Duration of strong ground motion in terms of earthquake magnitude,epicentral distance,site conditions and site geometry

The physical bases and empirical equations for modelling the duration of strong earthquake ground motion in terms of the earthquake magnitude, the epicentral distance and the geological and local soil site conditions are investigated. At 12 narrow frequency bands, the duration of a function of motion f(t), where f(t) is acceleration, velocity or displacement, is defined as the sum of time intervals during which the integral S f² (τ) dτ gains a significant portion of its final value. All the records are band-pass filtered through 12 narrow filters and the duration of strong ground motion is studied separately in these frequency bands. It is shown that the duration of strong motion can be modelled as a sum of the source duration, the prolongation due to propagation effects and the prolongation due to the presence of the sediments and local soils. It is shown how the influence of the magnitude on the duration of strong ground motion becomes progressively stronger, in going from low to moderate frequencies, and that the duration is longer for ‘soft’ than for ‘hard’ propagation paths, at low and at moderate frequencies. At high frequencies, the nature of the broadening of the strong motion portion of the record with increasing distance is different, and is most likely related to the diffraction and scattering of the short waves by the velocity inhomogeneities along the wave path. It is also shown that the geological and local soil conditions should both be included in the model. The duration can be prolonged by 3–5 sec at a site on a deep sedimentary layer at frequencies near 0.5 Hz, and by as much as 5–6 sec by the presence of soft soil underneath the station, at a frequency of about 1 Hz. An empirical equation for a probabilistic estimate of the discrepancies of the predictions by our models relative to the observed data (distribution function of the residuals) is presented.     

Duration of strong ground motion during Northridge,California, earthquake of January 17, 1994

Contours of spatial variations of the duration of strong earthquake ground motion during the 1994 Northridge, California earthquake are smooth over distances as large as tens of kilometers. Visual comparison of those contours with the depth of sediments and with vertical offsets of the basement rocks along the faults in the Los Angeles basin are in excellent qualitative agreement with the trends predicted by the previously published empirical scaling equations of strong-motion duration. It is argued that if the source-to-station distances were measured along the three-dimensional wave paths through the sedimentary wave-guides, rather than along straight lines as is common at present, the accuracy of empirical scaling equations could be improved significantly.     

2018年2月12日河北永清4.3级地震强震动记录分析

2018年2月12日河北省廊坊市永清县发生4.3级地震,震中位于廊固凹陷内的河西务断裂附近,国家强震动台网中心获取74组强震动记录,本文对触发台站进行场地分类,结果显示大部分为Ⅰ类场地(51.35%),其次为Ⅱ类场地(35.14%)和Ⅲ类场地(13.51%)。根据5个典型台站加速度时程记录及反应谱分析本次地震强震动特征,并对各台站峰值加速度反应谱与方位角和震中距的关系进行分析。采用克里格插值方法绘制地震仪器烈度分布图,由于台站分布不均,导致极震区附近烈度影响场计算缺值现象,本文通过拟合本次地震强震动记录得到该地区地震动衰减关系,在空白区域建立空间随机假设台站进行补点插值,解决了计算缺值问题,为缺少台站记录的震区提供准确快速制作烈度分布图的思路,为震害调查和地震应急救援提供重要依据。     

Topographic site effects of Xishan Park ridge in Zigong city,Sichuan considering epicentral distance

Ground motion amplification induced by topography plays a vital role in engineering seismology. A topographic array of 8 accelerographs has been operating along the ridge in Xishan Park since 2007. The topographic site effects in Zigong city are studied based on the strong motion data of 2008 M_s 8.0 Wenchuan earthquake (the epicentral distance?=?225 km) and 2019 M_s 5.2 Zizhong earthquake (the epicentral distance?=?29 km). We compare the peak ground acceleration (PGA) of the two earthquakes and find that the PGA of Station 7#, which locates on a relatively steep slope, is amplified by 4.41 times comparing with the reference station in Zizhong earthquake, while this value is only 1.62 in Wenchuan earthquake. Fourier amplitude spectrum shows that the high frequency content of Zizhong earthquake is more abundant because of its smaller epicentral distance. By using the standard spectral ratio (SSR) method, we conclude that the amplification occurs because high-frequency ground motion is likely to resonate at small-scale features. Finally, the 3D numerical simulations are used to verify these conclusions. Our work indicates that more sophisticated numerical models need to be established for more accurate topographic site effects quantification. In addition, the influence of nearby topographic features should be considered when selecting reference stations.

