地球物理学进展 ›› 2019, Vol. 34 ›› Issue (4): 1481-1488.doi: 10.6038/pg2019CC0221

• 应用地球物理学Ⅰ(油气及金属矿产地球物理勘探) • 上一篇    下一篇

基于频移目标函数的包络反演方法

梁展源,吴国忱,张晓语   

  1. 中国石油大学(华东)地球科学与技术学院,山东青岛 266580
  • 收稿日期:2018-09-14 修回日期:2019-06-12 出版日期:2019-08-20 发布日期:2019-08-30
  • 作者简介:梁展源,男,1989年生,山东莱阳人,博士研究生,主要从事非均匀介质全波形反演方法研究. (E-mail: lzy09017216@163.com)
  • 基金资助:
    国家科技重大专项(2016ZX05024-001-008)

Envelope inversion method based on frequency-shifted objective function

LIANG Zhan-yuan,WU Guo-chen,ZHANG Xiao-yu   

  1. School of Geosciences, China University of Petroleum, Shandong Qingdao 266580, China
  • Received:2018-09-14 Revised:2019-06-12 Online:2019-08-20 Published:2019-08-30

摘要:

地震数据的包络携带着超低频信息,能够在缺失低频的情况下,为全波形反演提供一个精确的初始模型.包络反演虽然利用超低频数据进行反演,但需与全波形反演采用同样精细的网格,具有几乎相同的计算量.根据包络数据的特性,对于给定的频带,在频率移动时,地震数据会发生变化,但是它的瞬时振幅包络是不变的.基于此性质,我们提出一种频移目标函数降低地震数据的频带,推导了相应的梯度算子.新方法可以在保持精度不变的情况下减少包络反演的计算量,提高计算效率.模型测试采用重采样后Marmousi模型,对比了频移前后的地震数据,分析了不同频移数据反演速度的精度,验证了本文方法的可行性与有效性.

关键词: 有限差分, 全波形反演, 包络反演, 低频数据, 多尺度反演

Abstract:

The envelope of seismic data carries ultra-low frequency information and can provide an accurate initial model for full waveform inversion without the low frequency. Although envelope inversion uses ultra-low frequency data for inversion, it needs to use the same fine mesh as full waveform inversion with almost the same amount of computation. According to the characteristics of the envelope data, for a given frequency band, the seismic data will change when the frequency moves, but its instantaneous amplitude envelope is constant. Based on this property, we propose a frequency shift objective function to reduce the frequency band of seismic data and derive the corresponding gradient operator. The new method can reduce the calculation of envelope inversion and improve the computational efficiency while maintaining the accuracy. The model test uses the Marmousi model after resampling. We compare the seismic data before and after the frequency shift and analyse the accuracy of the inverted velocity of different frequency shift data which verify the feasibility and effectiveness of the proposed method.

Key words: Finite difference, Full waveform inversion, Envelope inversion, Low frequency data, Multiscale inversion

中图分类号: