摘要
发生在新生代早期的印度板块和欧亚板块间的碰撞不仅形成了喜马拉雅造山带,还造就了一个以独特的地势高度、地貌、地质环境、自然环境等特征而闻名于世的青藏高原。自19世纪中期普拉特和艾利等创立地壳均衡论以来,青藏高原的形成、演化及隆升机制的问题一直就是国际大陆动力学理论研究的核心和前缘热点。目前,已有许多学者提出了各种动力学模型试图解释青藏高原的隆升、演化之谜,这些模型和模式的提出对我们理解青藏高原的隆升机制起到了积极的启发和推动作用。不同的隆升机制会在青藏高原周缘尤其是东缘地区产生完全不同的形变模式。因此,研究青藏高原东缘三维地壳形变可为高原深部结果和动力学演化过程提供重要的定量数值边界条件,有助于理解高原隆升的地球动力学机制。作为青藏高原东边界和中蒙大陆中轴构造带中南段的青藏高原东缘地区,北起鄂尔多斯地块西缘,跨越秦岭,穿过龙门山,再沿安宁河-小江断裂带向南延至缅甸境内,成为分隔中国大陆东部相对稳定的鄂尔多斯高原、四川盆地和华南地块与西部强烈隆升的青藏高原之间的边界活动构造带。深部地球物理探测表明,西部青藏高原的地壳厚度达60~70km,东部华南和华北的地壳厚度只有40~50km,高原东缘地区则为非常明显的重力梯度带和地壳厚度突变带。晚新生代和现今构造变形在高原东缘两侧的差异非常明显,规模巨大的活动断裂和强烈地震主要发生在该带以西,而其以东不仅不发育大规模活动断裂,强震活动水平也远远低于西部。2008年5月12日,在龙门山地区发生了Ms8.0级地震,更证明了青藏高原东缘地区的构造活动性。由于青藏高原地震地质构造复杂等多种原因,利用地质学和地球物理学方法确定其构造变形的运动场在短时间内很难做到,近年来发展起来的GPS观测技术为测定现今构造变形的速度场提供了前所未有的有效手段。一些学者已经利用GPS观测获取了青藏高原东缘地区高分辨率的水平运动速度场图像。虽然GPS观测技术可以提供三维的地壳运动信息,但由于大气折射、发射和接收天线相位中心误差等因素的存在,GPS垂向定位精度较低。另一方面,GPS观测技术真正在我国开始大规模实施是在1999年中国地壳运动观测网络项目的运行,距今也就是十多年的时间。而我国地震水准监测网开始大规模观测始于1966年邢台地震之后,至今已有近50年的历史。因此,在地震科学各领域内以精密水准测量为主要手段的地壳垂直形变研究仍占重要地位。利用精密水准数据研究青藏高原东缘地区的地壳垂直形变是贯穿本文的一条主线,可分为以下两个研究内容。1青藏高原东缘地区现今地壳垂直运动研究利用青藏高原东缘地区40年的精密水准观测资料,获取了研究区内的现今地壳垂直运动速度场图像,为区域地壳垂直运动和强震中长期危险性预测研究提供了重要基础资料。在获取的区域垂直运动速度场基础上,结合前人得出的该地区现今地壳水平运动速度场结果,综合分析青藏高原东缘地区三维地壳运动速度场特征,并对地壳形变的动力学机制进行了初步探讨。取得的主要结论如下:(1)收集、整理青藏高原东缘地区的精密水准测量数据,包括1970年以来的地震水准监测网(主要资料来源)、全国二期和二期复测水准网(占比例较小)和“中国综合地球物理场观测-青藏高原东缘地区”项目于2010~2011年在滇中和滇西地区的观测资料,做出所有高差观测值随时间变化的曲线,剔除由于地震事件、地下水抽取等因素导致的不稳定水准点。共找出重复观测的水准点3439个,一等水准观测高差占总观测数据的97.5%,二等观测高差占2.5%。(2)在整体平差处理之前,我们选择于2010~2011年在滇中和滇西地区(26°N以南)观测的水准网来估计高差每公里的中误差。由于观测时间间隔短,基本保证了观测时间的同步性,可以采用静态平差方法。平差后得到高差改正数的每公里中误差为1.2mm。根据青藏高原东缘地区构造变形强烈和水准资料多年、多期复杂的特点,我们采用线性动态平差模型,以研究区内9个GPS测站(网络工程中的基准站和基本站)垂直运动速度结果作为先验约束,可以有效减小水准测量中系统误差的累计,统一处理获取了青藏高原东缘地区现今地壳垂直运动速度场图像。整体平差后得到的验后每公里高差改正数的中误差为0.97mm,与验前每公里中误差1.2mm相近,从小范围区域证明了结果的可靠性。(3)垂直运动速度场结果揭示出,青藏东缘地区地壳长期垂直运动趋势与已有地质学方法、GPS和水准观测得到的结果一致。青藏高原东缘大部分区域都存在上升趋势,其中贡嘎山地区上升速率最大达到了5.7mm/a,而西秦岭天水地区垂直速率最大达到6.4mm/a。(4)从横跨断裂带的垂直运动速度剖面上可以估计出断裂带的垂直滑动速率,可为一些不易于通过地质学方法得出垂向滑动量的断层提供定量约束。结果表明,垂直滑动速率最大的为龙门山断裂达到3.4±0.4mm/a,其次为大凉山断裂带,垂直滑动速率为2.0±0.4mm/a。贺兰山东麓、六盘山、龙日坝和小江断裂的垂直滑动速率为1~1.6mm/a,而则木河和红河断裂的垂直滑动速率不明显。(5)利用小波分解技术获取了青藏高原东缘不同波长的垂直运动速度场图像。其中,长波长(500-1000公里)的地壳垂直变形运动可能与青藏高原深部地幔的变形有关,区域短波长的变形则主要与地壳的变形有关。(6)将获取的青藏高原东缘地区现今地壳垂直运动速度场,与网络工程1999~2007年该区域的长期水平运动速度场相结合,分析和研究东缘地区的地壳运动学特征和地壳形变的动力学机制。物质通量分析表明,区域内大部分地区存在上下地壳差异性运动,即维持当今各地块隆升速率需要下地壳与上地幔物质注入速率高于上地壳物质注入速率,为本地区下地壳流的存在提供佐证。例如松潘-甘孜地块西部的隆起速率为0~1mm/a,其中部地区隆起速率增大至2~3mm/a,到其东部靠近四川盆地地区垂直速率下降为0~1mm/a,而如果假定岩石圈内物质水平运动均一则只能解释本地区平均~0.7mm/a的隆升。这种垂直运动现象可能揭示出在松潘-甘孜地块中下地壳存在管流层。可以认为青藏高原内部地壳物质向东扩展,由于受到四川盆地强硬地壳的阻挡,中下地壳物质以塑性流变的方式在龙门山及其以西川西高原之下堆积,导致川西高原中下地壳的显著增厚,并对上部脆性地壳施加垂直隆升作用,从而造成龙门山和川西高原的隆升。(7)区域三维地壳运动速度场结果揭示,贺兰山上升、银川地堑继承性沉陷;六盘山地区的抬升速率主要以地壳缩短的形式实现;川滇地块中南部地区由于东西向水平拉张而表现为下沉运动。(8)通过区域垂直形变速率场,结合5级地震震中分布,发现了滇西南地区龙陵-澜沧活动断裂带永德-镇康地震空段的现今高速率异常隆起特征,揭示了该地震空段的强震中长期危险性。2研究区内两大典型地震的同震及震后垂直形变研究利用水准数据分析和研究发生在青藏高原东缘地区的两次典型地震的同震及震后垂直形变,取得的主要结论如下:(1)利用1990年4月26日青海共和Ms7.0级地震同震垂直形变资料,在已有结论的基础上,修正了此次地震的同震滑动模型。基于震后多期精密水准数据,发现震后相邻测站高差观测值的时间序列明显具有对数衰减特征或指数衰减特征。发展了一个利用水准数据研究震后形变的方法。这种方法从相邻水准点之间的高差出发,可以有效减少系统误差累积带来的偏差并充分利用观测到的水准数据。发展了一个流变学形变模型解释共和地震震后形变。模型结果表明,震后滑移模型和粘弹性松弛模型共同导致了震后产生的地表垂直形变。滑移主要发生在同震破裂断层面及沿原断层面向上和向西北方向的延伸部分;粘弹性松弛主要出现在下地壳与上地幔,其粘滞系数为9×10~(19)Pa·s。这个结果揭示出柴达木盆地底部的下地壳和上地幔要比以往研究预想的脆性强、粘滞性高。(2)基于2008年5月12日四川汶川Ms8.0级地震GPS获取的中远场同震水平、垂直形变和精密水准测量获取的近场同震垂直形变,对同震断层的几何破裂模型以及断层面上的滑动分布进行反演。反演结果除了与前人已有结论一致外,还发现北川断裂深部15~18km范围内的滑动量值也比较大,约为2~4m。根据震后2008~2011年龙门山断裂带及周缘地区的GPS区域站观测资料获取了汶川地震震后水平形变的弛豫过程(对数松弛时间为12天),有助于认识汶川地震破裂机制,为深入研究震后形变的物理学机制提供基础资料。利用震后2008~2011年的龙门山断裂带附近的精密水准观测得到了震后垂直形变场的演化过程。
