Multiple Elimination Research Based On Deconvolution In τ-p Domain

In seismic exploration, especially in marine oil and gas exploration, multiples reduce seismic data's S/N and disturb the recognition of primary. It will be more difficult to do velocity analysis and pre-stack and post-stack migration. Even more, multiples will affect the authenticity and reliability of imaging, result in fake seismic imaging and seriously influence seismic interpretation.Multiples can be separated into long and short period, also surface related and internal multiples. Because of complicated generation of multiples, it's difficult to separate simply with primary and eliminate in seismic data processing.This article analyzes the characters of multiples systemically, and discusses detailedly about the generation, classification and recognition of multiples. At the same time, filter methods and wave-equation methods of multiples attenuation are introduced here. Both kinds of methods have their conditions, disadvantages and advantages. So different kinds of multiples must be applied adaptive methods to eliminate. Then the article discusses forward modeling based on acoustic wave higher difference, then analyzes numerical dispersion including time and spatial dispersion and presents its stability conditions. Both primary and multiple inτ?pdomain are more periodically than that in t ? xdomain. Meanwhile, there is a better result after applied pre-deconvolution inτ?pdomain. Consequently, the Remul method which is different from traditional methods is introduced here. The Remul method comes from forward model of seismic data, and its elimination factor is deduced inω?pdomain (τis transformed toωby FFT inτ?pdomain). This operator bases on the reflection model of seismic data, and take both water layer multiples and water layer peg-legs into account. The consistent of operator and model data is more suitable to apply deconvolution and make multiples attenuation more effectively. At last, the complicated theoretical model and seismic data examples show that this operator works well, and some cognitions and advises about multiples attenuation have been presented.