声介质波动方程反射层析反演方法研究
发布时间：2018-12-20 19:05
【摘要】：勘探地震学中,背景速度是决定地震波传播的关键因素,是进一步反演速度场更高成分的基础。对于透射波和折射波,其具有较大的入射角孔径,可以用来恢复速度模型中的长波长分量;但是对于反射波,由于较小的反射角孔径,通常用于恢复速度场短波长分量。本文针对上述问题围绕反射波路径的推导方法,反射波层析目标泛函的构建及其反射波层析对全波形反演效果的改善进行了研究。从而实现了利用反射波信息来反演速度场长波长分量的过程。“波路径”能更准确地描述地震波在地球介质中的传播过程,相对传统射线路径,波路径能够避免高频射线路径在高速区域聚焦、在低速区域发散的问题。波动方程层析成像正是在波路径的基础上提出的。本文提出的反射波路径由两部分构成,一部分是背景入射场入射到地下一个散射点,产生二次场,二次场对地下所有其它点激励后产生的场的变化;另一部分代表从源点出发的Green函数激励地下一个散射点,产生二次场,二次场对地下所有其它点激励后产生的场的变化。这两部分的叠加产生完整的梯度量,即反射波路径。我们将数据残差沿着该反射波路径进行反投影,便可实现反射波层析反演。反演过程中根据不同的初始速度场选择不同的目标函数。由于模拟地震数据中地震波振幅的不确定性,建立在传统最小二乘误差泛函基础上的波形反演严重受到了制约。但是,各种形式的波动方程基本遵循共同的程函方程,其旅行时信息基本不会受到影响。故在波形反演中应当更注重旅行时信息。当初始速度场与真实速度场相差较大时,我们选择旅行时反演,降低其对波动方程的依赖程度。只有在初始速度场较为准确时我们方可采取波形反演。本文利用偏移与反偏移算子从背景速度场中获取反射波信息,构建了反射波路径,进而建立波动方程反射层析的梯度公式。在反演过程中,本文对比了不同的目标函数的反演效果。在波动方程反射旅行时反演中,结合Ma Y和Luo Y的思想改进了反射层析中其伴随源的形式。同时运用动态图像校正方法来计算反射波旅行时差,实现了利用反射波旅行时信息来反演速度场中低波数成分的过程。在波动方程反射波形反演中,实现了较为复杂的模型试算,并改善了FWI的效果。模型试算结果表明,该方法能够提高反演的稳定性,具有更高应用潜力。
[Abstract]:In exploration seismology background velocity is the key factor to determine the propagation of seismic wave and the basis for further inversion of the higher component of velocity field. For transmission wave and refraction wave, it has large incident angle aperture, which can be used to recover the long wavelength component in velocity model, but for reflection wave, it is usually used to restore the short wavelength component of velocity field because of small reflection angle aperture. This paper focuses on the derivation of the reflected wave path, the construction of the target functional of the reflection wave tomography and the improvement of the full waveform inversion effect by the reflection wave tomography. Thus, the long wavelength component of velocity field can be retrieved by using reflected wave information. The "wave path" can more accurately describe the process of seismic wave propagation in the earth's medium. Compared with the traditional ray path, the wave path can avoid the problem of high frequency ray path focusing in the high speed region and divergence in the low speed region. Wave equation tomography is based on the wave path. The reflected wave path proposed in this paper is composed of two parts. One is the background incident field incident to a scattering point under the ground, which produces the secondary field and the change of the field after the secondary field excites all the other underground points. The other part represents the Green function from the source point which excites a scattering point in the ground and produces the quadratic field and the change of the field after the quadratic field excites all the other points in the ground. The superposition of these two parts produces a complete gradient, that is, the reflected wave path. The reflection wave tomography inversion can be realized by back-projecting the data residuals along the reflected wave path. In the inversion process, different objective functions are selected according to different initial velocity fields. Because of the uncertainty of the amplitude of seismic wave in simulated seismic data, the waveform inversion based on the traditional least square error functional is seriously restricted. However, all kinds of wave equations follow the common equation, and the travel information will not be affected. Therefore, more attention should be paid to travel time information in waveform inversion. When the difference between the initial velocity field and the real velocity field is large, we choose travel time inversion to reduce its dependence on the wave equation. Only when the initial velocity field is more accurate can we take waveform inversion. In this paper, the reflected wave path is constructed by using migration and inverse migration operators to obtain the reflected wave information from the background velocity field, and then the gradient formula of reflection tomography for wave equation is established. In the process of inversion, the inversion effects of different objective functions are compared. In the reflection travel time inversion of wave equation, the form of accompanying source in reflection tomography is improved by combining the ideas of Ma Y and Luo Y. At the same time, the dynamic image correction method is used to calculate the travel time difference of the reflected wave, and the process of retrieving the low wavenumber component of the velocity field by using the travel time information of the reflected wave is realized. In the reflection waveform inversion of wave equation, a more complicated model trial calculation is realized, and the effect of FWI is improved. The model results show that this method can improve the stability of inversion and has higher application potential.