The collision between Indian and Eurasian plates that occurred in the earlyCenozoic not only formed the Himalayan orogenic belt, but also created the Tibetanplateau, which is famous for the unique elevation, geomorphology, geologicalenvironment and natural environment in the world. Since the mid-19th century Prattand Airy founded the isostasy theory, the formation, evolution and uplift mechanismof the Tibetan Plateau is one of the focused international studies of continentaldynamics. At present, many scientists have proposed a variety of geophysicaldynamic models and attempted to explain the uplift of the Tibetan Plateau. Thesemodels have played a positive inspiration and impetus to understanding of the upliftmechanism. Different models suggest rather different crustal deformation patterns inthe vicinity of the plateau, especially around its east margin. Thus, precisemeasurement of the three-dimensional crustal deformation in this region can provideimportant quantitative numerical boundary conditions for the dynamic evolution, andhelp to understand the kinematic and geodynamic models.As the eastern border of the Tibetan Plateau and the central and southern section ofthe Chinese-Mongolian continental mid-axis tectonic belt, the northern part of theeastern margin of the Tibetan Plateau is located on the western margin of the Ordosblock, and its central part crosses Qinling and through the Longmen Shan, its southernpart goes through the Anninghe-Xiaojiang fault. The eastern margin of the TibetanPlateau separates stable blocks of the Ordos, Sichuan basin and South China from thestrong uplift of the Tibetan Plateau. The deep geophysical exploration shows that thecrustal thickness of the Tibetan Plateau is about60~70km, while the crustal thicknessof North and South China is only about40~50km. The eastern margin of the TibetanPlateau is the gravity gradient and abnormal crustal thickness zone. Late Cenozoicand contemporary tectonic deformation on both sides of the eastern margin of theTibetan Plateau is obviously different. Large-scale active faults and strong earthquakes mainly occurred in the west of the zone, while its east does not hostlarge-scale active faults and the earthquake activity is much lower. On May12,2008,a devastating Ms8.0earthquake occurred in the Longmen Shan region, the middle partof the eastern margin of the Tibetan Plateau. This earthquake proves the tectonicactivity in the eastern margin of the Tibetan Plateau.Because of the complicated seismic geological structure in the Tibetan Plateau, it isdifficult to determine the tectonic deformation field by the use of geologic andgeophysic method in a short period. However, with the development of GPStechnique in recent years, it provides an unprecedented