     

盆地对三分量地震动持时的影响初探

利用关东盆地及其周边KiK-net台网井上台站记录的2004—2017年15次中强地震（矩震级为5.1～6.9级）构建三分量记录显著持时Ds_5-95数据库。针对该数据库,基于残差分析方法和3种水平向地震动持时参数预测方程,计算并给出事件间残差和事件内残差及其随不同类别参数的变化。在此基础上,初步探讨了水平向地震动持时预测方程应用于预测竖向地震动持时的可行性及盆地对三分量地震动持时的影响。研究结果表明,对于震源距和场地V_S30相当的情况,盆地内台站持时普遍大于盆地外台站持时,盆地内、外台站竖向地震动持时均大于水平向地震动持时;3种预测方程均可实现对盆地外台站水平向地震动Ds_5-95的合理估计,但在一定程度上低估了盆地内台站的水平向地震动Ds_5-95;3种预测方程均无法直接应用于竖向地震动持时预测。     

Evolution of accelerographs, data processing, strong motion arrays and amplitude and spatial resolution in recording strong earthquake motion

This paper presents a review of the advances in strong motion recording since the early 1930s, based mostly on the experiences in the United States. A particular emphasis is placed on the amplitude and spatial resolution of recording, which both must be ‘adequate’ to capture the nature of strong earthquake ground motion and response of structures. The first strong motion accelerographs had optical recording system, dynamic range of about 50 dB and useful life longer than 30 years. Digital strong motion accelerographs started to become available in the late 1970s. Their dynamic range has been increasing progressively, and at present is about 135 dB. Most models have had useful life shorter than 5–10 years. One benefit from a high dynamic range is early trigger and anticipated ability to compute permanent displacements. Another benefit is higher sensitivity and hence a possibility to record smaller amplitude motions (aftershocks, smaller local earthquakes and distant large earthquakes), which would augment significantly the strong motion databases. The present trend of upgrading existing and adding new stations with high dynamic range accelerographs has lead to deployment of relatively small number of new stations (the new high dynamic range digital instruments are 2–3 times more expensive than the old analog instruments or new digital instruments with dynamic range of 60 dB or less). Consequently, the spatial resolution of recording, both of ground motion and structural response, has increased only slowly during the past 20 years, by at most a factor of two. A major (and necessary) future increase in the spatial resolution of recording will require orders of magnitude larger funding, for purchase of new instruments, their maintenance, and for data retrieval, processing, management and dissemination. This will become possible only with an order of magnitude cheaper and ‘maintenance-free’ strong motion accelerographs. In view of the rapid growth of computer technology this does not seem to be (and should not be) out of our reach.     

柴达木盆地东部地震地面运动放大效应

柴达木盆地是青藏高原东北部大型断陷山间盆地,该地区的流动观测记录了2008年11月10日发生于大柴旦附近的M_W6.3地震。和附近的基岩上的记录相比,盆地内部的记录显示出非常显著的地面运动放大效应,表现为峰值速度的增大、持续时间的延长,其呈现出长持续时间的后续震相。傅里叶频谱分析表明盆地内部显著的后续震相的频率和直达波相比较低,地面质点运动轨迹图显示后续震相为面波运动特征。为了解释地面运动的差异,构建二维模型,通过交错网格高阶有限差分方法计算了地震波在盆地内部的传播过程,结果显示盆地内部低速层的存在造成直达波的放大以及多次反射与转换,盆地边缘结构造成的波的相干叠加产生了强烈的次生面波,其低频、大振幅、长持续时间的特征是盆地内部地面运动放大的主要原因。     

Determination and application of path duration of seismic ground motions based on the K-NET data in Sagami Bay,Japan

