董良国,,
王建华
同济大学海洋地质国家重点实验室, 上海 200092
基金项目: 国家重点研发计划深海关键技术与装备重点专项(2019YFC0312004)和国家自然科学基金项目(41874127)联合资助
详细信息
作者简介: 张建明, 男, 1996年生, 现为同济大学海洋与地球科学学院博士研究生, 主要从事地震波反演与近地表建模等方面的研究.E-mail: jm_zhang@tongji.edu.cn
通讯作者: 董良国, 同济大学海洋与地球科学学院教授, 博士生导师, 主要从事地震波传播理论与数值模拟、地震波反演等方面的研究.E-mail: dlg@tongji.edu.cn
中图分类号: P631收稿日期:2020-07-27
修回日期:2021-05-12
上线日期:2021-07-10
First-arrival multi-information joint inversion method based on wave-equation
ZHANG JianMing,DONG LiangGuo,,
WANG JianHua
State Key Laboratory of Marine Geology, Tongji University, Shanghai 200092, China
More Information
Corresponding author: DONG LiangGuo,E-mail:dlg@tongji.edu.cn
MSC: P631--> Received Date: 27 July 2020
Revised Date: 12 May 2021
Available Online: 10 July 2021
摘要
摘要:地震初至波中包含着丰富的近地表速度结构信息，如何分阶段、分尺度地利用这些信息进行近地表速度建模是地震勘探中的一个关键问题.在速度反演的不同阶段，综合利用初至波中的不同信息（如走时、包络和波形等）进行联合反演，可以有效地降低反演对初始模型的依赖程度，提高近地表速度模型的反演精度.为此，本文提出了一种统一基于波动方程正演引擎的初至波多信息联合反演方法，该方法同时匹配观测和模拟的初至波走时、包络和波形信息.在不同的反演阶段选择不同的权重因子调节不同信息的权重，这样不仅降低了反演对初始速度模型的依赖，而且自然地实现了多尺度反演.在每轮反演迭代中，一次正演模拟的波场同时应用于初至波走时、包络和波形匹配，无需额外的射线追踪.同时，联合反演方法在一定程度上缓解了串联反演中目标函数漂移问题，提高了近地表速度建模的精度.
关键词: 近地表建模/
目标函数/
多尺度/
联合反演
Abstract:The first-arrival of seismic wave contains abundant information about the near-surface structure. How to use these information to build the near-surface velocity model with different scales is a key issue in different exploration stages. Ideally, in different stages of velocity inversion, comprehensive utilization of first-arrival multi-information (such as travel time, envelope and waveform, etc.) for joint inversion can not only reduce the dependence on the initial velocity model, but also improve the inversion accuracy of the near-surface velocity model. Based on the unified seismic wave equation modeling engine, a new first-arrival multi-information joint inversion method is proposed which matches the observed and simulated first-arrival traveltime, envelope and waveform information at the same time. In different inversion stages, different weighting factors are selected to adjust the weight of different information, which not only reduces the dependence of inversion on the initial velocity model, but also realizes multi-scale inversion naturally. In every iteration of the inversion, one-time simulated wavefield are simultaneously used in first-arrival traveltime, envelope and waveform fitting. Therefore, there is no need for additional ray tracing. At the same time, the joint inversion method can partially mitigate the drift phenomenon of objective function in sequential inversion, and improve the accuracy of near-surface velocity model building.
Key words:Near-surface model building/
Misfit function/
Multi-scale/
Joint inversion
PDF全文下载地址:
http://www.geophy.cn/data/article/export-pdf?id=c2914897-c26b-4663-9684-2cbd9c8d589a
