【作者】 汪建军；

【导师】 许才军；

【作者基本信息】 武汉大学 ， 大地测量学与测量工程， 2010， 博士

【摘要】 应力触发是指地震(或火山)发生后造成后续断层(或火山)的力学性质及物理、化学性质的改变,抑制或加速断层错动(或火山活动)现象。在过去近二十年内,库仑应力是解释主余震间、主震间、地震与火山间、火山与火山间相互触发和断层相互作用现象以及评估未来潜在地震断层段的重要量化指标。应力触发已然成为研究地震发生的力学驱动机制的重要方向。然而,应力触发领域仍然存在一些根本性的问题尚待解决：其一为接收断层的合理选择；其二为库仑应力模型对其参数的敏感性；其三为地震周期中的应力转移模式。选择合理的接收断层有利具有可预测性的库仑应力图的构制；明确库仑应力模型对其参数的敏感性有利于应力触发分析结论的可靠性；确立地震周期中不同时段的应力转移模式有利于更完整地认识地壳应力场的时空演化过程以及定位未来潜在地震危险段。因此,本文针对这三个问题从理论分析和震例分析两个方面加以研究：建立库仑应力正反演分析解析模型,将最佳破裂面模型推广至三维情形以及构建库仑应力误差估计式；考察1997年Mw7.6玛尼地震与2001年Mw7.8年可可西里地震间的应力转移以及2008年Mw7.9汶川地震周边断层同震、震后和震间应力演化。基于库仑应力基本理念(断层面所受剪切力与摩擦力之和若为正则断层面倾向于滑动或者若断层面法向应力为拉张正应力则断层上下两盘易于张裂),本文建立了适于库仑应力正反演分析的统一解析模型并推广了常见的二维最佳破裂面模型至三维情形。以2003年伊朗Mw6.6 Bam地震和2008年Mw7.9汶川地震为震例,检测了三维最佳破裂面模型对提高库仑应力同余震空间相关性的性能。结果表明：对2003年Bam地震而言,采用均匀接收断层时库仑应力对余震触发率为47.06%,而采用三维最佳破裂面模型时仑应力对余震触发率为87.53%；对2008年汶川地震而言,采用均匀接收断层时库仑应力对余震触发率为45.68%,而采用三维最佳破裂面模型时仑应力对余震触发率为58.20%。库仑应力对余震触发率的提高源自三维最佳破裂面模型中的接收断层具有更大的自由度,因而可以普遍性地提高库仑应力对余震的触发效能。因此,对大量震源机制未定的余震情形可以采用三维最佳破裂面模型来更好地解释余震的空间分布。基于库仑应力正反演分析统一解析模型,建立了库仑应力模型的误差估计式并系统性地研究了库仑应力模型对源断层的滑动分布、发震断层激发的应力张量、摩擦系数、接收断层的走向、倾角、滑动角等的不确定性和构造背景应力场的敏感性以及库仑应力模型本身。结果表明：(1)若源断层的滑动分布愈精细,则库仑应力空间分布愈平滑；源断层激发的应力张量的误差所致的库仑应力相对误差大体为90%-100%。此两点表明为构制合理的库仑应力图,源断层的滑动分布需予以尽量反演得更精细。(2)走滑型源断层激发的库仑应力对模型系数不太敏感,而倾滑型和张裂型源断层激发的库仑应力对摩擦系数略敏感,因此库仑应力模型中常采用以断层的长期滑移历史确定断层的模型系数是合理的。(3)接收断层的走向误差相对其倾角误差更易于致使库仑应力相对误差偏高,而较之倾角误差其滑动角误差所致的库仑应力相对误差相对偏低,因此接收断层的走向较之其余二者更应予以更好地加以约束。(4)若构造应力场的量值较之地震激发的应力场的量值要小,则构造应力场的大小及方向对库仑应力空间分布影响甚若,反之则构造应力场的方向控制库仑应力空间分布。因此采用顾及构造应力场的最佳破裂面模型时,构造应力场的方向需尽量予以准确确定。(5)采用断层孔隙压力与断层正应力成正比假设的库仑应力模型所构制的库仑应力空间分布与摒弃该假设的库仑应力模型所构制的库仑应力空间分布略有不同,但因这一假设所致的二者量值差异甚为明显,因此在进行库仑应力定量分析时应采用不顾及该假设的库仑应力模型。本文考察了昆仑断层自1937年以来六次强震间的应力转移,结果表明这六次强震可以采用同震应力转移图像来解释。着重分析了1997年Mw7.6玛尼地震和2001年Mw7.8可可西里地震间的应力转移。结果表明：1997年玛尼地震对2001年可可西里地震有一定的促进作用,对可可西里地震断层面上的应力扰动平均为0.002bars,而2001年可可西里在1997年地震断层面上的库仑应力均为正且为0-0.08bars,它对未来于1997年地震断层面上的地震危险性有一定的促进作用。本文从同震库仑应力、震后库仑应力以及震间库仑应力三个方面分析了2008年Mw7.9汶川地震周边断层上的库仑应力及与余震的时空演化模式。地震断层的同震错动、震后下地壳和上地幔物质的粘性流动以及震间地震断层的浅部闭锁均对地壳应力场的调节起着重要的作用。顾及同震、震后和震间的库仑应力空间分布图像可以很好地解释1到2年内余震的时空变化模式。2008年周边断层上的总库仑应力表明玉树-玛曲断裂、鲜水河断裂、安宁河断裂、则木河断裂、大凉山断裂、东昆仑断裂以及海原断裂可能是未来地震危险段。更多还原

【Abstract】 The Coulomb failure function (CFF) quantitatively describes stress changes in sec-ondary faults near the source fault of an earthquake. It can be employed to monitor how stress transfers and then to shed some light on the probability of successive events occurring around a source fault. Accordingly, it has become a crucial quantitative index to investigate and interpret mainshock-mainshock, mainshock-aftershock, mainshock-volcano and volcano-volcano triggering phenomena as well as communication among major active faults over the past two decades. Nevertheless, there are still some funda-mental problems which should be figured out. One is which kind of receiver fault should be chosen for the analysis of earthquake stress triggering and the other is how sensitive the CFF is to its parameters. Besides, what the characteristic of stress evolution is when coseismic, postseismic and interseismic Coulomb stresses are completely considered also should be investigated, compared with that of conventional coseismic Coulomb stress. Therefore, the aim of this dissertation is twofold:(1) setting up a general analytical Coulomb stress model and estimator of CFF; (2) exploring stress transfer among the 1997 Mw7.6 Manyi earthquake and the 2001 Mw7.8 Kokoxili earthquake and stress evolution due to coseismic, postseismic and interseismic Coulomb stress of the 2008 Wenchuan earthquake.Based on the basic philosophy of Coulomb stress that the seismogenic