【作者】 王阎昭；

【导师】 沈正康；

【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】 大陆内部形变及其动力学机制是地学领域的重要研究方向和热点问题,多年来存在很多争论。大陆构造形变场模型研究对于大陆内部形变及其动力学机制的认识具有重要作用。随着对大陆形变机制认识的不断深化和观测数据的增加,大陆构造形变场模型研究得到了长足的进展;与此同时,大地测量技术特别是GPS技术的快速发展,不仅为大陆形变场模型研究提供了难得的机遇,也提出了更高的要求,暴露出了很多问题,主要表现在以下三个方面:首先,随着观测技术的迅速发展和观测数据的急剧增加,人们对地壳结构和地壳形变的认识越来越精细,地壳的分维结构特征也开始显现出来,如何在有限的数据约束下在模型中体现出地壳结构的分维特征?其次,随着模型的改进,更多参量被引入模型当中,但实际观测数据有限,如何在模型参量和数据量之间寻求平衡,在有限的数据约束下更好地模拟地壳形变?再次,许多模型在建立过程中虽然参考了地质、地球物理等方面的资料,但仍然带有一定主观性,缺乏客观的判断标准,如何让整个研究过程更客观合理?针对上述问题,开展了有关大陆构造形变场模型的研究工作,并将两种模型分别应用于青藏高原东北缘和东南缘,获得对大陆地壳形变模式及其动力学机制的认识。1形变场模型研究分别从两种最基本的构造形变场模型——刚性块体运动模型和断裂位错模型出发,发展可形变块体模型和连接断层元模型。1)可形变块体模型可形变块体模型的基本假设是:一、活动块体在发生相对运动的同时,内部还可能发生形变,圉于有限的数据,模型中假设若块体发生形变则为均匀形变;二、假设断裂带由上部脆性层和下部蠕滑层组成,上部脆性层在间震期处于锁定状态,下部蠕滑层的错动量则由块体间相对运动引起块体内部形变所决定。模型的具体内容分为块体划分、参量反演和应变能计算三部分。块体划分的方法是:依据GPS速度场和活动断裂以及其他地质地球物理资料,对研究区的活动块体做初步划分。用F检验的方法检验块体内部台站运动的奇异性,剔除奇异点或对块体边界的位置进行调整。用刚性块体模型反演块体运动的欧拉极,同时得到模型拟合后残差。用F检验的方法检验相邻块体的独立性,对独立性低于设定阈值的块体进行合并;重复这一过程,直至所有相邻块体的独立性均高于设定阈值。参量反演的方法是:以块体边界一定范围外的台站速率为约束,反演块体的运动变形参量;利用反演结果计算块体边界上的错动速率,用弹性位错模型估算块体边界锁定对速度场的影响,并对所有台站速率做改正;用改正后的速度场反演块体运动变形参量;重复上述过程,直至模型拟合后的残差达到最小值。在保持其他块体运动参量个数不变的情况下,将块体的参量个数逐个由3个增加到6个,即允许块体内部发生均匀形变,用F检验的方法比较某一块体运动参量增加前后模型拟合后残差,若均匀形变显著性超过设定阈值,则在后面的反演过程中始终保持该块体参量个数为6个;重复以上过程,直至所有块体参量都不再增加。应变能的计算方法是:发展块体内部应变率矩阵分解模型,将可形变块体模型反演得到的块体内部应变率矩阵进行分解,得到分别对应于两个断裂面走向平行的位错源的应变率矩阵,用于计算块体内部“等效地震矩积累率”;用可形变块体模型反演得到的块体边界上的错动速率计算块体边界上的“等效地震矩积累率”;用地震目录的震级资料计算块体边界带和块体内部地震释放的能量。2)连接断层元模型连接断层元模型的基本假设是:一、假设所有断层的倾角均为90°,断裂带由上部脆性层和下部蠕滑层组成,上部脆性层在间震期处于锁定状态;二、被断裂围限区域的地壳运动具有一定整体性,同时断裂交接部位可能发生局部形变。连接断层元模型的基本原理是:通过弹性位错模型建立GPS速度场与断裂错动速率之间的定量关系;通过误差方程的形式对断层的错动量施加约束,对断层错动的走滑分量施加连续性约束条件,对断层错动的挤压/拉张分量施加限制性条件;运用最小二乘方法通过GPS速度场反演断裂错动速率。其中对断层走滑速率施加的两种极端约束条件分别对应两种极端模型:当施加严格约束条件时,对应于块体模型;当不施加任何约束条件时则对应于断裂模型。通过施加适当的约束条件,使得模型在保证一定块体整体运动协调性的同时,允许相邻断裂段接合处发生局部形变,从而能够更合理地模拟地表形变场。对挤压/拉张分量施加的限制性条件主要是为了防止由于数据约束不足造成的断裂挤压/拉张速率被过高估计。具体的做法是:依据已有活动断裂和相关地质地球物理研究成果,初步建立断层模型,并用连接断层元方法反演断层的滑动速率,拟合GPS速度场;通过模型拟合结果和实际观测数据的比较,调整断层模型和约束条件;重复上述过程,在参量解析度与拟合后残差的折中关系中寻得最佳值,最终得到有限参量而与实际观测数据相吻合的最佳模型。2形变场模型在青藏高原东缘的应用青藏高原及其周边地区是目前大陆内部变形最强烈、地震活动性最强的地区。其构造演化机制一直是大陆动力学研究的热点和前沿。青藏高原东缘地区是高原隆升发展的前沿,是正在形成中的高原,是研究大陆内部变形和高原隆升机制的理想地点。这里地震活动频繁,是中国大陆内部最强烈的地震带,控制着一系列历史强震的发生,是研究地震孕育发生机制的天然试验场。对这一地区的地壳应变分配方式和断裂活动性的定量研究将为地震危险性评估提供数据支持,进而为地震预报服务,具有十分重要的理论价值和现实意义。这里又是中国大陆内部GPS台站分布密度最高的地区之一,能够为模型提供较好的约束,也便于对研究区内相对精细的结构开展研究。前人已经在这一地区开展了大量的研究工作,不仅能够为模型的建立提供更丰富的信息,而且也便于结果的比较和解释。依据青藏高原东北缘GPS速度场和前人在这一地区的研究成果,将研究区划分为16个活动块体并分析其运动特征。以青藏高原东北缘地区(90°~110°E, 28°~42°N)GPS速度场为约束,采用可形变块体模型,同时考虑边界锁定效应,反演块体运动变形参量的同时对块体边界上的错动速率做出估计。采用F检验的方法剔除奇异点、检验模型中相邻块体的独立性、块体内部均匀形变的显著性以及块体边界的活动性。F检验的结果显示10个块体内部形变的显著性超过95%。阿拉善、鄂尔多斯和民勤块体平动和转动速率都很小,内部形变不显著或应变率值较低,是研究区内相对稳定的块体,对印藏碰撞造成的北东向推挤起阻挡作用。祁连山和海原块体所形成的条带状区域发生强烈挤压变形,对其南侧块体的北东向运动起到缓冲作用。被海原、东昆仑和龙门山断裂所围限的区域内块体在整体向北东方向运动的同时,由北西向南东块体平动方向发生顺时针偏转,同时块体绕自身几何中心做顺时针旋转,块体内应变率值由西向东逐渐减弱。青藏高原东北缘地区的断裂按其走向和性质主要分为三组:北西西向断裂和甘孜-玉树-鲜水河断裂主要为左旋挤压性质,具有较高的滑动速率,对该地区的地壳运动起控制作用;北北西-近南北向断裂和北东向断裂主要为右旋走滑性质,滑动速率相对北西西向断裂较弱,对该地区地壳活动起调节作用。北东-北北东走向的断裂则既有左旋性质也有右旋性质。用可形变块体模型反演得到北西西向祁连山北缘、党河南山、香山-天景山、海原、青海南山、西秦岭北缘和东昆仑断裂以及甘孜-玉树-鲜水河断裂的左旋走滑速率分别为2.3±0.3、2.3±0.7、3.3±0.5、5.9±0.4、0.9±0.6、1.4±0.3、1.5~13.4和12.2~13.5 mm/a;北北西-近南北向鄂拉山、庄浪河和岷江断裂的右旋走滑速率分别为2.7±0.9、0.2±0.4和2.9±0.8 mm/a;北东-北北东向阿尔金和狼山山前断裂的左旋走滑速率为2.9±0.4和1.4~2.3 mm/a,银川、龙门山和龙日坝断裂的右旋走滑速率分别为3.3±0.3、0.6~1.6和6.1±0.9 mm/a。与此同时,采用剖面投影方法计算了主要断裂的错动速率,并用F检验的方法检验了各条断裂的活动性,所得结果与用可形变块体模型所得结果基本一致。