复杂介质频率域弹性波方程全波形反演研究

项目名称： 复杂介质频率域弹性波方程全波形反演研究

项目编号： No.41474102

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 天文学、地球科学

项目作者： 韩波

作者单位： 哈尔滨工业大学

项目金额： 100万元

中文摘要： 利用弹性波方程进行弹性数据的全波形反演，由于能够包含体波和面波的全部信息，真正符合复杂介质和各向异性的勘探需要，成为了地震勘探全波形反演的研究热点和发展趋势，正在引起人们极大的关注。但由于这是一个计算量和存储量巨大的问题，同时受到计算资源的限制、局部极值问题以及介质复杂性的困扰，利用传统的反演理论与算法很难开展弹性波全波形反演研究。为此，本申报项目采用频率域弹性波方程，拟从快速正演模拟、基于线性采样法的初始模型、稀疏约束与全变差正则化反演方法、贝叶斯推断、并行计算等几个方面进行综合研究，解决这一计算挑战问题。探索频率域全波形反演问题解的稀疏性，实现对复杂介质弹性波方程的多参数精细反演成像，为石油勘探及复杂的工程地球物理勘探问题提供理论支撑。本项目预期在快速算法与并行计算、异常体边界的精细识别、以及多参数反演等方面的研究有较大的突破。

中文关键词： 全波形反演；弹性波方程；稀疏约束正则化；全变差正则化；并行计算

英文摘要： The full waveform inversion (FWI) benefits from the recorded data of the elastic wave modeling, incorporates the information of both the body wave and the surface wave and takes into account both the heterogeneity and anisotropy. It has recently become a spotlight and a development trend of the research in seismic exploration. However, due to the huge computational cost, FWI suffers from the constraints of the computational resources. The strong nonlinearity and the highly heterogeneous medium also make FWI an optimization problem with multiple local minima. Therefore, the conventional inverse theory has difficulties in dealing with the elastic FWI. In order to overcome the challenging problem, in this project, we intend to utilize the frequency-domain elastic wave equation, develop a fast forward modeling algorithm, construct a good initial estimation model from the linear sampling method, formulate the regularization method based on the sparsity-constraints and the total variation, incorporate the Bayesian inference and the parallel computing techniques. We will investigate the sparsity of the solution to the frequency-domain problem and endeavor to obtain a high-resolution multi-parameter inversion of the subsurface structure. We would like to supply a theoretical support for the oil exploration and the complex engineering geophysical exploration. The project aims to make a breakthrough in the investigation of the fast forward modeling algorithm, the parallel computing, the fine recognition of the abnormal edge and the inversion of multiple parameters.

英文关键词： full waveform inversion;elastic wave equation;sparsity constraint regularization;total variation regularization;parallel computing