The Study On Removing Internal Multiples With CFP Layer Algorithm And Curvelet Transform

Multiples has been the difficulty in the field of seismic exploration.With theincrease of the seismic exploration difficulty,especially in-depth of the lithologyexploration and reservoir characterization,the requirement of seismic imagingaccuracy are increasing too.As a coherent noise,the multiple seriously degrade thesignal to noise ratio of seismic data,interferes the identification of the effectivesignal,which cause the false geological phenomena of the seismic profile,affectingthe processing and interpretation of seismic data.Therefore,the study of multiplesuppression has the practical significance and application value.At present,most methods sudy on attenuation of multiple are for thesurface-related multiple,but the study on internal multiple is relatively small.But inactual seismic exploration,especially land exploration,because the instability offormation shallow structure.so the surface-related multiple not obvious like seadata.So that,for the land exploration,the internal multiple becomes particularlyimportant.We know,relative to the surface-related multiple,there is no more obviouscharacteristics of internal multiple.It is difficult to determine the positions wheregenerate layer multiple on the seismic profiles,so it is difficult to identify andremove.Furthermore,the dispersion of inernal multiple is very serious,which lead toour observation of seismic data has become very blurred,especially in the depth partof profile.In this paper,on the base of SRME method,we will study on internal multipleselimination based on wave-equation.Study on inernal multiples using the CFPtechnology emerged in recent years.Take advantage full-wave filed reconstructiondatum of the shot and receive downward continuation to interface where generatesinternal multiple,so that the interface underground is equivalent to the freeesurface,reuse SRME method to predicted multiple and match to subtract. Use Curvelet match instead of least-squares matching,use the Curveletthreshold value method instead of the least-squares minus.In the case of the selectionof least-squares matching,when matching parameters selected improper,the matchingeffect between the mathed multiple and the multiple in original data not very well,oreven no effect,and the process of amplitude matching will be bring the boundaryeffect and waveform distortion,weakening the effect of multiple suppression.Butafter amplitude matching in Curvelet domain,the matching effect id better thanairspace domain,this related to sparsity and directional characteristics of Curvelettransform.Curvelet threshold minus method can adapt to the matching error,it canminish the impact of waveform distortion and boundary effect,make the sifnal eventmore continuous.We know,the conventional subtraction method just uses the sampleas operational objective,subtract directly,this id bound to make the result not smoothenough.But the Curvelet threshold method goes in the Curvelet domain,given theCurvelet’s sparsity,a few Curvelet can describe the event sufficiently in a seismicprofile,illustrating that the Curvelet matches the seismic event very well.On the basisof previous work,we mainly finished the following work:(1)By the consult and collection of data,analyzing the generation types ofmultiple and feature recognition,and compare the advantage or disadvantage againstthe basic principles of common multiple suppression.Classify the multiple based onthe reflection interface of generate multiple,can be divided into two categories:freesurface multiple and internal multiple.The method of removal internal multiple canbe dividied into two categories too:one is de-noising method separate primary andmultiple which based on their characteristic differences in the space;the other isde-noising method which based on multiples cyclical and predictable,which uses theintrinsic link between the primaries and multiples.These technical are pre-stackdomain in principle;(2)Detailed introduce the basic principles of CFP technology to suppressinternal multiple.On the base of WRW model,by using of the model and thesingle-component data matrix notation,then derived the generation of multiple andformulas of the boundary algorithm and the layer algorithm which based on CFPtechnology to suppress the internal multiple,including model-driven and data-drivenin two forms.Based on the theoretical formula for data-driven forms,and furtherstudy of the implementation process of the algorithm,incluing the construction of CFP gathers and grid-point gathers;(3)Combination the Curvelet technology which developed rapidly to match andsubtract with multiple in recent years.The Curvelet is anisotropic wavelet developedin recent years,can decomposite optimizly of the seismic record.It’s application tomultiple matching and threshold subtraction is superior to the conventional Wienerfilter matching and adaptive subtraction;(4)The effects test of theoretical model and actual data.To further prove thecorrectness and validity of the proposed method,use the CFP layer technologymentioned in this article to the aim of internal elimination about the theoretical dataand the actual data,achieved better effect,demonstrate that the method has the certainfeasibility and adaptability,have a certain effect on the actual production,has goodprospects for development;(5)Summarize above work.