另外,研究区的两条速度阶跃带金昌-民乐和玛曲-洛须速度阶跃带的活动性分别达到100%和98.8%,由可形变块体模型反演得到前者的左旋走滑速率为4.3±0.3 mm/a,后者的右旋走滑速率为3.2±1.1 mm/a。利用可形变块体模型的反演结果估算块体内部和边界“等效地震矩积累率”的比值约为0.31,利用地震目录估算块体内部和边界地震能量释放的比值约为0.36,二者相当吻合。以GPS数据给出的川滇地区(96°~108°E, 21°~35°N)速度场为约束,依据研究区已知断裂分布情况建立断层模型,用最小二乘方法反演了该地区主要活动断层的现今错动速率。结果显示,印藏碰撞引起的北北东向推挤和高原隆升引起的重力势能作用造成青藏高原物质东向挤出。遇到来自稳定华南块体的阻挡后,高原东南部物质相对稳定欧亚板块转向南东方向继而向南运动,使得川滇地区围绕喜马拉雅东构造结作顺时针转动,造成川滇地块东侧断裂作左旋走滑活动,而其西侧断裂以右旋走滑活动为主。其中甘孜-玉树、鲜水河、安宁河、则木河、大凉山、小江断裂及其向南西方向延伸的部分和打洛-景洪、湄沾断裂构成青藏高原东南部东向挤出的东北边界和东边界,左旋速率分别为0.3~14.7、8.9~17.1、5.1±2.5、2.8±2.3、7.1±2.1、9.4±1.2、10.1±2.0、7.3±2.6和4.9±3.0 mm/a。青藏高原东南部东向挤出的西南边界似乎不是由单一断裂带构成,而是在较宽范围内形成的一条右旋剪切带。位于红河断裂北东侧的南华-楚雄-建水断裂和西南侧的无量山、龙陵-澜沧断裂活动性较强,分别具有4.2±1.3、4.3±1.1和8.5±1.7 mm/a的右旋走滑活动。但金沙江断裂目前基本不活动,红河断裂的活动性不强。龙门山一带在汶川地震发生前地壳活动较弱,龙门山断裂宝兴-北川段和北川-青川段缩短速率分别为1.4±1.0和1.6±1.3 mm/a,而龙门山断裂西北方向的龙日坝断裂有5.1±1.2 mm/a的右旋走滑分量。川滇菱形块体内部的一些断裂表现出较强的活动性,其中理塘断裂左旋走滑速率为4.4±1.3 mm/a,拉张速率2.7±1.1 mm/a;玉农希断裂及其周边地区右旋剪切形变速率为2.7±2.3 mm/a,地壳缩短速率6.7±2.3 mm/a。丽江-小金河断裂中段活动性强于北段和南段,达到左旋走滑5.4±1.2 mm/a,拉张0.5±1.0 mm/a。与此同时,讨论了不同断裂锁定深度对结果的影响,并得到鲜水河断裂的锁定深度为15 km,70%置信区间为11~19 km。3对大陆形变模式及其动力学机制的认识连接断层元模型在青藏高原东南缘地区的研究结果显示,该地区存在多条错动速率非常有限的活动断裂,将地壳分割成多个相互运动的地块。可形变块体模型在青藏高原东北缘地区的研究结果显示,该地区地壳变形方式表现为一系列百公里尺度的小型块体的相对运动,块体内部和边界应变能的积累速率之比约为0.31:1,块体内部和边界上地震释放能量之比约为0.36:1。整个青藏高原东缘地区的地壳应变能积累主要集中在断裂带上,但活动块体内部的应变能也不容忽视。结合青藏高原东缘的深部结构和各向异性研究结果,认为青藏高原及其周边地区的地壳厚度自高原内部向外逐渐减薄,青藏高原东缘地区下地壳和(或)上地幔出现不连续的软弱带,上下地壳或地壳与地幔之间发生部分解耦,造成该地区地壳强度有所降低。在印藏碰撞造成的推挤作用下,以甘孜-玉树-鲜水河断裂为界,青藏高原东北缘地区的地壳物质同时受到北东向推挤作用和阿拉善、鄂尔多斯块体的阻挡,地壳或岩石圈被北西西和北北西向的两组断裂切割成若干小型块体,通过块体之间的相对运动和块体内部的变形实现地壳整体的顺时针旋转,地壳物质向南东方向的运移造成龙门山地区的应力积累,同时推挤着四川盆地一起向南东方向运动。甘孜-玉树-鲜水河断裂以南地区的地壳物质则在青藏高原内部物质东向挤出和重力势能的作用下向阻挡作用很弱的南部地区运动,造成东侧鲜水河-小江断裂带的左旋走滑活动和西侧宽阔的右旋剪切形变带。青藏高原东缘地区的构造形变场既不表现为“大陆逃逸”模式,也不表现为“连续形变”模式,而是介于两者之间。大陆构造形变场主要取决于岩石圈的流变学结构和岩石圈所受构造作用,当中下地壳或上地幔不存在软弱层,地壳和地幔的变形是耦合的,岩石圈具有较高的强度,在同等的构造应力作用下岩石圈不易发生破裂和变形,仅存在数量很少的切割整个岩石圈的断裂,因而能够被“大陆逃逸”假说较好的解释。当中下地壳或上地幔出现大范围连续软弱层,上下地壳或地壳与地幔发生解耦,地壳强度较低,在同等构造应力的作用下容易发生破裂和变形,断裂大多切割至软弱层所在的深度,能够被“连续形变”假说较好的解释。当中下地壳或上地幔存在不连续软弱层,上下地壳或地壳与地幔发生部分解耦,地壳形变模式则介于两者之间。更多还原

【Abstract】 Continental deformation and mechanism has been a focusing point of geoscience since the 1970s. Controversies yet still exist on how the continental deformation field can be best described and how the continental lithosphere is deformed. It is very important to develop continental deformation models for the understanding of mechanisms of continental deformation. Thanks to the growth of understanding on continental deformation mechanisms and accumulation of observations, deformation models have been greatly improved. Rapid development of geodetic technology, especially GPS technique, not only provides an opportunity for improving deformation models, but also raises more questions. New problems emerge, particularly as follows. First of all, along with the development of observational technology and mounting observations, more details on crustal structure and deformation pattern are discovered, and a fractal structure of crust starts to emerge. Therefore how to account for the fractal behavior in a model with only limited data constraints? Second, more and more parameters have been introduced along with the improvements of models, with yet still finite observations. How to balance between the amounts of parameter and data in order to best model