【作者】 刘志坤；

【导师】 黄金莉；

【作者基本信息】 中国地震局地震预测研究所 ， 固体地球物理学， 2010， 硕士

【摘要】 2008年5月12日四川省汶川县发生了Ms8.0级强烈地震,造成了重大的人员伤亡和经济损失,这是新中国成立以来破坏性最大的地震,也是自1976年唐山地震后发生在中国大陆伤亡最惨重的一次地震。大地震发生在四川盆地与川西高原交界的龙门山断裂带上,发震断层向北东方向单侧扩展,全长超过300km,沿龙门山中央断裂和前山断裂形成约240km和72km的地表破裂带,在以龙门山断裂带为中心的较大范围内都有显著的同震地表形变;同时,地下介质的属性和状态可能也会发生一些变化,研究这些变化对于了解汶川地震机理、断裂带演化规律及地震危险性分析有着极其重要的意义。本论文利用国际上近两年最新发展的背景噪声互相关方法并结合尾波干涉的思想研究了汶川地震前后震源区及周边相对地震波速度的变化。基于背景噪声的地下介质速度结构及其变化的研究是目前地震学的研究热点之一。大气、海洋和固体地球相互作用产生了全球的背景噪声场,对两个地震台站的背景噪声进行互相关运算可提取其间的经验格林函数,进而反演地下介质的速度结构,而不同时期的经验格林函数波形能保持很高的稳定性和相似性,对这些波形应用尾波干涉法测量能高精度地检测地下结构随时间的变化。由于背景噪声数据的连续性,可以实现对波速变化的全时监测。本研究应用了2001年-2009年汶川地震震源区周边的四川区域地震台连续波形资料的垂直分量。数据处理包括三个主要步骤:首先,通过单台数据预处理得到地震背景噪声;然后,对两个台站的背景噪声进行互相关运算提取台站间的经验格林函数;最后,在0.1-0.5Hz频带下计算不同时段的经验格林函数与参考经验格林函数的走时偏移,进而得到相对地震波速度随时间的变化。数据预处理中的关键步骤是时域归一化,研究中对比了已有的几种时域归一化方法,并对滑动窗绝对值平均方法作了适当的改进,取得了较好效果。通过大量的试算对移动窗口互相关法、移动窗口互谱法、压缩-拉伸法及移动窗口压缩-拉伸法四种相对波速测量方法进行对比分析,最终选取了误差较小、可靠性较高的移动窗互谱法计算相对地震波速度变化。应用上述数据处理方法得到汶川地震震源区及周边上百个台站对的相对波速变化结果,根据台站对间距及经验格林函数的信噪比和钟差等参数对这些结果进行筛选,最后深入分析了41个台站对的相对波速变化。结果表明,2008年5月12日汶川Ms8.0级地震造成了震源区地震波速度的急剧降低,最大降幅达0.4%;从西南至东北300多公里的余震带内,波速变化呈现空间分段特征,大致以安县为界,余震带西南部地区在汶川主震后波速降即达到最大值,而东北部地区的最大波速降一般出现在主震后的1-4个月,相对地震波速度变化的这种分段特性与地震序列的时空分布特征有较好的对应关系;在震源区外围的四川盆地也观测到了震后波速降低,而川西高原内部则没有出现明显的波速变化。为探究汶川地震前较长时段该区地震波速度变化的“背景”,将所用资料的时间范围向前延伸到2001年,发现在汶川地震发生前的几年时间里各台站对的波速变化一直比较平稳,没有明显的线性或周期性的变化。本文还对检测到的地震波速度变化的物理机制进行了分析。利用背景噪声互相关方法得到的波速变化主要反映了台站对间一定的三维空间内地壳介质状态及属性的变化,尽管还不能确定这种变化的准确空间位置及引起变化的物理机制,但通过分析计算并结合其它地质与地球物理资料,我们认为强地面运动引起的地表破坏和地震的静态应力变化都不能很好地解释本文得到的地震波速度变化,这种变化可能反映了一定深度范围内断层区内部的结构破坏和断层周围的应力变化。更多还原

