Specific Barrier Model Research Articles

Near-Fault and Far-Field Strong Ground-Motion Simulation for Earthquake Engineering Applications Using the Specific Barrier Model

Codes for aseismic design may require use of recorded ground motions as input in dynamic analysis. When records are not available, motions must be simulated. The specific barrier model (SBM) is particularly useful in this context because (1) it provides the most complete, yet parsimonious, self-consistent description of the earthquake faulting processes that are responsible for the generation of high-frequency radiation; (2) it has been calibrated to earthquakes of three different tectonic regions; and (3) its key parameter, the barrier interval, is related to the duration of the near-fault pulses (NFP), the most damaging feature of near-fault motions. We carry out “blind” (i.e., using the minimum amount of a priori source information) simulations of strong motions of well-recorded earthquakes of magnitudes between 6.2–7.6. We assess the quality of fit of the simulated time histories to the recorded motions and show that the simulations exhibit close to zero bias over frequencies of 0.1–20 Hz for the data set used. This exercise illustrates that the method will provide earthquake motions that can be used with confidence in aseismic design.

Analysis of Iranian strong-motion data using the specific barrier model

Two decades after the development of the specific barrier model (SBM) by Papageorgiou and Aki (1983a Bull. Seism. Soc. Am. 73 693–722, 1983b Bull. Seism. Soc. Am. 73 953–78), it remains to be the most sophisticated, yet simple, theoretical description of the earthquake faulting process. In this study, strong-motion accelerograms recorded in Iran are used in the context of 'random vibration theory' to calibrate the SBM. The database consists of 136 three-component records from 19 earthquakes of magnitude ranging from Mw 5.2 to Mw 7.4, recorded at hypocentral distances up to 200 km. Regression analysis has been conducted for pseudospectral velocity (PSV) using the 'random-effects' model that accounts for correlations in the data recorded by a single earthquake. A rough estimation of site amplification functions, is first obtained by the incorporation of mean frequency-dependent site-amplification factors, based on a gross characterization of the site class. However, a repetition of regression analysis with the use of a site-specific amplification function estimated by the horizontal-to-vertical ratio technique improved the fit to the observed amplitudes over the lower frequencies. The average values of the local and global stress drops, which are the two key parameters of the SBM, are determined to be 90 and 44 bars, respectively. The predicted PSV values agree well with available Iranian strong-motion data, as evidenced by the near-zero average of differences between the logarithms of the observed and predicted values (residuals) for all frequencies and the lack of any significant residual trends with distance and magnitude. A comparison of the SBM source spectra of this study with those of California and eastern North America (ENA) reveals that the spectral amplitudes are more like typical Californian source models than ENA in the frequency range studied here, from 1 to 10 Hz. Moreover, the calibrated model of this study which provides an efficient and physical representation of Iranian plateau source characteristics is found to be in reasonably good agreement with other local and regional attenuation relationships. Investigation of the residuals showed only a weak influence of soil nonlinearity; however, because of the relatively weak levels of acceleration in our database, it is not yet adequate to clearly distinguish this effect.

Calibration of the specific barrier model to Iranian plateau earthquakes and development of physically based attenuation relationships for Iran

Earthquake ground-motion relationships for soil and rock sites in Iran have been developed based on the specific barrier model (SBM) used within the context of the stochastic modeling and calibrated against up-to-date Iranian strong-motion data. A total of 171 strong-motion accelerograms recorded at distances of up to 200 km from 24 earthquakes with moment magnitudes ranging from Mw 5.2 to 7.4 are used to determine the region-specific source parameters of this model. Regression analysis was conducted using the “random effects” methodology that considers both earthquake-to-earthquake (inter-event) variability and within-earthquake (intra-event) variability to effectively handle the problem of weighting observations from different earthquakes. The minimization of the error function in each iteration of the “random effects” procedure was performed using the genetic algorithm method. The residuals are examined against available Iranian strong-motion data to confirm that the model predictions are unbiased and that there are no significant residual trends with distance and magnitude. No evidence of self-similarity breakdown is observed between the source radius and its seismic moment. To verify the robustness of the results, tests were performed to confirm that the results are unchanged if the number of observations is changed by removing different randomly selected datasets from the original database. Stochastic simulations, using the derived SBM, are then performed to predict peak ground-motion and response spectra parameters for a wide range of magnitudes and distances. The stochastic SBM predictions agree well with the new empirical regression equations proposed for Iran, Europe and Middle East in the magnitude–distance ranges well represented by the data. It has been shown that the SBM of this study provides unbiased ground-motion estimates over the entire frequency range of most engineering interests (1–10 Hz) for the Iranian earthquakes. Our results are also important for the assessment of hazards in other seismically active environments in the Middle East and Mediterranean regions.