the crustal deformation with limited observations? Third, geological and geophysical observations with various qualities are often used in deformation modeling, but the lack of objective criterions on their quality control makes the modeling process subjective. How to make the whole research process more objective and reasonable?In order to answer the above problems, continental deformation models have been developed. Two of such models have been applied to study the deformation fields of the northeastern and southeastern margins of Tibetan plateau, respectively, to better understand the deformation patterns and mechanisms of the regional continental crust.1 Research on deformation modelsA deformable block model and a linked fault patch model are developed, based on two basic deformation models, e.g. the block motion model and the fault dislocation model.1) Deformable block modelThe deformable block model is developed under two assumptions. First, besides relative motions between active blocks, internal deformation may also take place. Due to limitation of observation data, internal deformation is assumed to be uniform within each block, if there is any. Second, faults are composed of two parts, an upper brittle layer and a lower creeping layer. The brittle layer is locked during the interseismic period. The amount of dislocation of the creeping layer is determined by the relative motion between blocks and the associated internal deformation.Realization of the deformable block model consists of three parts: block differentiation, parameter inversion, and strain energy estimation.Blocks are differentiated through five steps. Step 1, the studied region is divided into several initial blocks based on deformation patterns in GPS velocity field, locations of active faults, and other geological and geophysical data. Step 2, the F-test is applied for testing outliers of station velocities inside the blocks, outlier stations are eliminated, and/or block boundaries are adjusted. Step 3, Euler pole of block motion is inverted for each block and the corresponding post-fit residuals are calculated, based on the assumption that there is no internal deformation for any block. Step 4, independence of adjacent blocks is tested by F-test; adjacent blocks whose independent test is not significant above a confidence threshold are combined into one. Step 5, steps 3 and 4 are iterated until independent tests for all the adjacent blocks are significant above the confidence threshold.Model parameters are inverted through four steps. Step 1, block kinematic parameters are inverted under constraints using station velocities within a block. Slip rates along block boundaries are estimated from the block kinematic result. Step 2, station velocities are corrected to remove deformation due to fault locking effect during the interseismic period. Step 3, kinematic parameters are inverted under constraints of corrected station velocities. Step 4, steps 2 and 3 are iterated until the post-fit residualχ2 reaches minimum. Furthermore, significance of internal deformation is tested. For each block its number of parameters is increased from 3 to 6 while keeping other block parameters fixed, and the F-test is performed to evaluate whether the reduction of data post-fit residualχ2 due to the increase