fault would slide if the shear stress on it overcomes its frictional force and it would unclamp if its normal stress is extensional and vice versa, I build a unified model to determine an optimally oriented plane and its corresponding Coulomb stress, then apply the model to the 2003 Mw 6.6 Bam (Iran) earthquake and the 2008 Mw 7.9 Wenchuan (China) earthquake, thereby checking its effectiveness. The results show that spatial correla-tion between positive Coulomb stress changes and aftershocks are, for the 2003 Bam earthquake,47.06% when elastic Coulomb stress changes are resolved on uniform planes and 87.53% when these are resolved on optimally oriented planes at depth; for the 2008 Wenchuan earthquake the correlations are 45.68% and 58.20%, respectively. It is recommended that account be taken of optimally oriented planes when drawing a Coulomb stress map for analyzing earthquake triggering effects between mainshock and aftershocks, in which case the source mechanisms of aftershocks are often undetermined.I next construct an error estimator of Coulomb stress model based on the unified model built previously and then systematically investigate the sensitivity of Coulomb stress model to the uncertainties of the slip distribution of source fault, the stress tensor relative to seismogenic earthquake fault, the friction coefficient of receiver fault, the strike, dip and rake of receiver fault and tectonic background stress as well as Coulomb stress model itself. The results show that:(1) The larger the standard deviation of the slip distribution of source fault inverted, the more fluctuational the spatial pattern of Coulomb stress is, especially in the vicinity of source fault, and the relative error of Coulomb stress arising from the uncertainty of stress tensor imparted by source fault is generally 90%-100%, both of which indicate that the slip distribution of source fault should be carefully constrained in order to draw a reasonable Coulomb stress map; (2) The Coulomb stress is not dramatically sensitive to the uncertainty of friction coefficient for strike-slip source fault but it is slightly affected for dip-slip and tensile source faults, implying that choosing an empirical friction coefficient is reasonable according to the secular slip history of seismogenic fault; (3) The uncertainty of the strike of receiver fault affects Coulomb stress more heavily than that of the dip of receiver fault does and for the rake of receiver fault case its effect is more trivial, and thus the strike of receiver fault should be better constrained for the analysis of Coulomb stress than the other two;(4) When the magnitude of tectonic background stress chosen is relatively smaller than those of stress induced by earthquake, both the magnitude and direction of tectonic background stress have minor role in changing the spatial pattern of Coulomb stress, whereas when the magnitude of tectonic background stress chosen is larger, say tens of times of the magnitude of earthquake stress, the direction of tectonic background stress begins to dominate the spatial pattern of Coulomb stress. Given that tectonic background stress is commonly larger than the stress induced by earthquake, the direc-tion of tectonic background stress should be carefully determined when one considers Coulomb stress on optimally oriented plane receiver fault;(5) The spatial pattern of Coulomb stress calculated with a simpler Coulomb stress model where an assumption that pore pressure is proportional to normal stress of fault exits is not dramatically different