Duration models are one of the important parameters in ground-motion simulations. This model varies in different study areas, and plays a critical role in nonlinear structural response analysis. Currently, available empirical models are being globally used in ground-motion simulations, with limited research focusing on path duration in specific regions. In this study, we collected 6,486 sets of three-component strong-motion records from 29 K-NET stations in the Sagami Bay, Japan, and its surrounding areas between January 2000 to October 2018. We extracted the effective duration of 386 pieces of ground-motion records by manually picking up the S-wave arrival time and calculating the significant duration. We then obtained the path duration model of the study area based on the empirical equation of dynamic corner frequency and source duration of [7]. Compared with the results of the available empirical models, the Fourier spectrum of the simulated ground motion from our effective duration model showed higher accuracy in the long-term range, with less fitting residuals. This path duration model was then applied to simulate two earthquakes of M_W5.4 and M_W6.2, respectively, in the region using the stochastic finite-fault method with a set of reliable source, path, and site parameters determined for the study area. The simulation results of most stations fit well with observation records in the 0–30 Hz frequency band. For the M_W5.4 earthquake, the simulated ground motions at KNG005/KNG010/SZO008 stations were relatively weak in the mid to high frequency band (1–30 Hz) because the quality factor and geometric diffusion model used in the simulation were the averages of the entire Sagami Bay region, causing a bias in the results of a few stations owing to local crustal velocity anomalies and topographic effects. For the M_W6.2 earthquake, the simulated ground motions were relatively weak at all SZO and TKY stations, mainly because of the close distance from these stations to the epicenter and the complex seismic-wave propagation paths. The analysis suggests that the differences between the simulation results of the two earthquakes were mainly related to complex geological conditions and seismic-wave propagation paths.     

2018年陕西宁强5.3级地震强地面运动特征及局部场地效应分析

2018年9月12日陕西省宁强县发生5.3级地震,中国数字强震动台网的39个专业台站在此次地震中触发。文章中通过处理捕获的117条三分向加速度记录,给出近场台站的地震动参数,绘制震中附近区域峰值加速度等值线图,其长轴呈西南-东北方向展布。采用实际观测数据与几种常用地震动衰减关系对比,发现霍俊荣衰减预测模型能更好地反映此次地震的影响场。将振幅最大的51GYD台的反应谱与我国抗震设计反应谱比较,采用最小二乘法拟合出不同震中距5个台站各周期谱加速度衰减特性,总结出此次地震的反应谱基本特征。运用H/V谱比法对51GYD土层台和62ZM台阵进行局部场地地震反应分析,研究覆盖土层对地震动的放大作用,及局部地形对峰值加速度和峰值速度的影响过程。     

Response of three Athens metro underground structures in the 1999 Parnitha earthquake

The M_s 5.9 earthquake of 1999 produced valuable records in three underground structures, as follows: (a) in the just completed cut-and-cover station of Sepolia two accelerographs recorded the free-field and the station-base motion; (b) in the still under-construction tunnelled station of Monastiraki an accelerograph recorded the ground surface motion, and (c) in the nearby Kerameikos station, abandoned for non-technical reasons, the temporary prestressed-anchor piled (PAP) wall was still in place and produced a record of total seismic displacement at its top. Directly or indirectly utilising these records, the article outlines the results of numerical analyses aimed at ‘recovering’ the complete seismic response of the three underground structures. Particular emphasis is given to Sepolia station, where the developed accelerations (with PGA of about 0.17 g at the station base and 0.43 g at the station roof) are shown to have been almost exactly equal to the design accelerations according to the seismic code under the assumption that the station responds as an aboveground structure. The successful performance of the two temporary structures, in Monastiraki and Kerameikos (which had been designed against minimal acceleration levels but experienced ground-surface high-frequency accelerations of the order of 0.50 g) is explained through dynamic response analyses.     