of the number of block parameters is significant. If so, internal deformation within the block is retained and the number of parameters for the block is increased to 6. The above procedure is iterated until none of the numbers of block parameters are increased based on the F-test.The procedure of strain energy estimation is as follows. The inverted strain rate tensor for an individual block is decomposed into two, each corresponding to a dislocation source and used to calculate the equivalent seismic moment accumulation rate inside the block. The estimated slip rate across a fault patch is used to calculate the equivalent seismic moment accumulation rate along the corresponding block boundary. Seismic energies released by earthquakes inside and along boundaries of blocks are calculated using magnitude information of an earthquake catalog. 2) Linked fault patch modelThe linked fault patch model is developed based on following assumptions. First, all faults dip at 90°. Second, all faults consist of upper brittle layers and lower creeping layers. The upper brittle layer is locked during the interseismic period. Third, regions confined by faults move as a whole to a certain extent, with regional deformation occurring around the intersection points of linked fault patches.The basic theory of linked fault patch model is as follows. The quantitative relationship between GPS velocity field and fault slip rates can be obtained based on dislocation theory. Constraints are applied on fault slip rates in two ways: continuity constraints are applied on the strike-slip components, and amplitude constraints are imposed on the normal components, respectively. The inversion is performed using the least-squares method. Extreme constraints on the strike-slip components correspond to extreme end-member models. When the constraints are extremely strict, the linked fault patch model is equivalent to the block motion model. However, when no constraints are applied, the model is equivalent to the fault patch dislocation model. Appropriate constraints should be imposed on the model to balance between integrated motions of blocks and local deformation around intersection points of linked faults. Also, constraints on normal components of fault slip rates help reduce excessive estimation of these model parameters due to the lack of constraints from data.The initial fault model is constructed based on information of active faults and other geological and geophysical results. GPS velocity field is inverted for solution of the linked fault patch model. The fault model and constraints are adjusted according to the goodness of data fits to the model. The above process is iterated until a best fit to the model is obtained.2 Application of deformation models to tectonic deformation of the eastern margin of Tibetan plateauThe Tibetan plateau and its surrounding region is a region undergoing the most severe deformation and with the most intensive seismic activity in continent. Its tectonic evolution mechanism has been the cutting edge and focusing point of geodynamic research for a long time. The eastern margin is the growing foreland of the Tibetan plateau. Thus it is an ideal place for research on continental deformation and tectonic evolution of the plateau. Intensive seismicity here also