【作者】 李正芳；

【导师】 周本刚；

【作者基本信息】 中国地震局地质研究所 ， 构造地质学， 2013， 博士

【摘要】 不均匀体的概念最初是在地震学中为了解释地震波的高频辐射成份提出来的，用来反映断层面上应力明显高于周围的部分。由不均匀体的研究引入的非均匀地震破裂模式，能较好的解释地震波中的复杂成份、主震前破裂的成因以及主破裂之后的应力集中，因此，不均匀体被认为是断层面上破裂的起始器、阻力器和集中器。根据不均匀体在地震破裂运动过程中发挥的作用，可分为凹凸体和障碍体两大类进行研究。凹凸体可理解为震前断层面上存在的一些高强度的未发生破裂的区域，可为下次地震发生的起始点或破坏最严重的点。障碍体则被称为强硬的应力集中区域，可作为断层破裂段的边界，起到限制破裂，在极端的情况下还会终止破裂的作用。目前为止，国内外学者对破裂段上的不均匀体大多从强震动记录和地震波反演的结果进行研究，从地表破裂方面的信息入手探讨不均匀体的方法仍比较缺乏，出于这种角度考虑，本文将在前人研究成果的基础上，以国内研究程度较高的断裂带为实例，研究断裂带上不均匀体的识别方法及相关特性，并探讨其在地震危险性分析中的应用。1、断裂带上凹凸体的识别凹凸体被定义为断裂系统中应力积累的强硬闭锁段，最终以大震的形式释放其主要能量，在地震破裂模型中被广泛使用，一般来讲，对已经发生过大震的断层面上的凹凸体可以由地震波分析和地表位错模型资料联合反演的方法确定其大小和位置，而未发生过地震且地震活动较活跃的断裂带，如何识别断裂带上的凹凸体至今为止仍是多数研究者一直讨论的问题。本文将着重从地震活动性、同震位移分布等方面给出凹凸体的识别方法。1）地震活动性分析本文对龙门山断裂带和鲜水河断裂带上1970年以来记录的小震数据进行了收集、整理和分析，采用了基于matlab平台的zmap软件，去除了断裂带上的丛集数据和余震，划定了有效地震数据的时间和震级范围，通过最大似然法求取了断裂带所在区域的b值分布图。基于b值大小与应力高低成反比的原理，通过断裂带上低b值区识别凹凸体的位置。在龙门山断裂带，通过低b值区识别出的现凹凸体的位置与汶川地震发生前所处的起始破裂位置和极震区的位置基本保持一致；而鲜水河断裂带由于受到小震数据的限制，部分段缺失b值分布，但整条断裂带仍可清晰识别出凹凸体位置，且1725年以来的历史强震和1970年以来5级以上的历史地震基本上都位于此区域。从断裂带的实例分析结果反映，利用小震数据通过最大似然法计算b值分布图，其相对低b值区与历年强震发生的位置存在较大的相关性，说明了利用低b值区识别凹凸体方法的可行性和实用性。2）基于地表破裂同震位移数据建立了凹凸体模型通过收集、整理和分析中国西部10条以走滑地震为主的破裂带的同震位移数据，采用统计学的方法拟定了凹凸体模型，建立了地表破裂参数与凹凸体模型参数之间的关系，是目前国内外利用地表破裂资料建立凹凸体模型的首次尝试。本模型可据不同断层的地表破裂位移值，给出其断裂带上最大凹凸体和所有凹凸体占断层破裂长度或破裂面积的比值，但该模型受到建模时同震位移数据的限制，利用该模型计算断裂带上凹凸体的大小时，需满足两个条件：一是断裂带上有较详细的地表破裂位移数据；二是该断裂带上地表破裂位移的最大值Dmax与地表破裂的平均位移值Dave的比值Dmax/Dave需小于等于3。基于拟合的模型，以汶川地震和昆仑山地震为例，识别了该破裂带上凹凸体的位置和大小，发现与前人地震波反演得到结果基本吻合。研究表明，本文建立的模型与前人的模型在识别断裂带上的凹凸体都具有实用性，区别在于，前人的模型是基于地震发生后，通过强震动反演得到的；而本模型是基于地表破裂位移的基础上拟合而成，对于那些具有历史地震和古地震破裂位移数据，而未有详细地震动记录的断裂带，本模型更具有实用性。尤其是随着科学技术的发展，可利用地面LIDAR技术识别出高清晰度地表破裂位移数据的技术支撑下，使得该模型在识别凹凸体方面具有更大的优势。因此，本文拟合的模型为断裂带上凹凸体的识别提供一种新途径，为判定断裂带上的强震危险性的分析提供了强有力的理论依据。2、断裂带上障碍体的识别地震发生时，岩层破裂并出现局部的滑动，但仍有未滑动即未受破坏的高应力强度区，震后该区域上的应力强度大于周围断层面上的应力强度，这种块体部分称为障碍体。当地震的破裂传播遇到障碍体时，障碍体可能被破坏，也可能在破裂通过后未破坏，或当时虽未破坏，后期随着周围的动应力即构造应力与障碍体强度值比值的增加，最终发生破坏，而破裂通过时是否发生破坏，取决于障碍体区域的大小及其自身的抗应力强度。文中以汶川地表破裂带和东昆仑断裂带为例，研究断裂带几何结构与障碍体的关系。认为破裂带存在的拉分阶区、挤压阶区、断裂的交汇处、断裂的急剧拐弯处等特殊构造部位及其同震位移趋势呈现突然下降拐点或波谷的位置，都可视障碍体存在的地方，根据其是否完全阻止破裂扩展，进一步划分为持久性障碍体和非持久性障碍体。通过收集大量国内外的震例，利用统计分析的方法分震级档给出了限制破裂传播的障碍体的止裂尺度，当走滑地震的震级介于6.0~6.9之间，阶区的最小止裂宽度为3km；走滑地震的震级介于7.0~7.5之间，阶区的最小止裂宽度为4km；当震级介于7.5~8.0之间，阶区的最小止裂宽度为6km；当震级介于8.0~8.5之间时，阶区的最小止裂宽度为8km，且拉分阶区比挤压阶区更容易被破裂所贯通。3、考虑不均匀性的潜在震源区强震复发行为的地震危险性分析在目前的科学认识水平下，地震的发生及地震动特性都具有一定不可预见性，必须以概率的方式来表达对未来地震及地震动的预测，即为概率地震危险性分析。潜在震源区作为概率地震危险性分析方法中一个十分重要的概念，其边界、震级上限及其地震活动性参数是决定地震危险性分析结果的关键因素。最早提出潜在震源区概念时，有一个很重要的假定，就是潜在震源区内各处地震发生概率是均匀的。其后在国际上采用概率地震危险性方法编制的区划图一般均沿用了这一假设，而且我国国标“重大工程地震安全性评价技术规范”（GB17741-2005）中，也是采用均匀模型来描述潜在震源区地震的发生特征的。事实上，在地震危险性分析工作中，潜在震源区划分的规模较大，可达几百乃至几千平方公里，比如鲜水河断裂带中的潜在震源区的划分就是一个例子，该断裂带划分为北段、中段和南东段，四代区划图编图组将其划分为三个Ms8.0级的潜在震源区，这种均匀的分布会造成地震危险性的“稀释”，从而降低了对地震危险性的估计。目前随着对断裂带的强震复发行为定量研究的深入，在地震区划与工程场地地震安全性评价工作中，考虑活动断裂破裂的复发行为来划分潜在震源区和确定地震活动性参数是当前国际上流行的趋势。鉴于地震孕育和发生的复杂性，目前不可能有一个统一的模型来描述大陆内部强震的复发行为，在地震危险性分析中，需要在多活动断裂段的复发行为研究的基础上，建立具体的地震复发模型。本论文以具体断裂带为研究对象，建立适合研究区的发震概率模型，给出潜在震源区内地震年发生率的确定方法，并将断裂带上识别出的不均匀体融入到潜在震源区的划分中，勾画出潜在震源区中不均匀体的边界并计算其震级上限，充分反映出潜在震源区内各处地震发生概率的不均匀性。1）潜在震源区的强震复发概率模型通过收集、整理和分析青藏高原东北部22条断裂带上古地震数据，拟定了该区的地震复发概率密度函数。根据此函数可计算出对区内断裂带未来百年内强震原地复发的条件概率。论文中将本文拟合的模型与目前通用的发震模型计算的概率值进行比较，发现通用模型的自变量t/R越接近1的时候，计算的复发概率值P增长的幅度不如本文拟合模型敏感。因此，对于古地震数据研究程度较高的断裂带，利用本文拟合的模型评价其未来大震的危险性可能更为准确，尤其是对平均复发间隔小，离逝时间长的段；而目前通用的复发模型针对那些古地震研究程度较低的断裂带，复发间隔较长的段落，可能更适用。2）潜在震源区震级上限估算的不确定性活动断裂定量研究的资料在评价特定断裂上的强震危险性方面发挥较大的作用，但受种种条件的制约不是每一条活动断裂上都可轻易获取所需的定量数据，并且这些数据本身通常含有较大的不确定性。