Response spectrum predictions for potential near-field and far-field earthquakes affecting Hong Kong: rock sites

To realistically assess the seismic risk relating to built infrastructures in Hong Kong and in the neighbouring coastal cities of southern Guangdong province, it is necessary to predict ground shaking induced by different earthquake scenarios with good accuracy. A companion paper has described the modelling of the spatial and temporal distribution of the diffused seismic activities in the region, based on the newly-developed ‘Expanding Circular Disc’ (ECD) method. Representative Magnitude–Distance (M–R) combinations for both near-field and far-field earthquakes (in relation to Hong Kong) have been derived using the ECD method. The present paper describes the modelling of the response spectrum on rock sites associated with the predicted M–R combinations, using the Component Attenuation Model (CAM) that was also developed recently by the authors, based on stochastic simulations of the seismological model. The significant effects of soil resonance on the response spectrum are described in a separate publication.The accuracy of CAM in modelling ground motion properties on rock sites has been tested here by comparisons with (i) strong motions recorded in Taiwan and South China from the 1999 ‘Chi-Chi’ earthquake in Taiwan (M=7.6), (ii) motions recorded in South China from another earthquake occurring in the southern Taiwan Strait in the same year (M=5.1), and (iii) historical seismic intensity data obtained within South China. The overall capability of CAM in modelling both near-field and far-field attenuation has been shown to be unmatched by existing empirical models. Results of the comparison studies confirm the accuracy of CAM, particularly within an epicentral distance of 300–400 km.This study shows that the developed serviceability response spectra (i.e. at short return periods) are controlled mainly by the earthquake recurrence behaviour of major distant seismic sources. In contrast, the ultimate response spectra (i.e. at long return periods) relate to events with magnitudes close to the maximum credible earthquake (MCE) limit, the effect of which may also be represented by the Characteristic Response Spectrum (CRS). Both types of earthquake scenario can be significantly affected by the regional crustal properties. The proposed response spectrum envelopes have been compared with previously developed recommendations, and a critical review has been conducted. The intrinsic advantages of the ECD–CAM modelling approach have been highlighted, emphasising its directness and transparency when compared with the more complex process required to implement traditional Probabilistic Seismic Hazard Assessment (PSHA).     

Empirical correlations between cumulative absolute velocity and spectral accelerations from NGA ground motion database

Considering multiple ground motion intensity measures is important in seismic hazard analysis and ground motion selection process. Using the NGA strong motion database and recently developed ground-motion prediction models, empirical correlations are developed between cumulative absolute velocity (CAV) and spectral accelerations (Sa) at periods from 0.01 to 10 s. The CAV–Sa correlations at long periods are significantly influenced by rupture distance due to modification of the frequency content and duration of the acceleration time history through travel path. Similarly, the presence of strong velocity pulses in near-source ground motions also affects the correlations at moderate to long periods. On the other hand, the correlations are not particularly sensitive to the earthquake magnitude, orientation of the ground-motion recordings, selection of ground-motion prediction models and local site conditions. Piecewise linear fitting equations are provided to quantify the correlations for various cases. The application of the CAV–Sa correlations in ground motion selection process is also discussed.     