makes it a natural laboratory for research on seismogenic processes and their associated physical mechanisms. Quantitative analyses of strain energy distribution and fault activity in this region will provide data for seismic risk evaluation, and be helpful for earthquake prediction research. This is also one of the regions with the highest GPS station concentration in continental China, capable of providing reasonable constraints on deformation modeling with fine details. Many studies have been performed in the region in the past, not only providing abundant information for model input, but also being helpful for result comparison and explanation.Crustal deformation in the northeast margin of the Tibetan plateau is modeled using a deformable block motion model, constrained by a GPS derived horizontal velocity field. The studied region spans 90-110oE, 28-42oN. Deformation field is assumed to be the result of boundary slips associated with relative block motion and uniform deformation within the blocks. The studied area is divided into 20 blocks initially, based on the a priori information from previous geological and seismological studies, and velocity gradients shown in the GPS velocity field. Kinematic parameters within each block and slip rates along block boundaries are evaluated, taking into account of contribution of fault locking effect to the horizontal velocity field. Since it is yet determined whether the motion of any station within each block is consistent with that of the others, and if all the blocks are independent of each other and all the blocks deform internally, F-test is used to screen out station velocity outliers within each block, justify independence of neighboring blocks, and determine significance of strain rate parameters through an iteration process, each time eliminating a station from the database or adjusting the block boundary, removing a block boundary, or adding a set of strain parameters within a block. Sixteen blocks have been identified as a result, among that 10 blocks demonstrate significant internal deformation. Alashan, Ordos, and Minqin blocks, with small translation and rotation rates and little or no internal deformation, are relative stable blocks in the studied region, blocking the northeastward push caused by the collision between the India Plate and the Tibetan plateau. The Qilian and Haiyuan blocks form a narrow stripe with strong internal deformation, absorbing the northeastward motion of blocks located at its southwest. The region confined by the Haiyuan, East Kunlun, and Longmenshan faults rotate clockwise, with the blocks moving from northeastward to southeastward, and the strain rates decreasing from west to east. Regional faults are categorized into three groups, based on their strike orientations and slip mechanisms: the NWW-trending faults and the Garzê-Yushu-Xianshuihe fault zone are transpressional, controlling crustal motion of the studied region. The sinistral slip rates across the Qilian Mountain northern frontal, Danghenanshan, Xiangshan-Tianjingshan, Haiyuan, Qinghainanshan, West Qinling Mountain northern frontal, East Kunlun, and Garzê-Yushu-Xianshuihe faults are of 2.3±0.3, 2.3±0.7, 3.3±0.5, 5.9±0.4, 0.9±0.6, 1.4±0.3, 1.5~13.4, and 12.2~13.5 mm/a, estimated from the deformable block model. Right