【作者】 郭飚；

【导师】 刘启元； 赵大鹏；

【作者基本信息】 中国地震局地质研究所 ， 固体地球物理学， 2009， 博士

【摘要】 大陆动力学已经成为当今固体地球物理各领域研究的主导方向。大陆动力学涉及问题非常广泛,但核心问题是大陆变形及其动力学。地震各向异性是地球动力学过程的指示器,根据地震各向异性的研究结果,可以推断上地幔物质的流动或变形,有助于了解地球内部的应力状态和地球的演化过程。地震层析成像是研究地球地壳上地幔结构的主要地球物理方法。地震观测技术的进步、数据的积累以及计算技术的发展使得利用体波走时数据反演各向异性参数成为可能。利用地震走时层析成像方法同时研究介质的非均匀性和各向异性,对于认识地球的结构及动力学过程都具有非常重要的意义。本文系统地讨论了弱各向异性条件下P波速度扰动的近似表达式,并在此基础上构建了地震各向异性P波走时层析成像算法。利用多个不同理论模型对本文提出的方法进行了数值检验,系统地讨论了该方法的应用条件,并将本文提出的理论和方法成功地应用在中国境内天山地区和川西龙门山地区,获得了上述地区地壳上地幔的P波速度结构和P波各向异性参数。本论文的主要研究成果主要包括以下四个部分:一、理论构建在Backus(197D)的弹性张量调和展开理论基础上,导出了弱各向异性介质中的P波相速度扰动调和展开的的各阶系数,校正了Smith & Dahlen(1973)和Bokelmann(2002)文献中的错误。根据P波速度扰动的调和展开公式,导出了在不同观测系统和不同各向异性介质情况下的简化公式。根据Bond旋转矩阵证明了绕坐标轴旋转时弹性矩阵元素变化规律,证明了在正交对称系垂直对称轴情况下,仅需4个参数即可模拟方位各向异性,而在任意各向异性情况下,仅需6个参数即可描述三维各向异性。二、技术实现基于弱各向异性近似条件下弹性张量的调和展开理论和地震走时层析成像方法,构建了地震各向异性P波走时层析成像算法。该算法利用均匀的网格节点和三次B样条插值函数描述各向同性速度场:利用各向异性块描述各向异性速度扰动场。这种模型参数化方法的优点在于:a)模型光滑,在反演时不需要加入光滑项:b)各向同性和各向异性的分辨率可以分别考虑,以减少介质的非均匀性与各向异性间的耦合。另外,总结对比了地震波走时计算的各种方法,构建了各向同性介质中快速行进波前扩展程序,并在此基础上构建了适合弱各向异性介质的各向异性快速行进算法,利用互易性定理减少了相应的走时计算的计算量。在该算法中,首先计算以台站位置为假想震源的走时场,再计算台站与近震震源或远震射线与模型边界的交点间的射线路径。这种走时计算方式特别适用于多源多台站的观测方式,可有效地减少正演计算量。反演方法采用,Tarantola(1982,1987)的非线性反演方法,对于非均匀和各向异性参数可以分别设置不同的阻尼因子。三、数值检验本文给出的多个理论模型测试结果表明:a)本文给出的各向同性和弱各向异性条件下的快速行进算法的计算精度可以满足地震走时层析成像的要求(误差小于震相的检测误差):b)各向异性的强度与非均匀结构是耦合的,而各向异性方向与非均匀结构是可以解耦的;c)在数据分布较好的情况下,P波走时方法可以分辨多层各向异性;d)对于远震,只要数据的方位覆盖大于180°即可较好的反演各向异性参数;e)三维各向异性的确定需要大于50°的震中距覆盖。四、实际数据解释1,利用横跨中国境内天山的库车—奎屯宽频带流动地震台阵和区域地震台网记录的近震和远震P波走时数据和Zhao et al(1992,1994)地震层析成像方法,重建了沿该地震台阵剖面400 km深度范围内地壳上地幔的各向同性P波速度结构。在此基础上,利用弱各向异性介质条件下的P波速度扰动公式和线性反演方法,进一步解释了剩余残差,获得了该地震剖面上的P波快波方向。我们的结果表明:a)沿新疆库车—奎屯剖面,天山地壳具有明显的横向分块结构,且南、北天山地壳显示了较为强烈的横向变形特征,表明塔里木地块对天山地壳具有强烈的侧向挤压作用。在塔里木和准噶尔地块上地幔顶部有厚度约60～90 km的高速异常体.塔里木—南天山下方的高速异常体产生了较为明显的弯曲变形,而准噶尔—北天山下方的高速异常体向南一直俯冲到中天山南侧边界下方300 km的深度。两者形成了不对称对冲构造.在塔里木和准噶尔地块下方150～400 km深度存在上地幔低速体。其中,塔里木地块一侧的上地幔低速体上涌到南天山地块的下方。在塔里木—南天山200～300 km深度范围的上地幔存在高速异常体,它可能是地幔热物质向上迁移过程融断的塔里木岩石圈拆离体。b)塔里木地块的俯冲可能涉及整个岩石圈深度,但其前缘仅限于南天山的北缘;青藏高原隆升的远程效应可能驱动了塔里木岩石圈向北俯冲,同时还造成天山造山带南侧上地幔物质的涌入。天山造山带上地幔广泛存在的低速异常有助于上地幔的变形。上地幔物质的强烈非均匀性应有助于推动天山造山带上地幔范围小尺度地幔对流的形成.根据研究区地壳上地幔速度结构特征可以推断,新近纪以来,天山快速隆升的主要力源来自青藏高原快速隆升的远程效应,相对软弱的上地幔为加速天山造山带的变形和隆升创造了必要条件。c)塔里木盆地和南天山的P波快波方向为近南北向,这与塔里木地块对天山南北向挤压变形作用是一致的。进人中天山褶皱带和北天山,P波快波方向逐渐偏转为北西向,北天山山前褶皱带和准噶尔盆地前缘的P波快波方向为近东西向,总体上,该区P波快波轴方向平行于天山造山带的走向,垂直于板块相对运动方向。塔里木盆地和南天山的壳幔为连续一致变形,而中天山、北天山和准噶尔上地幔的各向异性方向可能受到地幔小尺度对流和构造方向的影响,壳幔的形变发生解耦。这表明,由于两侧盆地的挤压,天山山体处于挤压状态,而上地幔物质的运移沿着山体平行的方向,两者之间构成了立交模型。根据研究区地壳上地幔速度结构特征和各向异性方向推断,上地幔物质的流动变形在天山造山带的变形和隆升过程中起着重要住用。2,利用在龙门山及邻区布设的川西地震台阵记录到的远震P波走时数据和本文提出的各向异性层析成像方法,同时反演了该区地壳上地幔远震P波各向同性速度结构和P波快波方向。我们的结果表明:a)研究区地壳上地幔P波速度结构具有较为明显的分区特征,松潘-甘孜地块和川滇地块速度较低,龙门山断裂断裂带及四川盆地西部速度较高。四川盆地西部地壳上地幔的高速异常厚度从南侧250 km向北逐渐减薄至100 km,推测这个高速异常体可能代表四川盆地的岩石圈。相对于四川盆地,松潘—甘孜地块的地壳上地幔较为软弱,但不存在四川盆地向松潘.甘孜地块的俯冲。松潘-甘孜地块的抬升可能与地幔上涌有关。四川盆地与松潘-甘孜地块和川滇地块间的动力学机制完全不同,川滇地块和四川盆地仅是垂直接触关系,而在四川盆地北部靠近龙门山一侧,发现四川盆地前缘自东向西减薄的现象,推测松潘-甘孜的上地幔物质侵蚀了四川盆地下方的岩石圈。鲜水河断裂带和龙门山断裂带都为超壳的深大断裂,鲜水河断裂带在深部结构上显示为较宽的低速特征.b)龙门山断裂带与四川盆地的地壳均显示为高速特征,四川盆地与松潘-甘孜地块之间的构造边界可能是汶川-茂县断裂。以汶川为界,龙门山断裂带被从松潘—甘孜地块侵入的低速物质分为两段,南北两段均显示为高速特征。汶川大地震及余震序列均发生在龙门山北段的高速介质区域内。这种深部结构特点对汶川大地震孕育和余震序列发育过程具有明显的控制作用。根据本文给出的松潘.甘孜地壳上地幔速度结构。我们可以进一步推断,松潘-甘孜地块的抬升应与地幔物质上涌有关,在坚硬的四川盆地的阻挡下作用下,青藏高原向东挤压和地幔上涌的双重作用下造成松潘—甘孜地块隆升,并造成了了龙门山断裂带的逆冲推覆。龙门山南北两段间的地壳强度较低,在长期的缓慢变形过程中,易于在龙门山北段的坚硬的上地壳内形成了巨大的应力积累,而其汶川处于龙门山北段的最南端,应力容易在此集中,并成为汶川大地震破裂的起始点。c)P波各向异性研究表明川滇地块的上地幔流动方向与GPS的观测结果相吻合。反映了川滇地块地壳和上地幔的形变一致,属于壳幔耦合型。松潘—甘孜地块的上地幔流动方向为北东向,而GPS的观测结果为近东西向。说明松潘甘孜地块壳幔的形变不一致。属于壳幔解耦型。四川盆地北部的P波快波方向与SKS的结果较为一致,而四川盆地南部的P波快波方向平行于主要构造方向。与SKS的结果相互垂直。四川盆地内部南北两部分具有不同的壳幔变形特征,这与速度结构给出的结论相一致。d)本文的研究结果并不支持四川盆地向松潘—甘孜地块的俯冲的动力学模型。与其结果相反,上地幔的物质流动在该区的演化变形过程中起主导作用,在上地幔深度上,松潘—甘孜地块的上地幔物质可能侵入四川盆地下方。更多还原