from that of Coulomb stress calculated with the Coulomb stress model without such assumption. However, the difference of the magnitude of Coulomb stress owing to this assumption is not ignorable. Therefore, the Coulomb stress model without this assumption should be employed for analysis of Coulomb stress triggering.After presenting a unified Coulomb stress model and systematically investigating the sensitivity of Coulomb stress model to its parameters and thereby bearing some creteria in mind, I explore stress transfer and stress evolution of three strong earth-quakes, the 1997 Mw7.6 Manyi earthquake, the 2001 Mw7.8 Kokoxili earthquake and the 2008 Mw7.9 Wenchuan earthquake in Tibet Plateau as case studies of Coulomb stress triggering.I examine the coupling of stress transfer among strong earthquakes in the vicinity of Kunlun fault such as 1937 M7.5 Huashixia earthquake,1963 M7.0 Doulan earthquake, 1973 M7.3 Manyi earthquake,1997 Mw7.6 Manyi earthquake,2001 Mw7.8 Kokixili earthquake and 2010 Mw6.9 Yushu earthquake with focus on stress transfer between the 1997 Mw7.6 Manyi earthquake and the 2001 Mw7.8 Kokoxili earthquake with Coulomb stress model. The result show that the occurrences of these six strong earthquakes can be interpreted with coseimsic Coulomb stress map. Nevertheless, Only 2010 Mw6.9 Yushu earthquake can be interpreted with coseismic and interseismic Coulomb stresses remarkably well, which might arise from the fact that the effect of viscoelastic relaxation of lower crust and upper mantle is not considered. Besides, the Coulomb stresses im-parted by the 1997 Mw7.6 Manyi earthquake on ruptured fault planes of the 2001 Mw7.8 Kokoxili earthquake are-0.03-0.02bars with the average being 0.002bars and most of fault planes have positive Coulomb stress, indicating that the former weakly promoted the latter; The 2001 Mw7.8 Kokoxili earthquake imparted all positive Coulomb stresses no larger than 0.08bars on the fault planes of the 1997 Mw7.6 Manyi earthquake and hence it would hasten potential earthquakes to occur on ruptured planes of the 1997 Manyi earthquake.I probe the spatial pattern of aftershocks of the 2008 Wenchuan Mw7.9 earth-quake and stress evolution due to coseimsic, postseismic and interseismic deformation on major active faults around this strong earthquake. The coseismic stress imparting from the 2008 Mw7.9 Wenchuan earthquake, postseismic stress due to postseismic re-laxation of lower crust and upper mantle and interseismic strain accumulation are now completely considered, compared with previous analysis of Coulomb stress triggering of the 2008 Mw7.9 Wenchuan earthquake. The results show that the spatiotemporal pattern of aftershocks of the 2008 Mw7.9 Wenchuan earthquake is consistent with the net Coulomb stresses remarkably well, that is, most of the aftershocks lie in zones with positive Coulomb stresses. The coseismic stresses on major faults imparted by the 2008 Mw7.9 Wenchuan earthquake is changed by time-dependent postseismic and interseis-mic Coulomb stresses and therefore complete Coulomb stresses coming from different phases such as coseismic, postseismic and interseismic ones in earthquake cycle should be considered in order to make a robust assessment of the segments of faults with po-tential earthquake hazard. The net Coulomb stress map shows that Yushu-Maqu fault, Xianshuihe fault, Anninghe fault, Zemuhe fault, Daliangshan fault, the eastern Kunlun fault and Haiyuan fault are all gradually being loaded with positive Coulomb stresses and pose earthquake hazards in future.更多还原