【Abstract】 On 12 May 2008, an Ms 8.0 earthquake, causing heavy casualties and economic losses,occurred in Wenchuan County, Sichuan Province, China. It was the most destructive earthquake since the founding of the PRC and the deadliest one to hit China since the 1976 Tangshan earthquake. This great earthquake occurred along the Longmenshan thrust belt, which is at the eastern margin of the Tibetan Plateau and adjacent to the Sichuan basin. It ruptured over a length of 300km along the NE direction and generated 240-km and 72-km surface rupture zone on the Longmenshan central fault and front-range fault, respectively. There was significantly large range of co-seismic deformation around the Longmenshan fault zone. In the meantime, the physical properties of subsurface material around the seisgenic zone also would be changed caused by the great earthquake. Studying these changes has important significance for understanding the mechanism of the Wenchuan earthquake and the evolution of fault zone and analyzing the seismic hazard. In this thesis, we applied ambient seismic noise correlation, combining with the coda wave interferometry technique, to detect the co- and post-seismic variations of the crustal velocity around the Wenchuan earthquake fault zone.The research of the Earth’s interior structure and its temporal variation based on the ambient seismic noise is one of the hottest topics in seismology recently. The interaction of the ocean and atmosphere with the solid Earth creates the ambient noise field throughout the Earth. Computing the cross-correlations of ambient noise at two stations can obtain the interstation empirical Green function (EGF), which can be used to invert the velocity structure. The waveforms of EGF in different period times maintain high degree of similarity, which makes it possible to detect the temporal variations of seismic velocity of the medium using coda wave interferometry. Furthermore, the continuous monitoring of velocity changes can be achieved due to the continuity of the available ambient noise.In this study, we used the vertical component seismograms recorded by seismic stations in the Sichuan digital seismic network during the year 2001 - 2009. The data processing procedure was divided into three phases: (1) single station data preprocess to obtain the ambient noise;(2) cross-correlation and temporal stack to extract the inter-station empirical Green function;(3) estimation of relative velocity change by measuring travel time shifts between the daily and the reference empirical Green function. The key procedure in single station data preprocess was time domain normalization. After comparing some existing normalization method, we made some improvement of running absolute mean normalization, from which we achieved better results. We did a lot of trials and comparative analyses of the techniques (moving-window cross-correlation technique, moving-window cross-spectrum technique, stretching technique and moving-window stretching technique) and parameters for estimating temporal change of seismic velocity. Finally, we chose the moving-window cross-spectrum technique to measure the velocity changes in this study because it can give a stable estimate of velocity changes and own small computing error.Applying the above data processing procedure, we obtained more than 100 station pair results of relative seismic velocity changes around the Wenchuan earthquake fault zone in frequency 0.1-0.5Hz. We selected the results based on the distance of station pairs and the SNR and clock error of EGF waveforms. Then 41 results were selected for further analysis. The results revealed a sudden post-seismic velocity drop for the station pairs across the Longmenshan fault, the largest of which was more than 0.4%. The seismic velocity change with time exhibited a spatial difference: in the southwest segment of the Longmenshan fault, the maximum value of velocity reduction occurred at the time immediately after the Wenchuan earthquake, whereas in the northeast segment of the fault velocity reduction appeared 1-4 months after the main earthquake, which was corresponding to the spatial-temporal distribution of the aftershocks. Furthermore, the spatial extent and magnitude of the post-seismic velocity change in the Sichuan Basin exceeded that in the mountainous regions west of the fault zone. In order to explore the background of seismic velocity change in this area before the Wenchuan earthquake, we processed the data from January 2001 and found that the velocity change in every station pair was always stable and no obvious linear or periodic trends before the Wenchuan earthquake.We discussed the physical mechanism of seismic velocity change observed above. The velocity change calculated by ambient seismic noise correlation reflected the changes of the states and properties of 3D crustal medium around station pair. We analyzed the results combined with other geological and geophysical observation and found that the static stress change and near-surface physical damage caused by strong ground motion could not entirely explain the measurements in this study. We inferred the temporal changes of seismic velocity maybe related to the damages from faults rupture and stress changes around the fault zones.更多还原