基于振幅和瞬时频率的震相自动识别方法

为了充分反映地震信号振幅增大和频率改变的特征,发展了一种以振幅和瞬时频率比乘积为特征参量的震相自动识别方法。谐波信号和有限实际地震记录的应用表明,本方法具有较高的识别精度。     

一种改进的STA/LTA震相自动识别方法

震相自动识别是地震速报和地震预警中的重要工作之一.针对震相自动识别中最常使用的方法STA/LTA算法,提出了一种能反映地震信号振幅和频率变化特征的新的特征函数CF(i)=Y(i)2-Y(i-1)Y(i 1)(其中CF(i)为i时刻的特征函数,Y(i)为i时刻的地震信号振幅).谐波信号和有限的实际地震记录的应用表明,本方法具有较高的识别精度.     

一种适用于地方震事件的S波到时自动拾取方法

基于特征值分解方法,本文讨论了一种适用于地方震事件S波震相到时拾取的自动处理算法。该算法计算参数少、简便快捷、易于实现,通过选用七个不同长度的时间窗,有效地减小了窗长选择不合理所引起的震相拾取误差。利用福建地震台网记录的9 855条三分向波形记录进行测试,结果表明:本文方法的S波平均拾取偏差为(0.003±1.34) s,其中79.6%的记录拾取偏差小于0.5 s,4.1%的记录拾取偏差超过2.0 s,说明本文方法能够满足日常工作基本需求。综上分析认为,波形记录质量是影响拾取算法结果精度的最主要因素,信噪比较高的记录,其S波到时拾取偏差显著优于信噪比较低的记录,对信噪比较低的部分记录进行带通滤波预处理后,S波震相拾取精度也有所提升。     

青藏高原东北缘深地震测深震相研究与地壳细结构

通过对青藏高原东北缘不同构造单元深地震测深资料震相的综合分析,利用反射率理论地震图方法对实际记录模拟计算,进一步研究东北缘区域内部不同构造单元地壳细结构.结果显示：西秦岭褶皱造山带分隔了南北不同性质的地壳结构,北侧为相对稳定的临夏—兰州新生代盆地、南侧为强烈改造的松潘—甘孜地块;松潘—甘孜地块在青藏高原东北缘的构造演化过程中改造为萎缩的若尔盖高原盆地和盆地边缘褶皱造山两类不同的地壳结构;青藏高原东北缘中下地壳普遍存在以多层高低速相间、低速度结构为主的破碎松弛结构,这种特征在缝合带和造山带尤为明显,显示为地壳形变增厚、流变滑动的重要场所;结合二维速度结构及GPS研究结果,对青藏高原东北缘地壳形变及动力学过程进行了讨论.     

基于样本增强的卷积神经网络震相拾取方法

为了快速、高效地从地震数据中识别地震事件和拾取震相,本文利用基于样本增强的卷积神经网络自动震相拾取方法,将西藏林芝地区L0230台站3个月数据作为训练集,该区内另外6个台站连续1个月的波形数据作为测试集,采用高斯噪声、随机噪声拼接、随机挑选噪声、随机截取地震事件等4种样本增强的方法扩增训练集,以提高自动震相拾取技术的准确率。结果显示:样本增强前模型在测试集上的地震事件识别准确率为80%,样本增强后提升至97%,表明样本增强有效地提高了模型的泛化性能和抗干扰能力;在0.5 s误差范围内,震相自动拾取准确率高于81%,在1.0 s误差范围内,准确率高于95%;利用基于样本增强的卷积神经网络震相拾取方法能够检测出人工拾取震相中误标和漏检的震相。      

地震检测与震相自动拾取研究

针对微震事件易受噪声干扰等特点,本文将STA/LTA方法和基于方差的AIC方法（var-AIC）相结合,在震相到时初步拾取的基础上,使用台站的德洛内（Delaunay）三角剖分及台站间最大走时差约束来减少噪声干扰的影响. 利用到时进行地震定位之后,根据台站预测到时,在设定的时间窗内对地震震相进行更精细的分析. 特别是针对微震事件信噪比低的特点,设计了基于偏振分析的拾取函数,根据窗内STA/LTA方法和var-AIC方法的拾取结果自动选择合适的值作为震相到时. 最后,对西昌流动地震台阵2013年304个单事件波形数据的分析处理和检验结果表明,本文方法较传统方法具有更高的地震事件检测能力和更高的震相拾取精度.      

