Modelling co- and post-seismic displacements revealed by InSAR, and their implications for fault behaviour
Modelling co- and post-seismic displacements revealed by InSAR, and their implications for fault behaviour
复制标题
DOI：
--
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
W. Feng
中科院分区：
文献类型：
--
作者：
W. Feng
地震学的最终目标是估计沿原有断层发生的未来地震事件的时间、震级和潜在的空间范围。基于速率-状态摩擦定律，针对这一目标提出了几种理论物理地震模型。在这些模型中，需要构造加载速率和断层的摩擦性质。现代大地测量，如GPS和InSAR，在大地震后提供了前所未有的近场观测。理论上，根据摩擦率和状态粗糙地震模型，断层上保持地震运动的速度弱化区应与断层仅在地震中滑动的速度增强区分开。然而，GPS测量揭示的2011年东北9.1级地震后的早期余滑大部分覆盖在历史断裂带上，这对速度减弱凹凸体模型提出了挑战。因此，基于实验室的摩擦定律在自然事件中的表现还需要进一步的研究，也需要系统地讨论可能影响通过大地测量模拟来估计滑动模型的因素。为了解决这一重要问题，本文对几次中强地震进行了研究。由InSAR形变时间序列确定的青海海西6.3级逆冲滑动地震后的最佳同震和震后滑动模型表明，最大余滑与同震滑动模型集中在同一地区，这与2011年日本地震中观测到的模式相似。在这种情况下，复杂的几何凹凸体可能在同震成核和震后断裂中起着至关重要的作用。由一张Cosmo-SkyMed震后干涉图揭示的2011-MW7.1级范主震后早期主要余滑发生在同震滑动模式的正上方。在这一事件中，也测试了不允许在同震凹凸体上滑动的震后模拟，表明没有凹凸体滑动的滑动模型可以解释震后观测，以及没有限制凹凸体滑动的余滑移模型。在2011年日本东北冲9.1级地震中，利用格雷斯同震重力变化和内陆同震GPS观测资料进行了联合反演，以重新研究主震的同震滑动模型。对这些不同数据集的滑动模型的比较表明，滑动模型可以观察到显著的变化，特别是最大滑动的位置。节理滑移模型表明，最大滑移量约42m出现在日本海沟附近的海底表面附近。同时，前人研究确定的累积余滑模式(滑动>；2m)与节理滑动模型产生的库仑应力变化具有空间相关性。作为一次走滑断层事件，2011年玉树6.8级地震也通过同震和震后建模进行了研究，所用的SAR数据比以往研究中使用的更多。最佳滑动模型表明，主要余滑集中在断层的浅部和两种主要的同震滑动模式之间，这表明速率和状态摩擦粗糙度模型在该事件中的表现是合适的。其他震后物理机制，如孔隙弹性回弹和粘弹性松弛也被研究过，这些都不能显著影响本研究中对浅层余滑模型的估计。认为在这种情况下，浅层余滑主要控制了主震后的震后行为。与使用其他研究的大地测量数据调查的另外21次地震相比，可以确定同震滑动模型和震后滑动模型之间的互补空间范围。2009年青海6.3级地震是一个特例，其断裂行为可能受控于断裂构造(如断裂弯曲)。综上所述，本文的主要贡献包括：1)摩擦定律在本文研究的大多数自然事件中具有一阶拟合度；2)几何粗糙度可能在地震旋回期间的断层形成中发挥重要作用；3)同震和震后滑动模型中的显著不确定性可能显著地偏离对断层摩擦性质的估计；4)本文从每一次地震中观察到关于其断层结构和复杂断层行为的新见解；以及(5)开发了一个新的大地地震模拟程序包，它可以处理包括InSAR、GPS和基于陆地/空间的重力变化在内的多种数据集。
The ultimate goal of seismology is to estimate the timing, magnitude and potential spatial extent of future seismic events along pre-existing faults. Based on the rate-state friction law, several theoretical physical earthquake models have been proposed towards this goal. Tectonic loading rate and frictional properties of faults are required in these models. Modern geodetic observations, e.g. GPS and InSAR, have provided unprecedented near-field observations following large earthquakes. In theory, according to the frictional rate and state asperity earthquake model, velocity-weakening regions holding seismic motions on faults should be separated with velocity-strengthening regions within which faults slip only aseismically. However, early afterslip following the 2011 MW 9.1 Tohoku-Oki earthquake revealed from GPS measurements was largely overlaid on the historical rupture zones, which challenged the velocity weakening asperity model. Therefore, the performance of the laboratory based friction law in the natural events needs further investigation, and the factors that may affect the estimates of slip models through geodetic modelling should also be discussed systematically. In this thesis, several moderate-strong events were investigated in order to address this important issue. The best-fit co- and post-seismic slip models following the 2009 MW 6.3 Haixi, Qinghai thrust-slip earthquake determined by InSAR deformation time-series suggest that the maximum afterslip is concentrated in the same area as the coseismic slip model, which is similar to the patterns observed in the 2011 Japan earthquake. In this case, complex geometric asperity may play a vital role in the coseismic nucleation and postseismic faulting. The major early afterslip after the 2011 MW 7.1 Van mainshock, which was revealed by one COSMO-SkyMed postseismic interferogram, is found just above the coseismic slip pattern. In this event, a postseismic modelling that did not allow slip across the coseismic asperity was also tested, suggesting that the slip model without slip in the asperities can explain the postseismic observations as well as the afterslip model without constraints on slip in the asperities. In the 2011 MW 9.1 Tohoku-Oki earthquake, a joint inversion with the GRACE coseismic gravity changes and inland coseismic GPS observations was conducted to re-investigate the coseismic slip model of the mainshock. A comparison of slip models from these different datasets suggests that significant variations of slip models can be observed, particularly the locations of the maximum slips. The joint slip model shows that the maximum slip of ~42 m appears near the seafloor surface close to the Japan Trench. Meanwhile, the accumulative afterslip patterns (slip >2 m) determined in previous studies appear in spatial correlation with the Coulomb stress changes generated using the joint slip model. As a strike-slip faulting event, the 2011 MW 6.8 Yushu earthquake was also investigated through co- and post-seismic modelling with more SAR data than was used in previous study. Best slip models suggest that the major afterslip is concentrated in shallow parts of the faults and between the two major coseismic slip patterns, suggesting that the performance of the rate and state frictional asperity model is appropriate in this event. Other postseismic physical mechanisms, pore-elastic rebound and viscoelastic relaxation have also been examined, which cannot significantly affect the estimate of the shallow afterslip model in this study. It is believed that the shallow afterslip predominantly controlled the postseismic behaviour after the mainshock in this case. In comparison to another 21 earthquakes investigated using geodetic data from other studies, complementary spatial extents between co- and post-seismic slip models can be identified. The 2009 MW 6.3 Qinghai earthquake is an exceptional case, in which the faulting behaviours might be dominated by the fault structure (e.g. fault bending). In conclusion, the major contributions from this thesis include: 1) the friction law gives a first order fit in most of natural events examined in this thesis; 2) geometric asperities may play an important role in faulting during earthquake cycles; 3) significant uncertainties in co- and post-seismic slip models can appreciably bias the estimation of fault frictional properties; 4) new insights derived from each earthquake regarding their fault structures and complex faulting behaviours have been observed in this thesis; and (5) a novel package for geodetic earthquake modelling has been developed, which can handle multiple datasets including InSAR, GPS and land/space based gravity changes.