【Abstract】 Continental dynamics has become one of the frontier fields in present earth sciences. Continental dynamics involved in a wide range of issues, while the key problems of continental dynamics are the tectonic deformation and its mechanisms. Seismic anisotropy is a useful indicator of geodynamic processes. Results of seismic anisotropy can be used to infer the mode of the upper mantle convection, and help us understand the stress regime and evolution of the Earth.Seismic tomography is the major method for studying the three-dimensional (3-D) velocity structure of the Earth, especially the structure of the crust and upper mantle. With the progress of seismic observational techniques, data accumulation as well as the development of computing technology, it is possible to invert anisotropic parameters using travel time data. It has great significance using seismic tomography methods to simultaneously invert for both isotropic and anisotropic structures for the understanding of Earth’s structure and dynamic processes.In this work, in light of some problems in the theoretical and applied investigation of anisotropic tomography method, the approximation expressions of qP velocity perturbation under the condition of weak anisotropy have been discussed systematically, and further the anisotropic P-wave travel-time tomography algorithm has been developed. The method was numerically tested by many different theoretical models, and the application conditions of the method were discussed. Finally, the method was successfully applied to study the Chinese Tianshan orogenic belt and Longmenshan area.This dissertation consists of the following four parts:ⅠTheory:Backus (1970) has proved that any fourth-order tensor can be uniquely represented as a linear combination of 21 canonical harmonic tensors. Based on Backus’s theory, I have reproduced the expression of Cartesian components of the canonical harmonic tensors and the coefficients of qP velocity perturbation expansion, and corrected the misprints in the literature of Smith & Dahlen (1973) and Bokelmann (2002). According to qP velocity of the harmonic perturbation expansion formula, the reduction formula for different observation systems and different anisotropic media is derived. Based on the Bond rotation theory, the change rule of elements of elastic tensor is proved. It proves that the azimuthal anisotropy of orthogonal symmetry with vertical axis can be described by only 4 parameters. For any anisotropy case, only six parameters are sufficient to describe the three-dimensional anisotropy.ⅡComputational algorithmBased on the theory of harmonic decomposition of elastic tensor under the conditions of weak anisotropy and the seismic tomography method, the anisotropic seismic-tomography algorithm is developed. The algorithm uses a uniform grid nodes and cubic B-spline interpolation function to describe the isotropic velocity field, and uses blocks to describe the anisotropic velocity perturbation field. This kind of velocity parameterization method has several advantages: a) The velocity model is smooth, without adding an additional