地震动模拟的两步随机经验格林函数方法研究

发布时间：2018-04-23 00:12

本文选题：强地震动 + 模拟　；参考：《中国地震局工程力学研究所》2017年博士论文

【摘要】：合理预测未来可能发生的地震中工程场地的地震动是地震危险性分析的关键,近年来提出的两步随机经验格林函数方法得到了广泛应用并成功模拟了一些地震的地震动,但是该方法仍然存在一些亟待解决的问题。本文基于两步随机经验格林函数方法开展地震动模拟的相关研究,重点解决该方法目前存在的一些问题,包括:地震应力降的表示、模拟地震动不确定性的定量分析、震源破裂方向性的模拟、缺乏经验格林函数的处理,本文旨在为两步随机经验格林函数方法在地震危险性分析中的应用打下基础。本文主要工作如下:(1)给出了用于地震动模拟的应力降估计模型并检验了模拟结果的有效性。基于地震应力降的研究结果,将地震应力降表示为服从对数正态分布的数据集,标准差(log10)确定为0.2-0.6,平均应力降根据地震平均滑动位错与破裂面纵横比的关系近似估计。基于两步随机经验格林函数方法模拟了2013年Mw6.6级芦山地震、2008年Mw6.9级岩手宫城内陆地震和2016年Mw7.1级熊本地震的地震动,震源模型只考虑高应力降的凹凸体区域。模拟结果发现,短周期(2.0s)模拟地震动与观测记录较为一致或相差不大,长周期(2.0s)模拟地震动则明显偏低,这可能与经验格林函数缺乏长周期信号有关,凹凸体之外的低应力降破裂区域对模拟地震动的贡献很小。(2)定量给出了两步随机经验格林函数方法模拟地震动的不确定性。首先采用不同应力降比值(0.5-8.0,固定间隔1.5)分别模拟芦山、熊本和岩手宫城内陆地震的地震动,发现模拟地震动随应力降比值的增大而增大,模拟地震动的残差与应力降比值的自然对数有明显的线性负相关,据此建立了模拟地震动与应力降比值的定量关系。其次,采用蒙特卡罗随机抽样方法随机给出服从对数正态分布的目标地震应力降,其中应力降标准差(log10)分别为0.2-0.6,分别模拟上述三次地震的地震动,建立了模拟地震动标准差与应力降标准差的定量关系,结果发现应力降标准差越大则模拟记录的标准差相对越小,长周期模拟地震动的不确定性更小;当应力降标准差为0.2-0.6时,模拟记录的标准差约为0.11-0.37。最后,讨论了随机破裂过程引起的模拟地震动的不确定性,根据K-S检验证实多次随机过程模拟地震动服从对数正态分布,200次随机过程模拟地震动的对数标准差约为0.05-0.15,且随周期(0.05-2.0s)增大而增大。(3)提出了利用两步随机经验格林函数方法模拟震源破裂方向性的技术思路。首先,选取两次震源破裂方向性显著的芦山地震余震(EQII和EQIII)作为经验格林函数模拟主震地震动,模拟结果发现模拟地震动表现出显著的方向性效应,且与小震方向性特性一致,说明经验格林函数的选取有必要考虑小震震源破裂方向性的影响。其次,提出以视拐角频率替代拐角频率改进现有的两步随机经验格林函数方法,基于改进方法采用EQIII的记录再次模拟芦山地震,结果显示模拟地震动并未表现出明显的方向性效应,且与实际观测记录较为一致,证实改进方法有效地表示(消除)了小震震源破裂方向性的影响。最后,基于改进方法模拟了一系列震源破裂方向性特征不同的地震,模拟地震动表现出相应的破裂方向性效应,主破裂前方的模拟地震动普遍高于主破裂后方,验证了改进方法实现地震动方向性效应模拟的可行性。(4)探索性地提出了广义经验格林函数的概念,并将其应用于唐山地震的地震动模拟。为弥补小震记录缺乏对经验格林函数方法的限制,提出了不考虑大小地震震源位置差异的广义经验格林函数的概念。分析日本强震动记录与中国地震动衰减关系的事件内残差,结果表明中国东部强震区和青藏地震区与日本地区传播路径差异对震中距150km的地震动PGA的影响可近似忽略,对PSA有很小的影响,说明日本地区的强震动记录可近似作为中国地区目标地震的广义经验格林函数。选取K-NET台站及KiK-net井下基岩台站记录分别模拟芦山地震具体台站的地震动及地震动场,结果表明,观测记录大体上位于模拟地震动平均值加减一倍标准差范围内,初步检验了广义经验格林函数的可靠性。最后,选取KiK-net井下基岩台站的记录作为模拟1976年Ms7.8级唐山地震基岩面地震动的广义经验格林函数,模拟地震动(3.0s)与地震动预测方程预测值较为一致,且模拟地震动PGA等值线与宏观烈度较为一致。
[Abstract]:The reasonable prediction of the ground motion of the engineering site in the future may be the key to the seismic hazard analysis. In recent years, the two step random empirical Green function method has been widely used and successfully simulated some earthquake ground motions, but there are still some problems to be solved in this method. This paper is based on two steps of random random. The empirical Green function method is used to carry out the related research on seismic simulation, focusing on some problems existing in this method, including the representation of the seismic stress drop, the quantitative analysis of the earthquake dynamic uncertainty, the simulation of the rupture direction of the earthquake source, the lack of the experience of the empirical Green function, and the two step random empirical Green function square. The main work of this paper is as follows. The main work of this paper is as follows: (1) the stress drop estimation model for ground motion simulation is given and the validity of the simulation results is tested. Based on the results of the seismic stress drop, the seismic stress drop is expressed as a data set obeying the lognormal distribution, and the standard deviation (log10) is determined to be 0. 2-0.6, the average stress drop is approximately estimated according to the relationship between the average slip dislocation of the earthquake and the longitudinal and transverse ratio of the fracture surface. Based on the two step random empirical Green function method, the 2013 Mw6.6 Lushan earthquake, the Iwate Miyagi inland earthquake in 2008 and the 2016 Mw7.1 Xiongben earthquake in 2016 are simulated. The source model only considers the concave and convex of the high stress drop. The simulation results show that the short period (2.0S) simulated ground motion is consistent with the observation record, and the long period (2.0S) simulated ground motion is obviously lower. This may be related to the lack of the long period signal of the empirical Green function. The contribution of the low stress drop zone outside the concave convex body to the simulated ground motion is small. (2) the quantitative analysis is given. The two step random empirical Green function method is used to simulate the uncertainty of ground motion. First, different stress drop ratios (0.5-8.0, fixed interval 1.5) are used to simulate the ground motions of the inland earthquakes in Lushan, Xiongben and Iwate, respectively. It is found that the simulated ground motion increases with the increase of the ratio of stress drop, and the ratio of the residual and the stress drop of simulated ground motion from the simulated ground motion. There is a clear linear negative correlation, and