活动断层长度作为活动断层定量数据之一相比其他的数据较容易获得，不确定性较小，因而，利用断层破裂长度估算震级的统计关系被广泛的应用于潜在震源区的震级上限的评估中。文中收集了青藏高原区7级以上以走滑为主的30个地震地表破裂参数资料，拟合出青藏高原区新的震级与破裂带长度统计关系式，并结合前人的统计关系式，分别通过破裂带长度估算震级，求出估算震级与仪器震级的差值，把差值为正值（即估算震级偏大）的归为一类，差值为负值（估算震级偏小）的归为一类作分析和对比。研究发现差值为正值的地震所处的走滑断裂带一般位于一级块体或次级块体的边界断裂带上，差值为负值的地震所处的走滑断裂带大多位于一级块体或次级块体内部断裂带或断裂带的交汇处。基于上述分类的差异，作者对不同研究者拟合的回归关系计算的差值数据进行了统计分析，分别给出了修正计算结果不确定性的参考值，为降低估算震级的不确定性提供了理论依据。3）潜在震源区的划分及危险性计算文中以具体实例的形式给出鲜水河断裂带潜在震源区中凹凸体的位置和大小。四代区划图编图组将鲜水河断裂带划分为炉霍、道孚和康定三个Ms8.0级的潜在震源区，划分的依据主要是鲜水河断裂的发震构造标志、几何分段特征及历史地震和古地震数据来确定的。而论文中提供的基于不均匀性的潜在震源区划分图，对潜在震源区的边界范围的确定仍按以前研究者提供的原则和方法，只是在其已有的潜在震源区范围内通过识别其上的凹凸体，进一步细化潜在震源区内应力不均匀性。首先利用小震活动给出的应力分布图，勾画出边界，然后按照论文中拟合的凹凸体模型，计算了相应各个潜在震源区中凹凸体的大小，炉霍潜在震源区中凹凸体A的长度为60.71km、道孚潜在震源区中凹凸体B的长度为38.13km、由于康定潜源中凹凸体C所在的断裂带的同震位移分布形态不易识别，文中只按照应力的不均匀性，勾画出边界，长度约78km。最后利用论文中建立的的凹凸体的长度L凹与震级Ms的回归关系式，同时考虑回归拟合关系和鲜水河断裂带特殊的地质构造环境的不确定性，分别求的凹凸体A、B、C的对应的震级Ms的震级上限分别为7.6级、7.3级、7.8级，并利用复发概率模型计算了相应的该三个凹凸体未来百年的年发生率，分别为5.558E-03、1.6693E-02和5.91E-04。更多还原

【Abstract】 The concept of heterogeneity on the fault rupture plane was firstly put forward to explainhigh frequency radiation components of seismic waves, suggesting these components aregenerated by the portions where the stress is significantly higher than the surrounding part of thefault plane. The models of inhomogeneous earthquake faulting can account for complexcomponents, rupture genesis before the main shock of seismic waves and stress concentration afterthe main rupture, so the fault heterogeneity is considered to be the important factor that determinesinitiating, resistance and stress concentration on the fault plane. According to the functions ofheterogeneity in the earthquake rupture process, two models have been suggested to describe suchheterogeneity: asperity model and barrier model. An asperity can be regarded as a high-strengthpart which does not rupture before the earthquake and could be the starting point or seriousdamage point for the next earthquake. A barrier is a high-stress concentration area which may bethe boundary of a fault rupture section to limit the rupturing, or will terminate the rupturing inextreme cases. So far, many foreign scholars have made detailed studies on this issue, providingsupportive evidence for seismic risk analysis. While little work results on this subject has beenreported at home. This thesis focuses on identification methods of heterogeneity on the faultrupture plane with case studies on several well-documented fault zones and discussion itsapplication to seismic hazard analysis.1. Identification methods of asperities on the fault zoneAsperity is defined as tough locked segment of stress accumulation in a fault system, whicheventually releases most energy to spawn a major quake. In the case that a fault has generatedmajor quakes that are instrumentally recorded and investigated, the asperities on this fault can bedetermined, including their size and location, by analysis of seismic waves and joint inversion ofsurface displacement data. In the case without these data available, how to identify asperitiesremain a problem to be solved. The thesis presents several methods based on b values from small quakes and coseismic displacements.