The Research Of Least-squares Reverse Time Migration Using Correlation Objective Function

At the present stage,seismic migration imaging technique is playing an increasingly important role in the underground structure imaging,from ray-tracing-based Kirchhoff migration and Beam migration,to the wave-propagation-based one-way wave equation migration and reverse time migration.For complicated structures,RTM is the state-of-the-art imaging technology.It based on the true amplitude migration imaging theory,which aims to compensate automatically for the geometric spreading occurring during the migration process and obtain the relatively accurate reflectivity,and have the advantages of no angle limits.It can image the special wave phenomenon such as multiple wave,converted wave,prismatic wave,etc.However,the imaging results using the conventional migration method are just blurry image,which can not be able to meet the needs of lithologic reservoir exploration.Based on the seismic inversion theory,the least-squares migration imaging method matches the predicted data and the observed data under the Born approximation,then obtained the higher-quality profiles which is conducive to better lithological reservoir imaging and reservoir parameter inversion.In practice,it is not an easy task to directly apply the conventional least-squares RTM(LSRTM)to the observed data and get the higher-quality profiles.First,the earth is at least a viscoelastic medium with density variations,and it is much more complicated than the models that we use the acoustic wave equation to propagate acoustic fields in the LSRTM algorithm.Also,it is extraordinarily difficult to define a good source wavelet in the modeling,and the source wavelets from different source records are different,moreover,the source wavelet are space-variant dramatically with strong absorption attenuation under the local media.These factors lead to a serious physical problem,which we are hard to get accurate amplitude information in the conventional LSRTM process,and the conventional LSRTM based on amplitude matching requires a high degree of accuracy of amplitude information.Therefore,all these practical issues need a considerable effort in preprocessing the observed and simulated data to correctly use the conventional LSRTM algorithm.Based on the analysis of the problems existing in the conventional LSRTM,we develop a phase-matching-based LSRTM.This method abandons the quadratic function which the traditional ways always used,and employs the cross-correlation function,in this way,this method emphasize more on the role of the seismic phase information instead of amplitude information in the new inversion framework.Hence,in theory,it can better apply to real data.Based on the correlation objective function building,synthesizing predicted data by reverse time demigration,precondition processing,pre-conditional conjugate gradient,we explore and realize a new LSRTM algorithm flow using the correlation objective function.Model measurement and actual data processing results show that the proposed method has applicability and effectiveness.