber filtering and the Radon transform. In the end, we talk about beamforming in detail. Basing on analyzing the ray tracing method and wave equation method theory, we designed geological medels proceeded with ray tracing and finite difference method forward modeling. Through the model and data analyse we have studied three popular methods of multiples attenuation. we analyzed and compared their results, and summarized their advantages and disadvantages. (1) When the parameters were chosen properly, predictive deconvolution is good for short multiples, but not for long multiples.In fact, the good periods are kept only in zero-offsets traces. (2) The method of frequency-wavenumber filtering is simple, and not need much computing time. If effective signal are confounded with noise in frequency-wavenumber domain, the multiples elimination will be not well. It do a very poor job in attenuating mutltiples at near-offsets traces. (3) compared with the other mutltiples attenuating method, Radon filtering can attenuate all kinds of multiples very well at traces on both near and far offsets. (4) The example of data shows that the beamforming method can attenuate multiples much better at traces on both near and far offsets than Radon filtering, and enhance the relatively weak primaries. In the end, we spontaneously concludes that beamforming is practicable.