【学位授予单位】：中国石油大学(华东)
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.4
本文编号：2388405
[Abstract]:In exploration seismology background velocity is the key factor to determine the propagation of seismic wave and the basis for further inversion of the higher component of velocity field. For transmission wave and refraction wave, it has large incident angle aperture, which can be used to recover the long wavelength component in velocity model, but for reflection wave, it is usually used to restore the short wavelength component of velocity field because of small reflection angle aperture. This paper focuses on the derivation of the reflected wave path, the construction of the target functional of the reflection wave tomography and the improvement of the full waveform inversion effect by the reflection wave tomography. Thus, the long wavelength component of velocity field can be retrieved by using reflected wave information. The "wave path" can more accurately describe the process of seismic wave propagation in the earth's medium. Compared with the traditional ray path, the wave path can avoid the problem of high frequency ray path focusing in the high speed region and divergence in the low speed region. Wave equation tomography is based on the wave path. The reflected wave path proposed in this paper is composed of two parts. One is the background incident field incident to a scattering point under the ground, which produces the secondary field and the change of the field after the secondary field excites all the other underground points. The other part represents the Green function from the source point which excites a scattering point in the ground and produces the quadratic field and the change of the field after the quadratic field excites all the other points in the ground. The superposition of these two parts produces a complete gradient, that is, the reflected wave path. The reflection wave tomography inversion can be realized by back-projecting the data residuals along the reflected wave path. In the inversion process, different objective functions are selected according to different initial velocity fields. Because of the uncertainty of the amplitude of seismic wave in simulated seismic data, the waveform inversion based on the traditional least square error functional is seriously restricted. However, all kinds of wave equations follow the common equation, and the travel information will not be affected. Therefore, more attention should be paid to travel time information in waveform inversion. When the difference between the initial velocity field and the real velocity field is large, we choose travel time inversion to reduce its dependence on the wave equation. Only when the initial velocity field is more accurate can we take waveform inversion. In this paper, the reflected wave path is constructed by using migration and inverse migration operators to obtain the reflected wave information from the background velocity field, and then the gradient formula of reflection tomography for wave equation is established. In the process of inversion, the inversion effects of different objective functions are compared. In the reflection travel time inversion of wave equation, the form of accompanying source in reflection tomography is improved by combining the ideas of Ma Y and Luo Y. At the same time, the dynamic image correction method is used to calculate the travel time difference of the reflected wave, and the process of retrieving the low wavenumber component of the velocity field by using the travel time information of the reflected wave is realized. In the reflection waveform inversion of wave equation, a more complicated model trial calculation is realized, and the effect of FWI is improved. The model results show that this method can improve the stability of inversion and has higher application potential.
【学位授予单位】：中国石油大学(华东)
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.4
【参考文献】
相关期刊论文 前2条
1 任浩然;黄光辉;王华忠;陈生昌;;地震反演成像中的Hessian算子研究[J];地球物理学报;2013年07期
2 王华忠;冯波;李辉;王雄文;胡江涛;;各种速度分析与反演方法的对比研究[J];岩性油气藏;2012年05期
相关博士学位论文 前1条
1 刘玉金;复杂介质条件下基于反演的地震成像方法研究[D];中国石油大学(华东);2014年
相关硕士学位论文 前1条
1 雷蕾;非均匀介质地震波散射波场正演模拟方法研究[D];中国石油大学;2010年
，本文编号：2388405
本文链接：https://www.wllwen.com/kejilunwen/diqiudizhi/2388405.html