Earthquake hazard in Marmara Region, Turkey

Earthquake hazard in the Marmara Region, Turkey has been investigated using time-independent probabilistic (simple Poissonian) and time-dependent probabilistic (renewal) models. The study culminated in hazard maps of the Marmara Region depicting peak ground acceleration (PGA) and spectral accelerations (SA)'s at 0.2 and 1 s periods corresponding to 10 and 2% probabilities of exceedance in 50 yrs. The historical seismicity, the tectonic models and the known slip rates along the faults constitute the main data used in the assignment. Based on recent findings it has been possible to provide a fault segmentation model for the Marmara Sea. For the main Marmara Fault this model essentially identifies fault segments for different structural, tectonic and geometrical features and historical earthquake occurrences. The damage distribution and pattern of the historical earthquakes have been carefully correlated with this fault segmentation model. The inter-event time period between characteristic earthquakes in these segments is consistently estimated by dividing the seismic slip estimated from the earthquake catalog by the GPS-derived slip rate of 22±3 mm/yr. The remaining segments in the eastern and southern Marmara region are also identified using recent geological, geophysical studies and historical earthquakes. The model assumes that seismic energy along the segments is released by characteristic earthquakes. For the probabilistic studies characteristic earthquake based recurrence relationships are used. Assuming normal distribution of inter-arrival times of characteristic earthquakes, the ‘mean recurrence time’, ‘covariance’ and the ‘time since last earthquake’ are developed for each segment. For the renewal model, the conditional probability for each fault segment is calculated from the mean recurrence interval of the characteristic earthquake, the elapsed time since the last major earthquake and the exposure period. The probabilities are conditional since they change as a function of the time elapsed since the last earthquake. For the background earthquake activity, a spatially smoothed seismicity is determined for each cell of a grid composed of cells of size 0.005°×0.005°. The ground motions are determined for soft rock (NEHRP B/C boundary) conditions. Western US-based attenuation relationships are utilized, since they show a good correlation with the attenuation characteristics of ground motion in the Marmara region. The possibility, that an event ruptures several fault segments (i.e. cascading), is also taken into account and investigated by two possible models of cascading. Differences between Poissonian and renewal models, and also the effect of cascading have been discussed with the help of PGA ratio maps.     

Internal waves and temperature fronts on slopes

Time series measurements from an array of temperature miniloggers in a line at constant depth along the sloping boundary of a lake are used to describe the ‘internal surf zone” where internal waves interact with the sloping boundary. More small positive temperature time derivatives are recorded than negative, but there are more large negative values than positive, giving the overall distribution of temperature time derivatives a small negative skewness. This is consistent with the internal wave dynamics; fronts form during the up-slope phase of the motion, bringing cold water up the slope, and the return flow may become unstable, leading to small advecting billows and weak warm fronts. The data are analysed to detect ‘events’, periods in which the temperature derivatives exceed a set threshold. The speed and distance travelled by ‘events’ are described. The motion along the slope may be a consequence of (a) instabilities advected by the flow (b) internal waves propagating along-slope or (c) internal waves approaching the slope from oblique directions. The propagation of several of the observed ‘events’ can only be explained by (c), evidence that the internal surf zone has some, but possibly not all, the characteristics of the conventional ‘surface wave’ surf zone, with waves steepening as they approach the slope at oblique angles.     

Empirical scaling of Fourier spectrum amplitudes of recorded strong earthquake accelerations in terms of Modified Mercalli Intensity, local soil conditions and depth of sediments

Empirical scaling equations of Fourier spectrum amplitudes in terms of the Modified Mercalli Intensity (MMI), local soil conditions (‘rock’, stiff soil, and deep soil), and local geologic conditions (depth of sediments) are presented. It is shown that both soil and geological site effects should be used together in estimation of the site specific Fourier amplitude spectra.     

Canyon topography effects on ground motion at Feitsui damsite

The free-field accelerograms along Feitsui Canyon are analyzed and modeled by a numerical scheme to study the effect of canyon topography. Since six strong-motion accelerometers (SC1–SC6) were deployed along the Feitsui Canyon in 1991; there are 14 earthquakes (4.9≤M_L≤6.6) recorded by these stations until June 1996, but only five triggered all six stations. The maximum PGA value is 68.6 cm s⁻² recorded at station SC1. According to the present data, the effect of the dam on the ground motions at canyon stations can be negligible. The amplitude of ground motion on the slopes of the canyon is bigger than that at its trough. The integral equation method is applied to a two dimensional model of Feitsui Canyon to study the effects of the canyon topography. We choose the ground motion of SC3 or SC4 station at the trough of the canyon as the input motion for the model, which is then used to predict the ground motion at the other five stations. Apart from the earthquake close to the damsite, the simple model can reproduce the observed accelerations at all frequencies below 4 Hz. Overall, the numerical method can well predict the ground motion along the canyon, although the high-frequency simulation is underestimated.     