A Fast and Efficient Simulation of the Far-Fault and Near-Fault Earthquake Ground Motions Associated with the June 17 and 21, 2000, Earthquakes in South Iceland

The two Mw 6.5 earthquakes on June 17 and 21, 2000, respectively, in the populated South Iceland Seismic Zone (SISZ) significantly augmented the Icelandic database of strong ground motions, and several strong velocity pulses were recorded at near-fault sites. The strong motions are interpreted via the Specific Barrier Model (SBM) and a mathematical model of near-fault velocity pulses. The data indicates self-similar source scaling and significantly greater attenuation of seismic waves than in other interplate regions. Through inversion of the data a new attenuation function for the SISZ has been adopted, which results in unbiased simulations. For the first time, the characteristics of the recorded near-fault pulses have been identified and compared to the worldwide database of such records. The SBM and the near-fault pulse model combine naturally in a fast and efficient synthesis of realistic, broad-band strong ground motions in the far-fault and near-fault region. Such simulations are showcased for the June 2000 earthquakes and indicate that the modeling approach adopted in this study is an effective tool for the estimation of realistic earthquake ground motions in the SISZ.

Calibration of the Specific Barrier Model to Earthquakes of Different Tectonic Regions

The specific barrier model, proposed and developed by Papageorgiou and Aki (1983a,b; 1985) provides the most complete, yet parsimonious, self-consis- tent description of the faulting process. It applies both in the near-fault and in the far-field region, thus allowing for consistent ground-motion simulations over the entire frequency range and for all distances of engineering interest. The model has been implemented in the stochastic method and calibrated with extended databases of response spectral amplitudes from earthquakes of intraplate regions (mainly east- ern North America events), interplate regions, and regions of tectonic extension (Spu- dich et al., 1999, database). The ensemble average value of a key parameter of the specific barrier model, the local stress drop Dr L ,i s� 161 bars for interplate earth- quakes, � 114 bars for extensional regime earthquakes, and � 180 bars for intraplate earthquakes. The high-frequency source spectral levels of interplate and extensional regime earthquakes deviate significantly from self-similar scaling. The deviation is most likely caused by the effective source area and/or irregularities in the rupture kinematics. We account for their overall effects through a high-frequency source complexity factor, f, in the source spectrum of the specific barrier model. As a result, inter- and intraplate source spectra show similar high-frequency levels at moderate magnitudes but intraplate earthquakes have higher spectral levels at the larger mag- nitudes. The interplate soil residuals show clear signs of nonlinear site response, whereas only slight signs of such nonlinearity are observed for the extensional data- set. The regional models calibrated in this study are in reasonably good agreement with other regional attenuation relationships and provide a reliable and physically realistic, yet computationally efficient, way to model strong ground motions with implications for seismic hazard and risk analysis. Online material: Strong-motion station and event-station pair information.

Engineering Ground Motion Modeling in the Near-Source Regime Using the Specific Barrier Model for Probabilistic Seismic Hazard Analysis

—The specific barrier model is used for the first time to simulate ground motion accelerations for the purpose of probabilistic seismic hazard analyses at sites near a dominant fault system. It incorporates the simulation of fault geometry and the relationship between the stress drop and seismic moment to estimate the number of cracks on the fault for the specific barrier model. Radiated direct shear waves are established following Boore’s (1983) procedure. The simulated peak ground accelerations (PGA) are then calibrated by strong-motion data. Basically, the model is of uniform source, and the directivity of the source is not taken into consideration. The results show that the calibrated PGA values are not sensitive to the relationship between the stress drop and seismic moment. However, the calibrated PGA values may increase about 20 percent for sites near the fault when the cut-off frequency, fmax , is raised from 5 Hz to 10 Hz. The variability of the simulated ground motion is, in general, smaller than that of the empirical strong-motion data shown in the literature. This may be improved by adding randomness into the parameter of fmax and uncertainties into the empirical relationships adopted in the model. The simulated attenuation curves may be used to judge which types of conventional attenuation equations are better at representing the attenuation of PGA for sites near the fault, especially for large earthquake events.