(1) Analysis of b valuesThis work collected, sorted and analyzed data of small earthquakes on the Longmen Shanfault zone and Xianshuihe fault zone since1970. Using the matlab platform and zmap software,the clustering data and aftershocks are removed, and time intervals and magnitude ranges ofeffective seismic data are defined. Then, in terms of the maximum likelihood analysis, the bvalues are mapped for the areas where the two fault zones are situated. On the assumption that thesize of the b value is inversely proportional to that of stress, the positions of asperities on the faultzone can be determined according to low-b value areas. On the Longmen Shan fault zone, theasperities derived from b-value distribution are largely consistent with the rupture initiation andmeisoseismal areas in their localities. While for the Xianshuihe fault zone, although partialsections lack b-value distribution due to limited data of small earthquakes, asperities can beestimated for most of the fault zone using this method, which are just the locations wherehistorical major events since1725and recent M5shocks since1970took place. These caseanalyses show that the low-b value areas are correlated with the epicenters of major quakes, thus itis possible to identify asperities of a fault zone using data of b values from small shocks.(2) Analysis of coseismic displacementsAssume that coseismic displacements along the fault zone observed on the ground are relatedto the slip distribution on the fault rupture plane. Considering that asperities of a fault can beinferred from slip displacement distribution as suggested of previous work, this thesis attempts toestablish the asperity model based on statistics to data of coseismic displacements. Such data werecollected from10strike-slip faults in western China for the model to be constructed. In this model,two parameters are defined in terms of coseismic displacements: one is the ratio of maximumasperity’ length or area to the whole fault rupture, and the other is all asperities to the wholerupture. It requires that there are enough data of surface coseismic displacements available and theration of maximum displacement to the average one cannot exceed3.0Taking the2001Kunlun Shan M8.1and2008Wenchuan M8.0as examples, locations andsizes of asperities on their faults are estimated using the method aforementioned. The results arelargely consistent with that derived from seismic wave inversion in previous work. If thisapproach is indeed correct and effective, it would be more applicable for those faults only with displacement data of historical or paleoearthquake earthquakes but lacking instrumental records.In particular, in the case that surface displacements of can be clearly measured by the LIDARtechnology, this method seems to have a big advantage as a new approach.2. Identification of barriers on fault zonesIn theory, barriers on a fault should be related with its geometry. For instance, at dilational orantidilational jogs, and bends of a strike-slip fault, coseismic displacement can decrease abruptly.These localities can be regarded as barriers to obstruct further spread of rupture. Such barriers canbe divided into persistent and non-persistent barriers.In light of many case studies at home and around the world, this thesis suggests a scalerelationship between earthquake magnitude and the width of a jog or barrier, which can stoprupturing, on a strike-slip fault using statistics. The results are as follows: the minimum widths are3km,4km,6km, and8km for M6.0~6.9, M7.0~7.5, M7.5~8.0, and M8.0~8.5, respectively. Andthe dilatational jog is easier to rupture through than the antidilational jog.3. Recurrence models of potential seismic sources