effective approach foracquiring the velocity field of the tectonic deformation. The image of thehigh-resolution velocity field for the eastern Tibetan Plateau has been obtained byGPS. Although GPS measurements can provide three-dimensional crustal deformation,due to the atmospheric refraction, the antenna phase centers of satellite and receiver,the accuracy of vertical positioning is lower than the horizontal. On the other hand,with the operation of Crustal Movement Observation Network of China (CMONOC)in1999, the observation history of GPS in China is only more than a decade. Bycontrary, the precise leveling networks were measured after the1966Xingtaiearthquake for nearly50years. Therefore, the precise leveling measurement is still amajor technique for obtaining the crustal vertical deformation which could be studiedin various fields of earthquake science research.Using precise leveling observations to study the vertical crustal deformation of theeastern margin of the Tibetan Plateau is a main theme through out this thesis, and thecontent can be divided into the following two aspects.1Present crustal vertical deformation of the eastern margin ofTibetan PlateauBased on the regional leveling observations collected in the eastern margin of theTibetan Plateau, we obtain the present crustal vertical velocity field, which couldprovide important basic information for the long-and middle-term seismic riskprediction. Combining the vertical velocity field acquired in this work and horizontal velocity obtained by previous studies, we analyze the characteristics of thethree-dimensional crustal movement of the eastern Tibetan Plateau, and discuss thegeodynamic mechanism of crustal deformation. The primary conclusions are asfollows.(1) The precise leveling data observed around the eastern margin of the TibetanPlateau are collected, including the leveling networks used in monitoring tectonicdeformation of main active faults in China since1970, the national leveling networksof China observed in1980s and1990s separately, and the central and western Yunnanregion surveyed between2010and2011which is a part of “Integrated GeophysicalField Observation–the eastern margin of Tibetan Plateau” project. All the heightdifferences between adjacent two benchmarks are plotted over time, so the instablebenchmarks caused by earthquake events or ground water drawing could be removed.Finally, there are3439benchmarks of which the ratio of first-order height differencesis97.5%, and the ratio of second-order is2.5%.(2) Before overall adjustment, we choose the leveling network located in the centraland western Yunan region (south of26°N) to estimate the prior uncertainty of eachkilometer of height difference. This leveling network was surveyed between2010and2011, and the movement of benchmarks can be considered as zero. So we can get theadjusted height elevation using the static adjustment. The result shows that theuncertainty of one kilometer of height difference is1.2mm.Because of the strong tectonic deformation in the studied region and the complexityof leveling data, the linear dynamic adjustment model is used to estimate the unknownparameters. The vertical velocities of9GPS stations within this region are as a prioriconstraints which can effectively reduce the accumulated systematic errors along theleveling routes. As a result, the present crustal vertical velocity field image is obtained.It shows that the posteriori uncertainty is0.97mm, which is consistent with the prioruncertainty. This indicates the reliability of velocity field result from a small region.