Data Driven Multiples And Primaries Separation Methods Research

Seismic migration imaging and inversion are two particularly important treatments in seismic processing.Before carrying out these processes on the seismic data collected in the field,various noise removal process is required to improve the signal-to-noise ratio and resolution of the seismic data.In the traditional seismic processing,because multiples will affect the accuracy of seismic migration and inversion,multiples are usually removaled regarded as a kind of important noise by seismic data handler,in the process of seismic waves in the downward propagation,it will encounters the reflection interface,and then it will return upwards.When encounters interface again,it will continue to reflect downwards.If the interface is a free surface,the multiple is called surface multiple;if the interface is a layer interface,the multiple is called internal multiple,and a number of multiples removal technology are developed at the past years.With the development of seismic migration and inversion technology,it has been found that multiples carry a lot of information that is not available from primaries,which is used as useful information from seismic data in the reverse time migration and full waveform inversion,and has achieved a better processing results than only primaries are used.And whether the multiples are regard as a noise or a useful signal,it is always necessary that multiples are extracted from the seismic data,so the multiples and primaries separation methods research is very meaningful.In this paper,we firstly discuss the one-dimensional SRME method and the two-dimensional SRME method,and the algorithm was tested using the simulation data.We analyze the suppression effect of the SRME method.We have come to two conclusions:1?the travel time and amplitude of the multiples predicted by the SRME method are not completely accurate;2? the energy of the primary signal will loss in the adaptive subtraction process.Aiming at the first problem of SRME method,we analyze multiple compression effect by various influencing factors such as the number of iterations,the length of the prediction factor and the noise.The suppression effect of SRME method is summarized for various influencing factors.Aiming at the energy loss of the primaries in the prediction and subtraction process,the CL-SRME method based on inversion is used to predict the primaries directly.This method avoids the prediction process of multiples,which avoids the energy loss of the primaries.Finally,a new extended internal multiple removal method is proposed,combining the concept of virtual seismic events with the CL-SRME method.This novel method produces an inversion-based internal multiple removal process,in which the primaries are estimated directly via a waveform inversion,which not only overcomes the problems such as low efficiency and the difficulty of identifying each interface but also avoids the internal multiples subtraction process.