波场重构精度、海量计算和存储需求是制约弹性逆时偏移成像发展的重要因素。本研究基于弹性波场矢量分解原理，将弹性波波场分解为矢量纵、横波波场，构建各自的递归积分时间波场外推公式；通过求解Christoffel 方程和低秩分解计算有限差分系数，将纵横波递归积分时间波场外推公式进行低秩有限差分表征，构建起低秩有限差分法弹性波场正向、反向延拓算子；基于辅助微分方程算法建立适用于二阶弹性波方程的复频移完全匹配层吸收边界，通过波场外推算子和完全匹配层辅助方程的迭代更新衰减来自边界的虚假反射；通过将激发振幅成像条件与炮域转换波极性校正方法结合，建立极性一致的激发振幅成像条件，将激发振幅成像条件从声波逆时偏移推广到弹性逆时偏移；研究CPU/GPU异构并行的高性能计算方法，实现高效率的弹性逆时偏移。随着对地震成像精度要求的不断提高，可以预计本项目的开展将会为复杂油气藏地震勘探提供一套行之有效的地震成像方法。

The accuracy of wavefield reconstruction and huge computation and storage requirement are major restrictions of elastic reverse time migration (RTM). To solve these problems, based on the elastic wavefield decomposition theory, we first decompose the wavefield into vector P and S wavefields and construct the corresponding elastic recursive integral time-extrapolation (RITE) formulas; second, we calculate Christoffel equations and Lowrank decomposition to derive the elastic wavefields lowrank finite difference operators; third, according to the method of auxiliary differential equations (ADE), we derive a kind of perfectly matched layer (PML) boundary condition which is suitable for two-order elastic wave equations to weaken the artificial boundary reflection; fourth, we propose the polarity-consistent excitation amplitude imaging condition by combining a shot domain polarity reversal correction method of converted waves and excitation amplitude imaging condition of acoustic RTM; Lastly, we study the highly-efficient elastic RTM based on the CPU/GPU high-performance computation platform. With the higher requirement of imaging accuracy, this project will provide an effective imaging methods with high efficiency and accuracy for complex reservoir exploration.