基于POT模型的昆仑山地区地震统计特征分析

极值统计是研究较少发生但一旦发生即产生极大影响的随机事件的有效方法。本文以地震活动频繁的昆仑山地区作为研究区域,建立了基于广义帕累托分布的超阈值（POT）模型,并讨论了该地区若干地震活动性参数,包括强震震级分布、潜在震级上限、强震平均复发间隔、一定周期内的强震发震概率、一定时期内的重现水平和超定值重现震级。经统计分析得到:该地区震级阈值选定为M_S5.5,超阈值期望震级为M_S6.81,潜在震级上限高达M_S9.08,M_S8.0的平均复发间隔仅为66.8年,未来3年该地区发生M_S5.5～M_S6.5的概率在80%以上,百年重现水平即可达到历史最大震级M_S8.1。     

基于广义帕累托分布的地震震级分布尾部特征分析

极值理论在地震危险性分析中有着重要应用,发震震级超过某一阈值的超出量分布可以近似为广义帕累托分布.基于广义帕累托分布给出了若干地震活动性参数的估计公式,包括强震震级分布、地震复发周期和重现水平、期望重现震级、地震危险性概率和潜在震级上限等;以云南地区震级资料为基础数据,讨论了阈值选取、模型拟合诊断和参数估计;在此基础上计算了该地区的地震活动性参数.结果表明,广义帕累托分布较好地刻画了强震震级分布,通过超阈值(POT)模型计算的复发周期与实际复发间隔统计基本一致,高分位数估计在一定阈值范围内表现稳定,为工程抗震中潜在震级上限的确定提供了一种途径.     

基于广义帕累托分布的地震震级分布尾部特征分析

极值理论在地震危险性分析中有着重要应用, 发震震级超过某一阈值的超出量分布可以近似为广义帕累托分布. 基于广义帕累托分布给出了若干地震活动性参数的估计公式, 包括强震震级分布、 地震复发周期和重现水平、 期望重现震级、 地震危险性概率和潜在震级上限等; 以云南地区震级资料为基础数据, 讨论了阈值选取、 模型拟合诊断和参数估计; 在此基础上计算了该地区的地震活动性参数. 结果表明, 广义帕累托分布较好地刻画了强震震级分布, 通过超阈值(POT)模型计算的复发周期与实际复发间隔统计基本一致, 高分位数估计在一定阈值范围内表现稳定, 为工程抗震中潜在震级上限的确定提供了一种途径.      

龙门山地区的地震活动性广义帕累托模型构建

利用1930—2016年龙门山地区M≥4.5历史地震目录,通过时空窗法减少余震对统计结果的影响,构建了该地区广义帕累托分布的超出量分布模型,估计了龙门山地区震级上限。结果表明:广义帕累托分布较好地拟合了龙门山地区强震数据,形状参数估计均为负值,最大震级分布存在有限上界,震级上限为8.3。     

基于广义极值理论的潜在地震海啸源震级上限及强震重现水平的估计——以琉球海沟俯冲带为例

以琉球海沟俯冲带作为研究区,将广义极值理论用于估计潜在地震海啸源震级上限,首先分析了琉球海沟俯冲带的地震地质构造特征以及历史地震资料,界定潜在地震海啸源区,然后根据地震活动性特征按时间域进行分割,并提取各时间段发生的极限震级的地震样本,最后通过广义极值分布模型估计了该区域的震级上限值和强震重现水平,并对其进行了不确定性分析。     

基于轮廓似然估计的广义极值分布在地震中长期预测中的应用

为描述强震预测的不确定性,在地震预报极值分析模型的参数估计中,引入轮廓似然估计法。对广义极值分布中形状参数和地震重现水平的轮廓似然估计原理及数值算法进行了详细地阐述,并利用构建的广义极值分布模型对东昆仑地震带进行了地震危险性分析。关于形状参数和重现水平的点估计,以及10年以内的重现水平置信区间的估计,轮廓似然估计法与极大似然估计法效果基本相同,但在中长期地震重现水平置信区间的预测中,轮廓似然估计法得到的关于置信水平不对称的置信区间,在强震水平下对预测震级的不确定性表达更准确,预测结果更加有效。      

