基于OC-seislet变换的三维叠前复杂地震波场迭代数据插值方法研究

项目名称： 基于OC-seislet变换的三维叠前复杂地震波场迭代数据插值方法研究

项目编号： No.41274119

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 天文学、地球科学

项目作者： 刘洋

作者单位： 吉林大学

项目金额： 90万元

中文摘要： 许多地震资料处理方法需要完整的数据信息，但是野外施工条件等因素使观测系统很难记录完整的地震波场，如空间采样率不足和地震道缺失等，尤其在复杂构造条件下的三维地震勘探中，缺失的叠前地震数据给后续处理带来严重的影响。传统数据插值通常只基于图像处理或波动方程的方法，每类方法很难同时快速并且准确地解决复杂地震波场（如散射波存在）的插值问题。结合两类方法的优点，开发地震波动力学预测理论、图像处理变换方法和压缩感知理论相结合的三维复杂地震波场数据插值方法具有重要的意义。本研究拟提出地震波动力学炮检距连续（OC）算子，构建能够表征复杂地震波场的三维OC-seislet稀疏变换，应用最佳化变换基函数对缺失部分数据的三维复杂地震波场进行压缩，利用波场在变换域内具有极其稀疏（少数元素非零）的特性，结合快速Bregman迭代算法进行变换域内处理，结果再反变换到数据域，最终实现复杂地震波场缺失数据的快速、准确插值。

中文关键词： 地震数据插值；炮检距连续算子；OC-seislet变换；Bregman迭代；压缩感知

英文摘要： Many seismic data processing methods need intact data information, however, an ideal uniform coverage is rarely achieved because of practical and economic constraints, eg., spatial undersampling and seismic trace missing, especially, prestack missing data in 3D complex structural condition will affect the further data processing workflow. Traditional data interpolation methods are usually based on imaging processing or wave equation methods, each class of methods are hard to both quickly and accurately solve the interpolation problem of complex wavefield (eg., presence of diffraction wave). It is important that one combines two methods and develops 3D complex wavefield interpolation method that includes seismic dynamic prediction theory, imaging processing transform, and compressive sensing theory. This research will propose offset continuation (OC) operator from seismic wave dynamics, build a 3D OC-seislet transform that can represent 3D complex seismic wavefield, use the sparse characteristics (only few elements are nonzero) of seismic wavefield in the transform domain, process the transform coefficients in the model space by using Bregman iterative algorithm, transform model space back to data space, and finally implement quickly and accurately missing-data interpolation method of complex seismic wavefield.

英文关键词： seismic data interpolation；offset continuation operator；OC-seislet transform；Bregman iteration；compressive sensing