considering heterogeneity and seismichazard analysisAt present, prediction of earthquake occurrence and ground motions contains uncertainties ofsome extent, thus it is usually expressed by a probability form. In probabilistic analysis of seismichazard, the potential seismic source model is an important factor which is commonly assumed tobe uniform. However, an earthquake source can be very large and is actually not uniform due tothe heterogeneity of the fault including its geometry and physical properties. Thus this factorshould be taken into consideration when a recurrence model of a seismogenic fault is constructed.（1）Recurrence models of potential sourcesThis work collected and analyzed data of paleoearthquakes on22faults in the northeasternTibetan plateau, and calculated the density function of earthquake recurrence in this region. Fromthis function, the conditional probabilities of in situ recurrence of major events in the future100years can be computed. Comparison of the model from this thesis with the currently used modelshows that when the independent t/R in the general model approaches1, the growth amplitude ofrecurrence probability is less sensitive than that of the fitting model in this thesis. It means that the common recurrence model is suitable for those faults with long recurrence intervals and few dataof paleoearthquakes. While for the faults with well-studied paleoearthquakes, particularly withshorter intervals and longer elapsed time, the model in this work would yield better estimates ofseismic risks in the future.（2）Uncertainties in estimation of upper limit of magnitude for potential sourcesUsing data of surface ruptures produced by30earthquakes of M7or greater in the Tibetanplateau, this thesis fitted the relationship between magnitude and rupture length by statistics. Withreference to previous work, earthquake magnitudes are estimated using this relationship and theirdifferences relative to instrumental magnitudes are calculated. These differences are classified intopositive (meaning greater values) and negative (smaller values) for comparison. It is found that thestrike-slip faults with positive differences lie mostly on boundaries of blocks or sub-blocks, whilethose with negative differences are within block or sub-blocks and at intersections of faults. Bystatistical analysis to these difference values, this work provides reference values of uncertaintiesfor correction, which can help reduce the uncertainty in upper limit of magnitude estimation.（3）Determination of potential seismic sources and seismic risk calculationFor the Xianshuihe fault zone, this work determines locations and sizes of asperities on itssources. First, using stress distribution from small earthquakes, all asperities in potential sources,which are determined by fault segmentation etc. in previous work, are delineated. Then, accordingto the asperity model in this thesis, the size of each asperity is calculated. The results show that theasperities A is60.71km long, B is38.13km long, and C is78km long which is estimated by stressheterogeneity. Based on the relationship between the length of asperities and magnitude (Ms),which is established in this work, the upper limit magnitudes of asperities A, B, and C are solvedto be7.5,7.3, and7.8, respectively. Finally, their recurrence probabilities in the future100yearsare calculated to be5.558E-03,1.6693E-02, and5.91E-04, respectively.更多还原