slip rates of 2.7±0.9, 0.2±0.4, and 2.9±0.8 mm/a are estimated across the NNW and NS-trending Elashan, Zhuanglanghe, and Minjiang faults. The Altyn Tagh and Langshan Mountain-front faults slip left laterally at rates of 2.9±0.4 and 1.4~2.3 mm/a, and the Yinchuan, Longmenshan, and Longriba faults slip right laterally at rates of 3.3±0.3, 0.6~1.6 and 6.1±0.9 mm/a, respectively. Slip rates of major active faults in the studied region are also estimated by a profile projection method, with their activity significance justified by F-test. The estimated slip rates by both methods are essentially consistent with each other. Deformation activities of the Jinchang-Minle and Maqu-Luoxu velocity gradient zones are justified as 100% and 98.8% confidence, with a 4.3±0.3 mm/a left slip and a 3.2±1.1 mm/a right slip estimated by deformable block model, respectively. The ratio of equivalent seismic moment accumulation rates within and along the boundaries of blocks is estimated as about 0.31, while the ratio of energies released by earthquakes within and along the boundaries of blocks as about 0.36, showing quite consistent results between the two.A linked-fault-element model is employed to invert for contemporary slip rates along major active faults in the Sichuan-Yunnan region (96°~108°E, 21°~35°N) using the least squares method. The model is based on known fault geometry, and constrained by a GPS-derived horizontal velocity field. The results support a model attributing the eastward extrusion of the Tibetan plateau driven mainly by the north-northeastward indentation of the Indian plate into Tibet and the gravitational collapse of the plateau. Resisted by a relatively stable south China block, materials of the Sichuan-Yunnan region rotate clockwise around the eastern Himalayan tectonic syntaxis. During the process the Garzê-Yushu, Xianshuihe, Anninghe, Zemuhe, Daliangshan, and Xiaojiang faults, the southwest extension of the Xiaojiang fault, and the Daluo-Jinghong and Mae Chan faults constitute the northeast and east boundaries of the eastward extrusion, with their left slip rates being 0.3~14.7, 8.9~17.1, 5.1±2.5, 2.8±2.3, 7.1±2.1, 9.4±1.2, 10.1±2.0, 7.3±2.6, and 4.9±3.0 mm/a, respectively. The southwestern boundary consists of a widely distributed dextral transpressional zone other than a single fault. Right slip rates of 4.2±1.3, 4.3±1.1, and 8.5±1.7 mm/a are detected across the Nanhua-Chuxiong-Jianshui, Wuliangshan, and Longling-Lancang faults. Crustal deformation across the Longmenshan fault is weak, with shortening rates of 1.4±1.0 and 1.6±1.3 mm/a across the Baoxing-Beichuan and Beichuan-Qingchuan segments. Right slip of 5.1±1.2 mm/a is detected across the Longriba fault northwest of the Longmenshan fault. Relatively large slip rates are detected across a few faults within the Sichuan-Yunnan block: 4.4±1.3 mm/a left slip and 2.7±1.1 mm/a shortening across the Litang fault, and 2.7±2.3 mm/a right-lateral shearing and 6.7±2.3 mm/a shortening across the Yunongxi fault and its surrounding regions. Besides, locking depth of Xianshuihe fault is estimated as 15 km with 70% confidence range of 11~19 km.3 Understanding on continental deformation pattern and its dynamic mechanismStudy of crustal deformation in the southeast margin of the Tibetan plateau reveals that the region