Research And Application Of High-quality Migration Method For Weak Signal Imaging

Reverse time migration(RTM)is an important subsurface imaging technique used in both seismic and electromagnetic(EM)dataproccessing,however,poor illumination of the subsurface target creates a difficult problem for RTM.Staining algorithm was proposed to improve the signal-to-noise ratio(S/N)of poorly illuminated subsurface structures in seismic modeling and imaging by extract traget related wavefield.The amplitudes of the full real wavefield and the stained wavefield,i.e.,the real and the imaginary wavefields,have a difference of several orders of magnitude with the waveform greatly distorted.We proposed a generalized staining algorithm(GSA)to achieve the amplitude preservation for the stained wavefield.In GSA,the real wave-field and the stained wavefield propagate in the same velocity model.A source wavelet is used as the source of the real wavefield to generate the real wavefield;meanwhile,the real wavefield at the stained area is regarded as the boundary condition to obtain the stained wavefield.GSA keeps some properties of the staining algorithm:the stained wavefield is synchronized with the real wavefield,and it contains only the information relevant to the target region.The most significant advantage of our method is ampli-tude preservation for the stained wavefield.The stained wavefield is regarded as the forward-propagating source wavefield reqquired to meet the imaging condition,which produces the desired high quality image.The amplitude presercation of stained wavei-field makes GSA promising in quantitative illumination analysis and velocity model building.Compared to the regular staining algorithm,GSA can be adopted easily by frequency-domain wavefield propagators as well as time-domain propagators.GSA extract the seismic wavefield passing through interested areas,however,the stained wavefield contains primary reflections,multiple reflections,scatterings and mul-tiple scatterings et al.The primary reflections from the target structure contribute to the image of this structure,others can introduce noise to the final image.Therefore,we choose a short period of time,during which the primary reflections from the target structure occur,instead of the whole computational time to stain the velocity model.In this way,we can extract only the primary reflection from the target structure and reduce the computational time.This method mutes the image of structures that is irrelevant to the target structure and the noise introduced by multiple reflections,scatterings and multiple scatterings et al.The attenuation of the electromagnetic wavefield limits the application of GPR techniques and unbalances the amplitude of images at different depths.GPR data also suffers from significant noise,such as scattering and multiple reflections,which inter-fere destructively with low-amplitude image of the signal at depth.In this paper,we employ the idea of extracting useful signals and present an electromagnetic wavefield reconstruction method to label and track the EM wavefield related to a given subsurface target.We use the real EM wavefield near the target as the boundary condition for solv-ing the transverse magnetic(TM)Maxwell equation with no active source.By solving this equation,we can obtain the EM wavefield passing through the target.This estab-lishes the correspondence between the reconstructed EM wavefield and the subsurface target structures.In this work,weak amplitude waves,such as EM waves below a highly lossy overburden,are isolated and used as the forward propagating source wavefield to obtain high quality image of the critical target area.RTM is widely used because it can handle overhanging structures with turning wave.However,it requires massive computational resources especially in 3D cases.Processing muitl-shot data in one migration is a promissing way to reduce the computa-tional time for RTM.However,multi-shot data processing method introduce cross-talk artifacts caused by interference between different shots.Shot-encoding method is pro-posed to reduce cross-talk artifacts.This paper propsed a new shot-encoding method which minimizes the terms of cross-talk artifacts and our method reduces the computa-tional time of RTM.