Secondary phase validation—Phase classification by polarization

A long-standing problem in operational seismology is that of reliable focal depth estimation. Standard analyst practice is to pick and identify a ‘phase’ in the P-coda. This picking will always produce a depth estimate but without any validation it cannot be trusted. In this article we ‘hunt’ for standard depth phases like pP, sP and/or PmP but unlike the analyst we use Bayes statistics for classifying the probability that polarization characteristics of pickings belong to one of the mentioned depth phases given preliminary epicenter information. In this regard we describe a general-purpose PC implementation of the Bayesian methodology that can deal with complex nonlinear models in a flexible way. The models are represented by a data-flow diagram that may be manipulated by the analyst through a graphical-programming environment. An analytic signal representation is used with the imaginary part being the Hilbert transform of the signal itself. The pickings are in terms of a plot of posterior probabilities as a function of time for pP, Sp or PmP being within the presumed azimuth and incident angle sectors for given preliminary epicenter locations. We have tested this novel focal depth estimation procedure on explosion and earthquake recordings from Cossack Ranger II stations in Karelia, NW Russia, and with encouraging results. For example, pickings deviating more than 5° off ‘true’ azimuth are rejected while Pn-incident angle estimate exhibit considerable scatter. A comprehensive test of our approach is not quite easy as recordings from so-called Ground Truth events are elusive.     

Modelling F2-layer seasonal trends and day-to-day variability driven by coupling with the lower atmosphere

This paper presents results from the TIME-GCM-CCM3 thermosphere–ionosphere–lower atmosphere flux-coupled model, and investigates how well the model simulates known F2-layer day/night and seasonal behaviour and patterns of day-to-day variability at seven ionosonde stations. Of the many possible contributors to F2-layer variability, the present work includes only the influence of ‘meteorological’ disturbances transmitted from lower levels in the atmosphere, solar and geomagnetic conditions being held at constant levels throughout a model year.In comparison to ionosonde data, TIME-GCM-CCM3 models the peak electron density (NmF2) quite well, except for overemphasizing the daytime summer/winter anomaly in both hemispheres and seriously underestimating night NmF2 in summer. The peak height hmF2 is satisfactorily modelled by day, except that the model does not reproduce its observed semiannual variation. Nighttime values of hmF2 are much too low, thus causing low model values of night NmF2. Comparison of the variations of NmF2 and the neutral [O/N₂] ratio supports the idea that both annual and semiannual variations of F2-layer electron density are largely caused by changes of neutral composition, which in turn are driven by the global thermospheric circulation.Finally, the paper describes and discusses the characteristics of the F2-layer response to the imposed ‘meteorological’ disturbances. The ionospheric response is evaluated as the standard deviations of five ionospheric parameters for each station within 11-day blocks of data. At any one station, the patterns of variability show some coherence between different parameters, such as peak electron density and the neutral atomic/molecular ratio. Coherence between stations is found only between the closest pairs, some 2500 km apart, which is presumably related to the scale size of the ‘meteorological’ disturbances. The F2-layer day-to-day variability appears to be related more to variations in winds than to variations of thermospheric composition.     

Earthquake ground motion modeling of induced seismicity in the Groningen gas field

A physics‐based numerical approach is used to characterize earthquake ground motion due to induced seismicity in the Groningen gas field and to improve empirical ground motion models for seismic hazard and risk assessment. To this end, a large‐scale (20 km × 20 km) heterogeneous 3D seismic wave propagation model for the Groningen area is constructed, based on the significant bulk of available geological, geophysical, geotechnical, and seismological data. Results of physics‐based numerical simulations are validated against the ground motion recordings of the January 8, 2018, M_L 3.4 Zeerijp earthquake. Taking advantage of suitable models of slip time functions at the seismic source and of the detailed geophysical model, the numerical simulations are found to reproduce accurately the observed features of ground motions at epicentral distances less than 10 km, in a broad frequency range, up to about 8 Hz. A sensitivity analysis is also addressed to discuss the impact of 3D underground geological features, the stochastic variability of seismic velocities and the frequency dependence of the quality factor. Amongst others, results point out some key features related to 3D seismic wave propagation, such as the magnitude and distance dependence of site amplification functions, that may be relevant to the improvement of the empirical models for earthquake ground motion prediction.