Least-squares reverse time migration with static plane-wave encoding
HUANG Jian-Ping;LI Chuang;LI Qing-Yang;GUO Shu-Juan;DUAN Xin-Biao;LI Ji-Guang;ZHAO Sheng-Tian;BU Chang-Cheng;Department of Geophysics,China University of Petroleum (East China);SINOPEC Geophysical Research Institute;Shengli Geophysical Research Institute of SINOPEC;
To solve the exploration problem of widely distributed fault block reservoirs in China,high-density seismic acquisition methods with small surface element and high coverage times become more popular recently which results in the huge dataset for seismic processing.Planewave migration is a kind of areal shot migration method which can process enormous data of highdensity acquisition with much higher computational efficiency compared with shot domain migration.But conventional plane-wave migration produces low quality images with crosstalk.To solve these problems,the paper introduces the plane-wave migration to the inversion framework and presents the theory and work flow of least-squares reverse time migration with static plane-wave encoding.The key points of plane-wave least squares reverse time migration can be divided into two parts.Firstly,the plane-wave decomposition is applied to the shot data which can be performed by using some delayed-shot variant of slant-stack processing.This involves applying a linear timedelay to the shot records.And the time delay is also applied to the corresponding sources.Note that there is one-to-one correspondence between source encoding and encoding to common shot gathers.The input data of migration is compressed a lot after plane-wave encoding which improves the computational efficiency.Secondly,the misfit function of plane-wave least squares reverse time migration is defined as the data difference between predicted plane-waves by Born modelling and observed plane-waves in the pre-stack migration.The conjugate gradient method is implemented to find the solution that minimizes the misfit function.However,the searching of CG method becomes very slow after several iterations in practical application,so we set the steepest descent direction as the gradient after several iterations to restart the CG method.One other thing to note is that the migration results are updated separately with different plane-waves and stacked together after the iterations are finished.The plane-wave least squares reverse time migration is firstly tested via the synthetic data set with the three layer medium model to verify the validity of the proposed method.The plane-wave reverse time migration is also applied to the model for better comparison.The comparison of imaging quality,vertical amplitude and transverse amplitude balance are presented which illustrates the advantages of the proposed method.The stable convergence also proves the robustness of the method.Then,the proposed method is applied to a complex fault block model in Shengli Oil-field to test its applicability.Several fault blocks are distributed in the mid-deep part and a high steep structure locates on the right of the model.The strong velocity variation,the burial depth and fuzzy boundary of fault blocks make it a good velocity model to test the resolution of the imaging method.The imaging results of the proposed method is compared with the reverse time migration,the split step migration and least-squares revere time migration results,which show that the proposed method can produce high quality images similar to leastsquares reverse time migration but only np/ns times computation is needed(ns is the number of the shots and npis the number of the encoded plane-waves).The paper presents the theory of prestack least-squares reverse time migration with a static plane-wave encoding technique.From the imaging test we can get the conclusions including:(1)the proposed method has the advantages of resolution enhancement and amplitude compensation in mid-deep part with the increase of iterations;(2)the crosstalk introduced by plane-waves can be suppressed by the stacking of images from different plane-waves and the optimization;(3)the computation and input/output cost is reduced a lot with the plane-waves encoding.Besides,the common image gathers can be extracted efficiently from the migration results of the proposed method which is also available for the migration velocity analysis and our next work will be focused on this part. However,the proposed method is only implemented with the fixed geometry and another point of our next work is to modify the method to fit irregular geometry.
【Fund】： 国家973课题(2014CB239006 2011CB202402);; 国家自然科学基金(41104069 41274124);; 山东省自然科学基金(ZR2011DQ016);; 中央高校科研业务费专项基金(R1401005A)联合资助
【CateGory Index】： P631.44
【CateGory Index】： P631.44