面波成像及三维横波速度方位各向异性反演研究

发布时间：2018-06-27 03:40

本文选题：面波层析成像 + 连续模型正则化　；参考：《中国科学技术大学》2017年硕士论文

【摘要】：面波成像是研究地壳上地幔速度结构和变形特征的最为重要的手段之一。基于连续模型正则化的面波层析成像被广泛应用于反演二维相速度或群速度分布。其中由于台站和地震事件空间分布不均匀,不可避免地导致射线路径分布不均匀,从而影响成像结果的空间分辨率。基于连续模型正则化方法,我们提出一种改进的空间自适应平滑方法。通过把约束反演模型平滑度的先验模型协方差矩阵中的固定相关长度,改进为基于射线密度的相关长度函数,实现了空间的自适应平滑。此外,我们提出了用以压制数据异常的两步反演策略。在合成数据测试中,我们设置不同尺度异常样式检测板模型,通过与传统方法反演结果对比,表明新的方法可以在射线路径覆盖密集区域有效提高模型恢复。对于青藏高原东南部地区的实际数据应用,空间自适应平滑方法反演出与传统方法相似的结构,而且新方法结果展现了更为精细的结构,并且对异常幅值有着更好的解析。通过合成测试和实际数据应用,表明我们的新方法在射线分布不均匀区域可以有效实现分辨率的空间变化。另一方面,地震面波所反映出的方位各向异性一直被认为反映了应力应变作用下矿物和岩石裂隙的定向排列,以及地壳、上地幔随深度变化的形变模式。传统的深度方位各向异性结构,通常由逐点的Rayleigh波纯路径频散及其方位各向性反演得到。我们提出了一个使用Rayleigh波混合路径频散直接反演横波速度方位各向异性变化的方法,从而不必进行二维群速度或者相速度的构建。在之前的研究中,我们已经构建了基于采用邻近算法(NA)的两步法反演一维深度横波波速和方位各向异性的程序(Yao 2015)。在采用了改进MINEOS程序后,我们可以更为精确地计算Rayleigh波相速度对于横向各向同性介质(TI)五个弹性模量(A,C,F,L,N)的敏感度核函数。同时,在我们之前提出的频散走时直接反演三维横波速度结构并进行面波射线路径追踪的基础上(Fangetal.2015),我们提出了基于射线追踪的三维横波各向异性直接反演方法,从而避免了面波研究中通常使用的大圆路径假设,进而可以应用于小区域复杂介质。通过一系列的合成数据测试,我们对新方法进行了系统的检验。首先,我们直接使用真实各向同性模型作为参考模型进行反演,其结果与输入模型较好拟合,从而证明了三维各向异性直接反演方法的可靠性。接着在第二个合成数据测试中,我们给出了使用Rayleigh波频散直接反演得到三维横波速度结构和三维横波速度方位各向异性结构的处理流程。最后通过使用加入不同比例高斯随机噪声的合成数据,我们证明了直接反演方法在一定程度噪声水平下对快波方位角反演的稳定性,尽管随着噪声水平的提高,很难对方位各向异性的幅值进行较好的恢复。合成测试结果表明,当使用数据的射线覆盖足够好,Rayleigh波频散测量数据足够精确时,我们的直接反演方法可以有效地反演出三维横波速度方位各向异性的变化。
[Abstract]:Surface wave imaging is one of the most important means to study the velocity structure and deformation characteristics of the crust on the mantle. Surface wave tomography based on continuous model regularization is widely used to invert two-dimensional velocity or group velocity distribution. The distribution of station and seismic events is inhomogenous, which inevitably leads to the distribution of ray paths. Based on the continuous model regularization method, we propose an improved spatial adaptive smoothing method. By using the fixed correlation length in the prior model covariance matrix of the smoothness of the constrained inversion model, we have improved the space based on the correlation length function based on the line density. In addition, we propose a two step inversion strategy to suppress the data anomaly. In the synthetic data test, we set up the different scale anomaly pattern detection board model. By comparing the traditional method inversion results, it shows that the new method can effectively improve the model recovery in the dense area of the ray path. For the Qinghai Tibet Plateau, the new method can be used to improve the model recovery. In the original data of the southeastern region, the spatial adaptive smoothing method was used to reverse the structure similar to that of the traditional method, and the results of the new method show a more fine structure and have a better analysis of the anomalous amplitude. It is shown that our new method can be used in the uneven distribution of the ray distribution through the synthetic test and the practical data application. On the other hand, the azimuthal anisotropy reflected by seismic surface waves has been considered to reflect the directional arrangement of mineral and rock fractures under stress and strain, as well as the deformation pattern of the crust and upper mantle with depth. The traditional depth azimuthal anisotropic structure, usually by point by point Rayleigh We have proposed a method to invert the azimuthal anisotropy of the transverse wave velocity using the Rayleigh wave mixing path dispersion, so we do not have to construct the two-dimensional group velocity or phase velocity. In the previous study, we have built the approach based on the proximity algorithm (NA). The two step method for inversion of the one-dimensional depth transverse wave velocity and azimuth anisotropy (Yao 2015). After using the improved MINEOS program, we can more accurately calculate the sensitivity kernel function of the Rayleigh wave velocity to five elastic moduli (A, C, F, L, N) of the transverse isotropic medium (TI). Based on the inversion of the three-dimensional transverse wave velocity structure and tracking the surface wave ray path (Fangetal.2015), we propose a three-dimensional transverse wave anisotropic direct inversion method based on ray tracing, thus avoiding the large circular path hypothesis commonly used in the study of surface waves, and thus can be applied to complex media in the area. We systematically test the new method. First, we directly use the true isotropic model as the reference model, and the results are well fitted to the input model, which proves the reliability of the three-dimensional anisotropic direct inversion method. We have given the second synthetic data tests. The processing flow of three-dimensional transverse wave velocity structure and three-dimensional transverse wave velocity azimuth structure is obtained by using Rayleigh wave dispersion directly. Finally, by using the synthetic data with different proportions of Gauss random noise, we prove the stability of the direct inversion method to the azimuth angle inversion of the fast wave in a certain degree of noise. With the improvement of the noise level, it is difficult to recover the amplitude of azimuth anisotropy. The results of synthetic test show that when the data's ray coverage is good enough and the Rayleigh wave dispersion measurement data are accurate enough, our direct inversion method can effectively retrieve the change of azimuthal anisotropy of the velocity of the three-dimensional transverse wave.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P315

【相似文献】