is divided into numerous blocks, which move relative to one another along faults with limited slip rates. Also study of crustal deformation in the northeastern margin of the Tibetan plateau using a deformable block model reveals a crustal deformation pattern of relative motions between numerous blocks of a hundred-kilometer scale. Ratio of strain accumulation rates within and along boundaries of blocks is about 0.31, while that of energies released by earthquakes is about 0.36. Therefore, strain energy of crust in the eastern margin of Tibetan plateau is accumulated mainly along active faults, yet there is still a significant portion accumulated within blocks. Based on seismic studies of crustal structures along eastern margin of the Tibetan plateau, the crust becomes gradually thinner from the plateau interior to the surrounding region, and weak zones develop at places in the lower crust and/or upper mantle of the eastern margin of the Tibetan plateau, implying partial decoupling between the upper and lower crust or between crust and mantle and decrease of crustal strength. Driven by the north-northeastward indentation of the Indian plate into Tibet, the crustal materials of the northeastern margin of the Tibetan plateau north of the Garzê-Yushu-Xianshuihe fault are blocked by the Alashan and Ordos blocks, resulting in a clockwise rotation, realized by relative motion and internal deformation of numerous micro-blocks confined by faults trending NWW and NNW. Southeastward motion of the crust results in strain accumulation across the Longmenshan region, and extrudes the Sichuan basin moving southeastward at the same time. Driven by the eastward extrusion of the Tibetan plateau and the gravitational collapse of the plateau, crust materials south of the Garzê-Yushu-Xianshuihe fault move southward toward a region of weak resistance, resulting in a left slip across the Xianshuihe-Xiaojiang fault to the east and a widely distributed dextral-transpressional zone to the west. In conclusion, the crustal deformation pattern in the eastern margin of the Tibetan plateau is not consistent with either of the“continental extrusion”or“continual deformation”model, but something in between. It is essentially determined by the rheological structure of the lithosphere and the tectonic forces imposing on the lithosphere. If no weak zone exists in the mid-lower crust or upper mantle, crust and mantle is coupled, and the lithosphere is strong enough to sustain high strength and is not easy to deform or rupture under tectonic stress. Only limited number of faults cut through the whole lithosphere, and the deformation pattern fits the“continental extrusion”hypothesis. If large scale weak zones exist in the mid-lower crust or upper mantle, upper and lower crust or crust and mantle are decoupled with reduced strength in the crust, and the crust is easy to deform or rupture under tectonic stress. A large number of faults break the upper crust and merge into the weak zones at depth, and the deformation pattern can be explained by the“continuous deformation”model. If fractional weak zones exist in the mid-lower crust or upper mantle, the upper and lower crusts or the crust and mantle are partly decoupled, and the crustal deformation pattern must be between the two extreme cases above.更多还原