smooth term in the inversion; b) The resolutions of the isotropic and anisotropic structures can be considered independently, which can reduce the coupling between effects of heterogeneity and anisotropy. By comparison of the various methods of seismic travel time calculation, I construct a fast marching method in isotropic media and anisotropic fast marching method for a weakly anisotropic media. The travel-time calculation algorithm adopts the reciprocity theorem to reduce the cost of forward calculation. First of all, the travel-time field was calculated by taking the locations of seismic stations as fictitious sources, and then by calculating the ray path between stations and local earthquakes or intersection between the teleseismic ray path and model boundary. Such a calculation procedure can significantly reduce the cost of forward calculation, especially for the observation system of multi-sources and multi-receivers. In inversion, the Tarantola (1982, 1987) non-linear inversion method was adopted, and the damping factor for heterogeneous and anisotropic model parameters can be set up separately.ⅢNumerical testsOur numerical tests for different theoretical models show the following results. a) The calculation accuracy of the isotropic and weak anisotropic fast marching algorithm proposed in this work meets the requirements of the seismic travel-time tomography (the calculation error is less than the data picking error). b) The magnitudes of anisotropy and heterogeneity are coupled. c) The direction of anisotropy and the heterogeneous structure can be decoupled. d) The P-wave travel time method can invert for multi-layered anisotropy when the travel-time data have good azimuth coverage. e) For the teleseismic inversion, the anisotropy parameters can be well inverted with azimuth coverage more then 180°. f) The inversion for the direction of anisotropy in the three-dimensional media needs an epicenter distance coverage great than 50°.ⅣApplications1. Using the P-wave travel time data recorded by the passive seismic array across the Chinese Tianshan and regional seismic network, P-wave velocity structure of the crust and upper mantle down to 400 km depth along the Kuqa-Kuytun profile is determined by using the seismic tomography technique of Zhao et al. (1992, 1994). Based on the obtained isotropic structure, we further explain the remaining travel time residuals and obtain the fast P-wave direction along the profile using the P-wave velocity perturbation formula in weak anisotropic media and the linear inversion method. The results show the following features.a) Along the Kuqa-Kuytun profile, the crust of the Chinese Tianshan has an obviously blocked structure, and the crust beneath the south and north Tianshan is subjected to an obvious lateral distortion. This manifests that the Tarim basin exerts the strong lateral compression on the Tianshan crust. There are high-velocity anomalies with thickness of 60-90 km on the top of the upper-mantle beneath the Tarim and Junggar basin. The high-velocity anomaly beneath the Tarim and south Tianshan is obviously distorted. The high-velocity anomaly beneath the Junggar and North Tianshan has thrusted down to the depth of 300 km underneath the south side of the central Tianshan. Both of them form an asymmetric bilateral thrust. In the depth range of 150-400 km beneath the Tarim and