青藏高原东北缘地震活动性广义帕累托模型的全域敏感性分析

基于广义帕累托分布构建地震活动性模型,因其输入参数取值难以避免不确定性,导致依据该模型所得的地震危险性估计结果具有不确定性。鉴于此,本文选取青藏高原东北缘为研究区,提出了基于全域敏感性分析的地震危险性估计的不确定性分析流程和方法。首先,利用地震活动性广义帕累托模型,进行研究区地震危险性估计;然后,选取地震记录的起始时间和震级阈值作为地震活动性模型的输入参数,采用具有全域敏感性分析功能的E-FAST方法,对上述两个参数的不确定性以及两参数之间的相互作用对地震危险性估计不确定性的影响进行定量分析。结果表明:地震危险性估计结果（不同重现期的震级重现水平、震级上限及相应的置信区间）对两个输入参数中的震级阈值更为敏感;不同重现期的地震危险性估计结果对震级阈值的敏感程度不同;对不同的重现期而言,在影响地震危险性估计结果的不确定性上,两个输入参数之间存在非线性效应,且非线性效应程度不同。本文提出的不确定性分析流程和方法,可以推广应用于基于其它类型地震活动性模型的地震危险性估计不确定性分析。      

中国大陆活动地块边界带强震震级分布特征研究

本文利用基于广义帕累托分布的超阈值分布模型,对中国大陆活动地块边界带强震震级分布特征开展研究,给出了各活动地块边界带强震震级的广义帕累托分布参数估计。结果表明,广义帕累托分布较好地拟合了各边界带强震数据,形状参数估计均为负值,说明对应震级应有上限,因此广义帕累托分布为潜在震级上限提供了一种自然的刻画。在此基础上,估计了震级上限,并给出了分布0.99997高分位数估计,通过与历史最大震级比较发现,高分位数估计相对稳健。在地震发生过程为泊松过程假设下,推导了广义帕累托分布与广义极值分布之间的联系,揭示了一种利用强震数据推断最大震级分布的可能途径。     

基于广义帕累托分布的马尼拉海沟俯冲带地震危险性估计

选取马尼拉海沟俯冲带作为潜源区, 基于广义帕累托分布, 通过对一定时段内超过某一阈值的震级数据进行拟合, 建立该潜源区地震危险性估计模型, 估计强震重现水平和震级上限, 并对估计结果的不确定性进行了分析, 得到马尼拉海沟俯冲带震级上限为9.0级, 10 a、 50 a、 100 a、 200 a马尼拉海沟俯冲带的震级重现水平期望值分别为7.1级、 7.6级、 7.7级、 7.9级。     

基于广义极值分布的巴颜喀拉块体中部地震危险性分析

本文对广义极值分布模型的构建机理进行了深入详细的阐述,给出了逻辑意义更加合理的的重现期和重现水平定义,以及相关的地震危险性评价指标。在此基础上,应用构建模型对巴颜喀拉块体中部的地震危险性做了客观的评价,得出巴颜喀拉块体中部每年的平均最大发震为M_s5.1,每20年发生M_s6.0以上强震可能性超过97%,M_s7.5左右的超强震约100年一遇,块体内部孕育地震的能量积累迅速。     

根据截断的G-R模型计算东北地震区震级上限

震级上限是指一个地区可能发生地震的最大震级,其概率意义为发生超过该震级地震的概率几乎为0.在有些地区,由于对其内部的地震构造研究和认识存有局限性,很难根据构造或者地质学的原则来确定震级上限.因此,根据数学模型,采用统计手段,使用地震活动性资料来计算震级上限的估计值是一种可行的方法.本文根据截断的G-R关系模型,采用最大似然计算方法,使用东北地震区的地震目录,计算了东北地震区震级上限,结果表明东北地震区的震级上限应为M_u=7.5左右.计算中我们考虑了不同震级的转换、震级误差的修正以及计算方法的影响.最终结果表明,不论采用何种方案进行计算,东北地震区的震级上限值均始终保持在7.5左右,这说明我们采用本文中方法计算得到的东北地震区的震级上限值是合理可信的,同时也说明在以往的研究中对东北地震区震级上限的估计大都是偏小的.     

广义极值分布在地震危险性分析中的应用

利用广义极值分布研究地震最大发震震级的规律,给出了广义极值分布下地震危险性分析的相关公式与方法,并对台湾地区历史地震资料进行极值统计分析,发现最大震级超过7级的地震理论发震次数与实际发震次数完全一致,在此基础上预测了未来几年发震的危险性.     