(3) The trend of long-term crustal vertical movement obtained in this thesis isconsistent with existing results inferred from geological methods, GPS and levelingmeasurements. Most regions of the eastern margin of the Tibetan Plateau are in the status of uplift, especially Gonggashan uplifts at a rate of5.7mm/a, and the uplift rateof the western Qinling is up to6.4mm/a.(4) The vertical velocity profile across the fault can be used to estimate its verticalslip rate, which is difficult to derive from the geologic method and can providequantitative constraint. The result shows that the vertical slip rate of the LongmenShan fault is up to3.4±0.4mm/a, followed by the Daliang Shan fault with the rate of2.0±0.4mm/a. The vertical slip rates of the Helan Shan, Liupan Shan, Longriba andXiaojiang faults are between1and1.6mm/a. There is no significant vertical slipacross the Zemuhe and Red River faults.(5) The wavelet decomposition technique is applied to obtain different wavelengthsof the vertical velocity field for the eastern margin of the Tibetan Plateau. The crustalvertical deformation of long wavelength (500-1000km) can be related to the deepmantle under the Tibetan Plateau, while the regional deformation of short wavelengthmay be related to the crsutal deformation.(6) Combining the present crustal vertical velocity field with horizontal velocityfield obtained in the eastern margin of the Tibetan Plateau, we investigate thecharacteristics of crustal movement and corresponding dynamic mechanism. Thehorizontal material influx result shows that the movement of upper and lower crust israther different in most regions, namely, to maintain the uplift rate of each sub-blockrequires the influx rate of the lower crust and upper mantle higher than that of theupper crust, which provides the evidence for the existence of lower crustal flow in thisregion. For example, the uplift rate of the western Songpan-Ganzi block is0~1mm/a,the uplift rate of its central part increases to2~3mm/a, and the vertical rate of itseastern part which is close to the Sichuan Basin drops to0~1mm/a. However, theaverage horizontal movement of material in the lithosphere can only lead to an upliftrate of~0.7mm/a. This vertical movement may reveal the existence of channel flowin the middle and lower crust under the Songpan-Ganzi block. While the TibetanPlateau moves eastward, due to the blocking of stable crust of the Sichuan Basin, thelower crustal material accumulated under the Longmen Shan and western Sichuanplateau in the way of plastic flow. Therefore, the middle and lower crust of the western Sichuan plateau thickened significantly, and exerted a vertical uplift to theupper brittle crust, which resulted in the uplift of the Longmen Shan and westernSichuan plateau.