Junggar basin, there exist low-velocity anomalies. Among them, the low-velocity anomaly beneath the Tarim block upwells below the south Tianshan. At the depth of 200-300 km beneath the Tarim and South Tianshan, there exists a high-velocity anomaly, which could be the lithospheric detachment of the Tarim caused by the upwelling of the upper-mantle’s hot material.b) The subduction of the Tarim block could involve the whole lithosphere, but its front is limited merely to the northern border of the South Tianshan. The far-field effect caused by the Tibet uplift could not only drive the subduction of the Tarim block, but also cause the upwelling of the upper mantle beneath the southern side of the Tianshan. The widespread low-velocity anomalies within the Tianshan upper mantle should facilitate the mantle deformation. The strong heterogeneity in the upper-mantle should be in favor of the small-scale mantle convection. In view of the velocity structure of the crust and upper mantle beneath the Tianshan, it could be inferred that the rapid uplift of the Tianshan since the Neogene is caused mainly by the far-field effect due to the uplift of the Tibetan plateau, and the relatively weak upper mantle beneath the Tianshan provides an essential condition for prompting the uplift and deformation of the Tianshan orogen.c) The fast P-wave direction is approximately in north-south in the Tarim Basin and South Tianshan, which accords with the Tarim block extrusion to the Tianshan. In the central Tianshan fold belt and North Tianshan, the fast P-wave direction turns gradually to NW-SE, and it is nearly in east-west in the North Tianshan fold belt and the front edge of the Junggar Basin. On the whole, the P-wave fast axis is parallel to the strike of the Tianshan orogenic belt, and perpendicular to the direction of relative movement of the plate. The crust and upper mantle have vertically coherence deformation in the Tarim Basin and South Tianshan, and decoupling in the North Tianshan and Junggar Basin because of the small-scale convection. These results suggest that the Tianshan Mountains was extruded by the two basins, the upper mantle materials have migrated along the direction parallel to the mountain, which can be described by a dynamic model something like the overpass. Based on the crust and upper mantle velocity structure and anisotropy directions, we infer that the upper mantle convection plays an important role in the process of the deformation and uplift in the Tianshan orogenic belt.2. The 3-D P-wave velocity structure of the crust and upper mantle down to 400 km depth and fast P-wave directions in the upper mantle are determined by applying our anisotropic tomography inversion technique to teleseismic travel time data recorded by the West Sichuan Seismic Array deployed in the Longmenshan area. The results show the following features.a) The crustal structure of the study area correlates with the surface geological features. The Sichuan Basin is imaged as a high-velocity feature, while the Songpan-Ganzi and Chuandian blocks are imaged as low-velocity features. The lithospheric thickness of the Sichuan basin with high-velocity has lateral variations from 250 km in south to 100 km in north. This high-velocity anomaly may represent the Sichuan Basin’s lithosphere. Relative to the Sichuan Basin, the crust and upper-mantle of the