基于振幅梯度预处理的多尺度全波形反演方法研究
Multi-scale full waveform inversion method based on gradient preprocessing
张建峰 1王志亮 1李江 1麻旭刚 1李博闻 2张凯3
作者信息
- 1. 中海石油(中国)有限公司天津分公司,天津 300459
- 2. 中海油能源发展股份有限公司工程技术分公司,天津 300459
- 3. 中国石油大学(华东)地球科学与技术学院,青岛 266580
- 折叠
摘要
在地震勘探中,全波形反演方法能充分利用地震资料中的动力学、运动学信息,高精度的反演地下介质参数.然而,由于全波形反演高度的非线性,在反演过程中易收敛的局部极小值.并且随着深度的增加,深层区域的反演效果不佳.因此,本文首先采用多尺度策略:随着迭代次数的增加,反演所需的地震资料由低频向高频逐渐增加,该策略能够有效降低了全波形反演的强非线性.并提出的基于地震波振幅的梯度预处理方法,模型试算结果证明了这种方法的正确性和有效性,可以提高深部能量,从而提高了深层区域的反演效果.
Abstract
In seismic exploration,the full-waveform inversion method can fully utilize the dynamics and kinematics information in the seismic data to invert the parameters of the subsurface medium with high accuracy.However,due to the high degree of nonlinearity of full-waveform inversion,it is easy to converge to the local minima in the inversion process.And with the increase of depth,the inversion effect in the deep region is not good.Therefore,this paper firstly adopts the multi-scale strategy;with the increase of iteration number,the seismic data required for inversion gradually increases from low frequency to high frequency,and this strategy can effectively reduce the strong nonlinearity of the full-waveform inversion.And the proposed gradient preprocessing method based on seismic wave amplitude proves the correctness and effectiveness of the method through model testing,which can improve the deep energy,thus improving the inversion effect in the deep region.
关键词
全波形反演/多尺度策略/梯度预处理/速度建模Key words
Full waveform inversion/Multi-scale strategy/Gradient preprocessing/Velocity modeling引用本文复制引用
基金项目
中海石油(中国)有限公司天津分公司重大攻关项目(CCL2022TJXONST1187)
出版年
2024
NETL
NSTL
万方数据