(7) The regional three-dimensional velocity field indicates that the Helan Shan isuplifting and the Yinchuan graben is subsiding. The crustal shortening of the LiupanShan region is the primary drive for uplift. The subsidence of the central and southernSichuan-Yunnan fragment is caused by the near east-west extension.(8) Through the regional vertical velocity field, combining with epicenterdistribution of more than Ms5.0earthquakes, we find an abnormal zone near theYongde-Zhenkang town with a high uplift rate, which is an earthquake gap on theLongling-Lancang fault.2Coseismic and postseismic vertical deformation of two typicalearthquakes occurred in the eastern margin of the TibetanPlateauLeveling data are used to analyze and study the coseismic and postseismic verticaldeformation of two typical earthquakes that occurred in the eastern margin of theTibetan Plateau. The main conclusions are described as follows.(1) We model leveling deformation data observed following the1990Ms7.0Gonghe earthquake in Qinghai Province to infer postseismic deformation sources andmechanisms. Using coseismic vertical displacements and previous study resultobtained from seismic body wave inversion, we update the main shock rupture model.Time series of elevation change between adjacent benchmarks can be modeled by alogarithmic or an exponential relaxation function. To model the postseismicdisplacements, we utilize the raw observations of elevation differences betweenadjacent benchmarks, not their integrals with respect to a reference benchmark, toconstrain a dislocation model in a continuum. It makes full use of the leveling dataand effectively reduces biases introduced from cumulative errors due to dataintegration. The postseismic modeling result suggests that two mechanisms operate simultaneously to produce the postseismic vertical deformation observed at thesurface: afterslip on the coseismic rupture fault plane of the main shock and itsperipheral extension, particularly upward into the sediment layer above the mainrupture, and viscoelastic relaxation of the lower crust and upper mantle, with aviscosity of9×10~(19)Pa·s. The result suggests more brittle and less viscous lower crustand upper mantle underneath the Qaidam basin than some of previous studiesenvisioned.(2) Using the coseismic displacements from GPS and near-field precise levelingmeasurements before and after the2008Ms8.0Wenchuan earthquake, the cosesimicfault geometry and slip distribution are modeled. Besides the similar results withprevious studies, we find that the slip at the depth of15~18km under the Beichuanfault is as large as2~4m. The postseimic relaxation process is acquired by utilizingthe GPS data observed around the Longmenshan fault between2008and2011. It ishelpful to understand the rupture mechanism of this quake, and provides basicinformation for the physical mechanisms of postseismic deformation. The evolutionof postseimic vertical deformation is derived based on precise leveling measuredbetween2008and2011near the rupture.