Songpan-Ganze block are relatively weak, but the velocity structure shows no evidence for the subduction of the Sichuan basin downward and channel flow. The uplift of the Songpan-Ganze block may be related to the mantle upwelling. The Sichuan basin vertically contacts with the Chuandian block, but the Sichuan basin fronts in the Longmenshan area become thinner from east to west with the feature of the Songpan-Ganzi block incursion into Sichuan basin in the upper-mantle. This feature shows different dynamic mechanisms between the Sichuan basin with the Chuandian block and the Sichuan basin with the Songpan-Ganzi block. The Xianshuihe and Longmenshan faults all cut through the whole crust. The Xianshuihe fault belt exhibits as a broad low-velocity zone.b) The crust of the Longmenshan fault belt and the Sichuan Basin shows high-velocity characteristics, and the boundary between the Sichuan Basin and Songpan-Ganze block may be the Wenchuan-Maoxian fault. The Longmen Shan fault belt was divided into two parts by the extrusion of low-velocity from the Songpan-Ganzi block bounded by Wenchuan: the southern part and the northern part. The two parts are all characterized by high-velocity. The great Wenchuan earthquake (Ms 8.0) and its aftershocks are all distributed in the high-velocity zone of the northern part of the Longmen Shan fault belt. This feature of deep structure may have controlled the occurrence of the Wenchuan mainshock and the distribution of its aftershocks. Based on the crustal structure of the Songpan-Ganzi block, we can infer that the uplift of the Songpan-Ganzi block and thrust nappe of the Longmenshan are a result of the eastward extrusion of the Tibetan Plateau and upper-mantle upwelling. The strength of rock between the two parts of Longmenshan is weak. During the process of long-term deformation, it is possible to accumulate high stress in the rigid crust of the northern part of Lomgmenshan. Wenchuan is located at the southern end of the northern part of Longmenshan, where large stress is easy to build up. These factors may be the reason why the rupture of the Ms 8.0 earthquake initiated there.c) P-wave anisotropy results show that the directions of upper-mantle flow in the Sichuan-Yunnan block are consistent with the GPS results, which means that the crust and upper mantle deformation of the Sichuan-Yunnan block is coupled. The flow direction of the Songpan-Ganze block’s upper mantle is in north-east direction, and the GPS results show a nearly east-west direction, suggesting that the deformation of the crust and upper-mantle in the Songpan-Ganze block is decoupled. The fast P-wave direction in the northern Sichuan Basin is accordance with the GPS results, but it is parallel to the tectonic line and perpendicular with SKS results in the southern part of the Suchuan Basin. Therefore there are different deformation modes of the crust and upper-mantle in the northern and southern parts of the Sichuan Basin, which are consistent with the velocity structure we determined.d) The results of this work do not support the dynamic model of subduction of the Sichuan Basin beneath the Songpan-Ganze block. In contrast, the flow of upper-mantle plays an important role in the evolution process of this area. In the upper-mantle, the material of the Songpan-Ganzi block may have invaded into the upper-mantle of the Sichuan Basin.更多还原