the quantitative relationship between simulated ground motion and the ratio of stress drop is established. Secondly, the seismic stress drop of the target is randomly given by the Monte Carlo random sampling method, and the standard difference of stress drop (log10) is 0.2-0.6, respectively, to simulate the ground motion of the above three earthquakes respectively. The quantitative relationship between the standard deviation of ground motion standard deviation and the standard deviation of stress drop is established. The results show that the higher the standard deviation of the stress drop is, the smaller the standard deviation of the simulated record is, the uncertainty of the long period simulated ground motion is smaller. When the standard deviation of the stress drop is 0.2-0.6, the standard deviation of the simulated record is about 0.11-0.37. last, and the random rupture process is discussed. The uncertainty caused by the simulated ground motion is confirmed by K-S test. The simulated ground motion of multiple random processes obeys the lognormal distribution. The logarithmic standard deviation of the simulated ground motion in the 200 random process is about 0.05-0.15 and increases with the increase of the period (0.05-2.0s). (3) a two step random empirical Green function method is proposed to simulate the rupture of the source. First, we select two aftershocks of Lushan earthquake (EQII and EQIII) to simulate the main earthquake ground motion as the empirical Green function. The simulation results show that the simulated ground motion shows significant directional effect and is in accordance with the specificity of the small earthquake direction, which indicates that the selection of the empirical Green function is necessary. Considering the influence of the rupture direction of the small earthquake source, the two step random empirical Green function method is proposed, which is based on the angular frequency instead of the corner frequency. Based on the improved method, the Lushan earthquake is simulated again by the EQIII record. The results show that the simulated ground motion does not show obvious directional effect and is compared with the actual observation record. In order to agree, it is proved that the improved method effectively expresses (eliminate) the influence of the rupture direction of the small earthquake source. Finally, based on the improved method, a series of earthquakes with different directional characteristics of the source rupture are simulated. The simulated ground motion shows the corresponding rupture direction effect. The simulated ground motion ahead of the main rupture is generally higher than the rear of the main rupture. The feasibility of improving the seismic directional effect is simulated by the improved method. (4) the concept of generalized empirical Green function is proposed and applied to the earthquake simulation in Tangshan earthquake. In order to make up for the lack of the empirical Green function method, the generalized experience of the difference of the magnitude of the magnitude earthquake source is put forward. The concept of Green's function is used to analyze the incident internal residual difference between the Japanese strong vibration record and the attenuation of ground motion in China. The results show that the difference of the propagation path between the strong earthquake zone in eastern China and the Tibetan earthquake area and the region of Japan is negligible for the earthquake PGA of the epicentral distance 150km, and it has a small influence on the PSA, indicating the strong vibration in Japan. It is approximated as the generalized empirical Green function of the target earthquake in China. The K-NET station and KiK-net underground rock station records are selected to simulate the ground motion and seismic dynamic field of the Lushan seismic station. The results show that the observation records are generally located within the range of the average value of the simulated ground motion. The reliability of the Green function of the sense experience. Finally, the record of the KiK-net underground rock station is selected as the generalized empirical Green function to simulate the ground motion of the base rock surface of the 1976 Ms7.8 Tangshan earthquake. The simulated ground motion (3.0s) is in agreement with the prediction of the ground motion prediction equation, and the simulated ground motion PGA equivalent line is more consistent with the macroscopic intensity.

【学位授予单位】：中国地震局工程力学研究所
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P315

【参考文献】