Research On Multi-source Wavefield Reconstruction And Marchenko Imaging

Currently,seismic interferometry and Marchenko imaging are frontier research hotspots.Seismic interferometry obtains new seismic records by interfering with the recorded seismic signals.Since the seismic interferometry can use passive source information,seismic interferometry techniques can be used to reconstruct records from passive noise signal and image the underground structure.Seismic interferometry reconstructs virtual source records at geophone locations.This condition greatly limits the application of seismic interferometry.The Marchenko method is based on the inverse scattering theory which is different from the conventional seismic interferometry.It realizes the reconstruction of a virtual source at the non-receiver locations without knowing the overlying structure of the underground target area.That is,the position of the virtual source can be anywhere in the subsurface medium,and this Green's function is a precise Green's function containing both primary and multiples.These excellent features of the Marchenko method are a great leap over the seismic interferometry that can only reconstruct the approximate Green's function at the geophone's position.Therefore,it is also known as beyond seismic interferometry.On the other hand,the reconstructed Green's function somewhere in the subsurface medium based on the Marchenko method can be decomposed into the upgoing Green's function and downgoing Green's function.Marchenko imaging is the new imaging method which is based on these two one-way Green's functions.Marchenko imaging requires only a macro-inaccurate velocity model as input during the reconstruction of the Green's function.Therefore,compared to the traditional imaging method that requires an accurate velocity model,it has a qualitative improvement.In addition,the artifacts in traditional imaging methods based on single-scattering theory are mainly due to their inability to distinguish and handle internal multiples,whereas Marchenko's imaging results are not affected by the internal multiples contained in the seismic data.Accurate imaging results can be obtained directly.Marchenko imaging can suppress the artifacts caused by the internal multiples without the need for precise models.Based on the two methods of seismic interferometry and Marchenko method,this paper studies the reconstruction of multi-source wavefield and Marchenko imaging.That is,active source Marchenko imaging,Marchenko imaging based on the combination of active source and passive source,and passive source Marchenko imaging.Firstly,this dissertation deeply studies the theory of seismic interferometry based on cross-correlation and multi-dimensional deconvolution,and proposes a low-frequency reconstruction method based on frequency advantage.Compared with the traditional low-frequency reconstruction methods,the method in this paper comes from the real passive source transmitted seismic information,so the low-frequency information is reliable.Based on the low-frequency reconstruction method based on the frequency advantage,the high-frequency end of the original active source data in the broadband seismic data has been completely preserved,the low-frequency end has been effectively reconstructed,and Marchenko imaging results do not include noise due to low frequency noise and false reflection interfaces.This paper presents the data-driven active and passive source joint Marchenko imaging theory.The combination of active and passive sources Marchenko imaging theory is a combination of seismic interferometry and Marchenko method.Firstly,broadband seismic data is obtained by using the frequency advantage methos,then the broadband data is processed by the data preprocessing method proposed in our paper.Then we obtained the reflection response needed by Green ' s function reconstruction.The results of numerical examples show that the side-lobe effect due to low frequency loss in Marchenko imaging results based on broadband seismic data disappears,and the true reflective interface is more focused.At the same time,this paper proposes the passive source Marchenko imaging and extends the application of Marchenko imaging to the passive source field.Firstly,based on the multi-dimensional deconvolution of noise sources,the reflection response without free surface multiples is reconstructed.This reflection response and an exact velocity model are applied to the Marchenko method to reconstruct the Green function and Marchenko imaging.A numerical example verifies the validity of the passive source Marchenko imaging.Secondly,in the numerical example,an inaccurate velocity model is applied to the Marchenko method.The final imaging results verify the robustness of the passive source Marchenko imaging.The numerical examples also compare the results of Marchenko imaging based on cross-correlation and multi-dimensional deconvolution.We find that the results of the multidimensional deconvolution based Marchenko imaging results are influenced by the stability of the least squares inversion,the accuracy of the Green's function,and the accuracy of the velocity model.The passive source Marchenko imaging effectively avoids the wavelet de-convolution processing and free-surface related multiples removal(such as SRME processing).From this perspective,passive source Marchenko imaging is the imaging method which accounts for primary waves,internal multiples,and free-surface related multiples.