1993年7—9月的全球地震动态

１９９３年第三季度，全球地震活动水平为中等偏高，明显高于上半年平均水平。日本北海道西南近海发生７．６级浅源地震，但不属于日本海沟地震。埃及西奈半岛发生５．７级地震，为今年亚欧带西段之最大地震。马里亚纳群岛发生８．１级中深震，使西北太平洋地区地震水平达到全球第一。兴都库什地区接连发生三次较大中深震，可能对我国西部地区地震活动有影响。墨西哥恰帕斯州近海发生７．３级地震，美洲带新的地震活动轮回正式开始。印度南部发生６．３级中强震，属于板内地震。     

THE THREE DIMENSIONAL DENSITY STRUCTURE OF CRUST AND UPPER MANTLE IN THE CENTRAL-SOUTHERN PART OF LONGMENSHAN

In recent years, strong earthquakes of M_S8.0 Wenchuan and M_S7.0 Lushan occurred in the central-southern part of Longmenshan fault zone. The distance between the two earthquakes is less than 80 kilometers. So if we can obtain the inner structure of the crust and upper mantle, it will benefit us to understand the mechanism of the two earthquakes. Based on the high resolution dataset of Bouguer gravity anomaly data and the initial model constrained by three-dimensional tomography results of P-wave velocity in Sichuan-Yunnan region, with the help of the preconditioned conjugate gradient(PCG)inversion method, we established the three dimensional density structure model of the crust and upper mantle of the central-southern segment of Longmenshan, the spatial interval of which is 10 kilometers along the horizontal direction and 5 kilometers along the depth which is limited to 0~65km, respectively. This model also provides a new geophysical model for studying the crustal structure of western Sichuan plateau and Sichuan Basin. The results show obvious differences in the crustal density structure on both sides(Songpan-Ganzê block and Sichuan Basin)of Longmenshan fault zone which is a boundary fault and controls the inner crustal structure. In Sichuan Basin, the sedimentary layer is represented as low density structure which is about 10km thick. In contrast, the upper crust of Songpan-Ganzê block shows a thinner sedimentary layer and higher density structure where bedrock is exposed. Furthermore, there is a wide scale low density layer in the middle crust of the Songpan-Ganzê block. Based on this, we inferred that the medium intensity of the Songpan-Ganzê block is significantly lower than that of Sichuan Basin. As a result, the eastward movement of material of the Qinghai-Tibet plateau, blocked by the Sichuan Basin, is inevitably impacted, resulting in compressional deformation and uplift, forming the Longmenshan thrust-nappe tectonic belt at the same time. The result also presents that the crustal structure has a distinct segmental feature along the Longmenshan fault zone, which is characterized by obviously discontinuous changes in crustal density. Moreover, a lot of high- and low-density structures appear around the epicenters of Wenchuan and Lushan earthquakes. Combining with the projection of the precise locating earthquake results, it is found that Longmenshan fault zone in the upper crust shows obvious segmentation, both Wenchuan and Lushan earthquake occurred in the high density side of the density gradient zone. Wenchuan earthquake and its aftershocks are mainly distributed in the west of central Longmenshan fault zone. In the south of Maoxian-Beichuan, its aftershocks occurred in high density area and the majority of them are thrust earthquake. In the north of Maoxian-Beichuan, its aftershocks occurred in the low density area and the majority of them are strike-slip earthquake. The Lushan earthquake and its aftershocks are concentrated near the gradient zone of crustal density and tend to the side of the high density zone. The aftershocks of Lushan earthquake ended at the edge of low-density zone which is in EW direction in the north Baoxing. The leading edge of Sichuan Basin, which has high density in the lower crust, expands toward the Qinghai-Tibet Plateau with the increase of depth, and is close to the west of the Longmenshan fault zone at the top of upper mantle. Our results show that there are a lot of low density bodies in the middle and lower crust of Songpan-Ganzê Block. With the increase of the depth, the low density bodies are moving to the south and its direction changed. This phenomenon shows that the depth and surface structure of Songpan-Ganzê Block are not consistent, suggesting that the crust and upper mantle are decoupled. Although a certain scale of low-density bodies are distributed in the middle and lower crust of Songpan-Ganzê, their connectivity is poor. There are some low-density anomalies in the floor plan. It is hard to give clear evidence to prove whether the lower crust flow exists.     

Regional Variation of the ω - Upper Bound Magnitude of GIII Distribution in and Around Turkey: Tectonic Implications for Earthquake Hazards

Complete data set of earthquakes in Turkey and the adjacent areas has been used in order to compute the ω values in 24 seismic regions of Turkey. The parameter is obtained through Gumbel’s third asymptotic distribution of extreme values and is well known as upper bound magnitude. This is an interpretation that no earthquake magnitude greater than ω can occur in a region. The results also estimate the most probable magnitude for a time period of 100 years. The estimates of ω exceed the value of 7.00 in 20 of the 24 seismic regions. An effort is also made to find a relation between the magnitude and the length of a fault in the complicated tectonics of Turkey and the surrounding area. Earthquake hazard revealed as tables and maps are also considered for Turkey and the surrounding area.