地震属性参数在札达—泉水沟深地震测深资料预处理与震相识别中的应用

地震属性分析技术在地球物理勘探领域的广泛应用,启发研究人员将其应用于人工源宽角反射/折射深地震测深剖面的资料预处理和震相识别。采用札达—泉水沟深地震测深资料,提取振幅、信噪比、主频、瞬时带宽、瞬时高频能量等地震属性参数,分析不同参数的物理含义,挑选其中对界面变化敏感的参数,对深地震测深资料进行预处理,并利用P波和S波的联合扫描,提高震相识别的准确性。走时互换结果显示,采用地震属性参数可有效提高震相拾取的准确性,进而提高后续地壳速度结构反演结果的精度。     

天然地震识别与震相自动拾取技术进展

震相拾取是地震数据处理过程中最基本的步骤之一。在传统的人工拾取技术不能满足庞大的地震数据处理需求的情况下,震相自动拾取技术从产生到发展至今经历了漫长的过程。本文回顾并总结了震相自动拾取技术的发展状况,重点介绍了长短时窗法、赤池准则法、模板匹配技术、基于自相关盲搜索的FAST法、S波偏振分析法、人工智能方法等,以及近年发展起来的多频率震相识别、全波形叠加、二次方自回归模型等方法,同时分析了每种方法的优势和局限性。     

近震S波震相实时自动识别方法研究

提出了一种用于地震早期预警的S波震相实时自动识别方法. 该方法不对原始信号进行任何滤波处理, 直接对三分向记录进行计算分析. 首先根据P波前0.5 s数据的卓越频率计算适用于该三分向记录的窗长, 采用由偏斜角和水平能量与总能量比值的平方积作为确定S波识别区间的特征函数, 将特征函数已有数据的5倍均值和5倍方差之和作为识别区间的触发阈值; 然后采用VAR-AIC方法对两个水平分向识别区间的数据分别计算分析, 对两个识别结果进行判断, 最终确定S波初动时刻. 经过对118个三分向记录的实际应用验证, 通过自动识别结果与人机交互震相识别结果相比, 本文方法对于S波相对P波尾波信噪比大于5 dB的地震记录, 其识别误差小于0.1 s的概率高达89.39%.      

一种识别宽频带震相的人工智能网络方法

提出了用反向传播智能网作为检测器来识别宽频带震相的一种方法。通过对3种反向传播智能检测器（长周期、中周期和短周期）的结果进行综合判断，认为这种方法既有短周期检测器准确性高的特点又有长周期检测器误警率低的特点。我们举例证明了适当地对数据进行预处理和反处理有助于改进系统的性能。本文也对反向传播智能网的结构和参数的设定进行了讨论。我们用研制成功的反向传播智能网检测器对美国地震学联合研究协会地震台网的1254张宽频带地震图上的地震事件进行了初至检测，希望能用这些走时资料进行地幔结构的层析成像研究。结果表明：1254张地震图中95％以上可识别出初至。自动识别的走时精确度尚可，85％以上的走时误差小于1s，约80％的走时误差小于0．5s。     

一种地震P波和S波初至时间自动拾取的新方法

地震P波、S波初至时间的拾取是地震波分析的一项基础性工作.本文提出了一种新的地震波初至时间自动拾取的方法:首先,把地震波的三分量时程曲线变换为一组空间向的能量变化率时程曲线;然后对能量变化率时程曲线进行STA/LTA(Short Time Average/Long Time Average,短时间的均值/长时间的均值)处理,拾取地震P波和S波的大致初至时间;最后提出采用一种二次方自回归模型对初至附近的能量变化率曲线进行二次方自回归处理,精确拾取出P波和S波的初至时间.本文采用了10组芦山地震的记录数据和150组汶川地震的记录数据对此方法的可靠性进行了检验.以人工拾取结果为参考,此方法具有很高的准确率和稳定性,同时,相比于常用的STA/LTA方法和AIC(Akaike Information Criterion,Akaike信息准则)方法,此方法在计算时间效率方面稍微逊色,但是对S波初至时间的拾取精度和可靠性更高.此方法丰富了地震P波、S波初至时间的自动拾取方法.     

基于双谱幅值和相位重构的地震子波提取

利用地震记录双谱中包含子波的幅值和相位信息,以及其超强的抗噪声干扰能力,采用一种基于双谱幅值和相位重构的地震子波提取方法,首先提取出子波幅值及相位信息,进而通过傅立叶反变换,使子波得以完全恢复.本文针对双谱相位重构递推公式,提出一种新的初值选取方法,使地震子波估计的稳定性得到了提高.仿真实验证实了该方法的可行性.     

Automatic onset phase picking for portable seismic array observation

Automatic phase picking is a critical procedure for seismic data processing, especially for a huge amount of seismic data recorded by a large-scale portable seismic array. In this study is presented a new method used for automatic accurate onset phase picking based on the proporty of dense seismic array observations. In our method, the Akaike＇s information criterion （AIC） for the single channel observation and the least-squares cross-correlation for the multi-channel observation are combined together. The tests by the seismic array observation data after triggering with the short-term average/long-term average （STA/LTA） technique show that the phase picking error is less than 0.3 s for local events by using the single channel AIC algorithm. In terms of multi-channel least-squares cross-correlation technique, the clear teleseismic P onset can be detected reliably. Even for the teleseismic records with high noise level, our algorithm is also able to effectually avoid manual misdetections.     

