利用高性能大规模计算平台进行三维地震波模拟及各向异性反演

发布时间：2018-04-08 13:43

本文选题：并行计算　切入点：地震波模拟　出处：《中国地震局兰州地震研究所》2015年硕士论文

【摘要】：随着计算机技术的发展,单核心的超级计算机已经逐渐退出人们的视野,被越来越多的廉价并行计算机所取代。并行计算技术的逐渐壮大为科技、工业、生活带来了翻天覆地的变化。地球物理行业一直依赖于大规模计算技术的发展,本文选择了地震学中最典型的两个问题,借助中国地震局兰州地震研究所的计算机集群,开展了关于地震波正反演的研究工作。本文的第一部分工作是运用谱元法研究地震波的正演模拟。震后的灾害评估一直是地震局所关心的问题,对于台站稀少的地区,想要在震后根据台站记录勾画出峰值分布变得异常困难。本文研究了谱元法的基本原理,借助普林斯顿大学的SPECFEM程序,以岷县漳县地震为例,展示了三维全波形地震波模拟对灾害评估的可行性。建立这样的模拟平台不仅仅可以用于灾后模拟评估,也可以用于破坏的预估计,为城市规划做出贡献。本研究的第二部分工作,也是本论文的重点,是关于地壳各向异性的研究。长期以来,人们把工作的重心放在上地幔,而忽略地壳的各向异性。一方面是由于地壳的各向异性极其复杂,另一方面是由于将研究上地幔各向异性的方法移植到地壳研究时会遇到困难。本文详细的对比了三种利用接收函数研究地壳各向异性的方法.由于加权叠加法和横波分裂法在研究分层倾斜对称轴各向异性问题时均存在局限性,本文重点以波形反演法展开研究。本文分析了各向异性参数和倾斜界面对波形的影响,并通过数值实验研究了反演的可靠性,提出通过二次反演改善反演结果的方法。本文的反演代码根据raysum程序和邻域算法重新构建,并在程序中加入H-k约束,保证结果模型的有效性。文章的最后以青藏高原东北缘的16个台站5年的资料为研究对象,尝试着揭开青藏高原复杂的动力学机制。通过对接收函数反演,发现在中下地壳确实存在着S波的低速层,这与中下地壳部分熔融的学说较为吻合。根据反演结果分析,发现上地壳各向异性基本满足裂隙成因假说。如果中下地壳流的假说成立,那么可以很好的解释本研究得到的中下地壳各向异性快轴的分布(东北缘地区,在中地壳受到相邻板块的阻拦,在下地壳沿北东方向溢出)。本文的研究尚处于起步阶段,揭示青藏高原的动力学机制仍需要结合其他地球物理方法,本文的工作只是试图通过各向异性的研究提供一些判据。
[Abstract]:With the development of computer technology, single core supercomputer has been gradually out of view and replaced by more and more cheap parallel computers.The development of parallel computing technology has brought about earth-shaking changes in technology, industry and life.The geophysics industry has been relying on the development of large-scale computing technology. In this paper, the two most typical problems in seismology have been selected, with the help of the computer cluster of Lanzhou Seismological Research Institute of China Seismological Bureau.The research on forward and inverse seismic wave modeling has been carried out.The first part of this paper is to use spectral element method to study forward modeling of seismic waves.The post-earthquake disaster assessment has always been of concern to the Seismological Bureau. It is extremely difficult to map out the peak distribution based on the station records for the regions where there are few stations after the earthquake.In this paper, the basic principle of spectral element method is studied. With the help of the SPECFEM program of Princeton University, taking the earthquake in Zhangzhou County, Minxian County as an example, the feasibility of 3D full-waveform seismic wave simulation for disaster assessment is demonstrated.Such a simulation platform can be used not only for post-disaster simulation assessment, but also for pre-estimation of damage, which can contribute to urban planning.The second part of this study, which is also the focus of this paper, is the study of crustal anisotropy.For a long time, the focus of work has been on the upper mantle, ignoring the anisotropy of the crust.On the one hand, the anisotropy of the crust is extremely complex, on the other hand, it is difficult to transplant the method of studying the upper mantle anisotropy to the crust.In this paper, three methods to study crustal anisotropy by receiving function are compared in detail.Due to the limitations of the weighted superposition method and the S-wave splitting method in studying the anisotropy of the stratified inclined symmetry axis, this paper focuses on the waveform inversion method.In this paper, the influence of anisotropic parameters and inclined interface on the waveform is analyzed, and the reliability of inversion is studied by numerical experiments, and a method to improve the inversion result by quadratic inversion is proposed.The inversion code of this paper is reconstructed according to raysum program and neighborhood algorithm, and H-k constraint is added to the program to ensure the validity of the result model.At the end of this paper, the data of 16 stations in the northeastern margin of the Qinghai-Xizang Plateau are taken as the research object, and the complex dynamic mechanism of the Qinghai-Xizang Plateau is tried to be revealed.By inversion of the receiving function, it is found that there is a low velocity layer of S wave in the middle and lower crust, which is in good agreement with the theory of partial melting of the middle and lower crust.According to the analysis of inversion results, it is found that the anisotropy of upper crust basically meets the hypothesis of fracture genesis.If the hypothesis of middle and lower crustal flow is established, the distribution of anisotropic fast axis of the middle and lower crust obtained in this study can be well explained (in the northeast margin, the middle crust is blocked by adjacent plates, and the lower crust overflows along the north and east direction.The study in this paper is still in its infancy, and it is still necessary to combine other geophysical methods to reveal the dynamic mechanism of the Qinghai-Xizang Plateau. This work only attempts to provide some criteria through the study of anisotropy.
【学位授予单位】：中国地震局兰州地震研究所
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P315.31-39

【引证文献】