首页|基于局部存储策略的声波全波形反演方法

基于局部存储策略的声波全波形反演方法

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
全波形反演作为一种高精度地下速度模型建模方法,自提出以来就受到了地震勘探领域的广泛关注.然而在时间域全波形反演中梯度计算需要存储或恢复震源波场,这会导致存储大量数据或额外的震源波场重构计算,使得全波形反演的实用化受到波场重构的制约.为此,本文提出了一种基于局部存储策略的声波全波形反演方法.依据奈奎斯特采样定理设置存储间隔,通过存储每个空间网格点有限时间区间内的有效波场来计算梯度,达到减少存储震源波场数据量、避免波场重构进而提升效率的目的.与其他方法相比,本方法无需波场重构且所需内存大小与正演模拟的差分精度和记录时长均无关.二维和三维推覆体模型合成数据的反演测试均表明:本方法能够在不重构波场的情况下保证反演结果的精度和信噪比,有效提高计算效率并显著降低对计算机内存的需求.
Acoustic full waveform inversion method based on local storage strategy
As a high precision underground velocity model building method,full-waveform inversion has been widely concerned in seismic exploration since it was proposed.However,in the time domain full-waveform inversion,the calculation of gradient needs to reconstruct or store the source wavefield to cross-correlate,which leads to extra storage,as well as the problem of wavefield reconstruction,which seriously restricts the practical application of full-waveform inversion.In this paper,a method of full-waveform inversion based on local storage strategy is proposed.According to the Nyquist sampling theorem,the storage interval is set to calculate the gradient by storing the effective wave field in the finite time interval of each spatial grid point,so as to reduce the storage of source wave field data,avoid wavefield reconstruction and improve efficiency.Compared with other methods,this method does not need to reconstruct the source wavefield and the required memory size is not affected by finite-difference accuracy and recording duration.The inversion results of 2D overthrust model as well as 3D overthrust model show that the proposed method can guarantee the accuracy and signal-to-noise ratio of inversion results without the source wavefield reconstruction,improve effectively the computational efficiency,and reduce significantly the requirement of computer memory.

Local storage strategyFull-waveform inversionVelocity model buildingNyquist samplingImprove the computational efficiency

杨瀚、张明坤、孙鹏远、李红辉、陈汉明、周辉

展开 >

油气资源与工程全国重点实验室,CNPC物探重点实验室,中国石油大学(北京),北京 102249

油气勘探计算机软件国家工程研究中心,西城区新康街5号东方地球物理公司,北京 100088

局部存储策略 全波形反演 速度模型建模 奈奎斯特采样 提高计算效率

国家重点研发计划变革性技术关键科学问题重点专项国家自然科学基金中国石油天然气集团有限公司前瞻性基础性项目

2018YFA0702502422741432021DJ3506

2024

10.6038/cjg2023R0441

地球物理学报
中国地球物理学会 中国科学院地质与地球物理研究所

地球物理学报

CSTPCD北大核心
影响因子：3.703
ISSN：0001-5733
年,卷(期)：2024.67(6)
  • 60