台湾海峡南部一次5.0级地震的sPn震相分析

运用Pn震相对齐测定sPn震相的方法分析了发生在台湾海峡南部的一次5．0级地震的 sPn震相,初步了解sPn震相的一些特征,并在不同地壳模型下用sPn-Pn的走时差计算了该地震的震源深度。结果表明:在该地震中,sPn震相特征明显;用Pn震相对齐测定sPn震相的方法可以快速、可靠地测定出sPn震相;用sPn-Pn的走时差计算震源深度时,不同模型计算出的结果相差不大。     

Study on the crust-mantle structure in the central and southern parts of Shanxi

Another comparative interpretation was conducted with respect to the data from 5 DSS profiles in the central and southern parts of Shanxi, leading to the conclusion that in Linxian, Linfen and Xingtai earthquake regions, through which the five profiles pass, there exist anomalous crust-mantle structure and abyssal crustal faults extending to Moho, all being regarded as the deep indications for earthquake occurrence.     

CRPN: A cascaded classification and regression DNN framework for seismic phase picking*

Current deep neural networks (DNN) used for seismic phase picking are becoming more complex, which consumes much computing time without significant accuracy improvement. In this study, we introduce a cascaded classification and regression framework for seismic phase picking, named as the classification and regression phase net (CRPN), which contains two convolutional neural network (CNN) models with different complexity to meet the requirements of accuracy and efficiency. The first stage of the CRPN are shallow CNNs used for rapid detection of seismic phase and picking P and S arrival times for earthquakes with magnitude larger than 2.0, respectively. The second stage of CRPN is used for high precision classification and regression. The regression is designed to reduce the time difference between the probability maximum and the real arrival time. After being trained using 500,000 P and S phases, the CRPN can process 400 hours’ seismic data per second, whose sampling rate is 1 Hz and 25 Hz for the two stages, respectively, on a Nvidia K2200 GPU, and pick 93% P and 89% S phases with the error being reduced by 0.1s after regression correction.     

考虑震相清晰度时的最优化定位方法

在文献[1]的基础上详细讨论了各台站直达波初动震相清晰度存在差异时如何利用井下和地面观测资料联合确定震源位置、发震时刻及区域平均波速的最优化方法。理论和实际检验表明,对于一个台网而言,考虑各台站震相清晰度间的差异,会使定位主要依赖于震相清晰的台站的资料,而震相不太清晰的台站的资料只对定位起参考作用。从而提高了整个台网定位的可信度。     

Stationary phase approximation in the ambient noise method revisited

The method of extracting Green's function between stations from cross correlation has proven to be effective theoretically and experimentally.It has been widely applied to surface wave tomography of the crust and upmost mantle.However,there are still controversies about why this method works.Snieder employed stationary phase approximation in evaluating contribution to cross correlation function from scatterers in the whole space,and concluded that it is the constructive interference of waves emitted by the scatterers near the receiver line that leads to the emergence of Green's function.His derivation demonstrates that cross correlation function is just the convolution of noise power spectrum and the Green's function.However,his derivation ignores influence from the two stationary points at infinities,therefore it may fail when attenuation is absent.In order to obtain accurate noise-correlation function due to scatters over the whole space,we compute the total contribution with numerical integration in polar coordinates.Our numerical computation of cross correlation function indicates that the incomplete stationary phase approximation introduces remarkable errors to the cross correlation function,in both amplitude and phase,when the frequency is low with reasonable quality factor Q.Our results argue that the distance between stations has to be beyond several wavelengths in order to reduce the influence of this inaccuracy on the applications of ambient noise method,and only the station pairs whose distances are above several (5) wavelengths can be used.     

On the treatment of uncertainty in seismic vulnerability and portfolio risk assessment

In a related study developed by the authors, building fragility is represented by intensity‐specific distributions of damage exceedance probability of various damage states. The contribution of the latter has been demonstrated in the context of loss estimation of building portfolios, where it is shown that the proposed concept of conditional fragility functions provides the link between seismic intensity and the uncertainty in damage exceedance probabilities. In the present study, this methodology is extended to the definition of building vulnerability, whereby vulnerability functions are characterized by hazard‐consistent distributions of damage ratio per level of primary seismic intensity parameter—Sa(T₁). The latter is further included in a loss assessment framework, in which the impact of variability and spatial correlation of damage ratio in the probabilistic evaluation of seismic loss is accounted for, using test‐bed portfolios of 2, 5, and 8‐story precode reinforced concrete buildings located in the district of Lisbon, Portugal. This methodology is evaluated in comparison with current state‐of‐the‐art methods of vulnerability and loss calculation, highlighting the discrepancies that can arise in loss estimates when the variability and spatial distributions of damage ratio, influenced by ground motion properties other than the considered primary intensity measure, are not taken into account.