Higher-order interrelations between seismogenic structures and earthquake processes

Aki, K., 1992. Higher-order interrelations between seismogenic structures and earthquake processes. In: T. Mikumo, K. Aki, M. Ohnaka, L.J. Ruff and P.K.P. Spudich (Guest-Editors), Earthquake Source Physics and Earthquake Precursors. Tectonophysics, 211: 1–12. To advance our understanding of higher-order (smaller-scale) interrelations between the fault zone structure and the earthquake process we start with the observation that the source parameters of major California earthquakes deviate systematically from the self-similarity. This departure from the self-similarity is interpreted in terms of both the specific barrier model and the asperity model. The asperity model, however, is found to be inconsistent with the following observations: (1) sharp impulsive displacements waveform in the direction perpendicular to the fault trace; (2) the complementary relationship between fault slip and aftershock; and (3) existence of barriers in the creeping segment of the San Andreas fault. The application of the specific barrier model, on the other hand, leads to a predominant subevent crack size of a few hundred meters in the creeping segment. This result is consistent with the negative correlation between coda Q −1 and b-value observed in the Stone Canyon area for 1.5 < M < 4, and suggests that the creeping brittle region may be the source of temporal change in coda Q −1. The existence of an extensive creeping brittle zone under southern California with the same predominant crack size can explain the observed strong positive correlation between the coda Q −1 and b-value for M > 3. Temporal increase in coda Q −1 is sometimes accompanied by an increase in the rate of surface deformation, but not always so. This is attributed to the erratic coupling between the creeping brittle region and the overlying brittle region. The applicability of the conclusions obtained from California earthquakes to other regions is investigated using the observed scaling laws of seismic source spectra, and it is suggested that the crack size of the creeping brittle zone may vary regionally from a few hundred meters to a few kilometers. As an alternative to the widely held view that the aseismic creep in the plate boundary may occur as a giant slow earthquake, our result suggests that it may be occurring as small fractures with scale lengths from a few hundred meters to a few kilometers.

AN EMPIRICAL SCALING OF STRONG-MOTION SPECTRA WITH APPLICATION TO ESTIMATE OF SOURCE SPECTRA

A new regression model of strong-motion spectra is derived by incorporating the conception of the dummy variables to the conventional regression model. The dummy variables are available not only to obtain the amplification factors due to local soil conditions of each observation site but also to estimate the non-linear dependences of spectral amplitude on earthquake magnitude and hypocentral distance. The new regression model is applied to 228 strong-motion spectra obtained in Japan. The regression results are used to estimate the source spectra, as well as being discussed in detail, in particular, from the view points of faulting and wave propagation mechanisms. It is finally coneluded that the specific barrier model is more connected with the present statistical source spectra than the stochastic ω-square model.

On two characteristic frequencies of acceleration spectra: Patch corner frequency andfmax

AbstractWe examine two issues related to the radiation of tectonic earthquake sources: (i) the complexity of the shape of observed source spectra, and (ii) the effect of recording site characteristics on the cut-off frequency fmax of acceleration spectra beyond which spectral amplitudes diminish sharply.We compute the far-field spectra of the 1979 Imperial Valley and the 1984 Morgan Hill earthquakes from the slip functions that were inferred from detailed inversion studies of strong motion and teleseismic data of these two California strike-slip events. These spectra, along with data from the 1983 Central Japan Sea earthquake sequence (main shock and aftershocks), demonstrate the existence of the so-called patch corner frequency, which is related to the scale-length of heterogeneities of the fault plane. Such a frequency is a distinct feature of the source spectrum of the specific barrier model of Papageorgiou and Aki.By analyzing accelerograms of the San Fernando earthquake of 1971, we found that fmax determined from records at soil stations is nearly the same as that determined from records at hard rock stations.

Sealing law of far-field spectra based on observed parameters of the specific barrier model

A set of acceleration source spectra is constructed using the observed parameters of the specific barrier model of Papageorgiou and Aki. The spectra show a significant departure from the ‘ω2-model’ at the high frequency range. Specifically, the high frequency spectral amplitudes of seismic excitation are higher as compared to the level predicted by the ‘ω2-model’. This is also supported by other observational evidence. The high frequency amplitudes of acceleration scale proportionally to the square root of the rupture areaS, to the rupture spreading velocityv, and to the local strain drop (Δσ/μ) (=strain drop in between barriers). The local strain drop in between barriers is not related in a simple way to the global strain drop, which is the strain drop estimated by assuming that it is uniform over the entire rupture area. Consequently, the similarity law does not apply. Using the source spectra which we constructed, we derive expressions for high frequency amplitudes of acceleration such asa rms anda max. Close to the fault both are independent of fault dimensions and scale as (Δσ/µ)(Δf)1/2, while away from the fault plane they scale asW 1/2(Δσ/µ)(Δf)1/2, whereW is the width of the fault and Δf is the effective bandwidth of the spectra.

Specific barrier model for the quantitative description of inhomogeneous faulting and the prediction of strong ground motion. Part I. Description of the model : Papageorgiou, A S; Aki, K Bull Seismol Soc AmV73, N3, June 1983, P693–722

Specific barrier model for the quantitative description of inhomogeneous faulting and the prediction of strong ground motion. Part I. Description of the model : Papageorgiou, A S; Aki, K Bull Seismol Soc AmV73, N3, June 1983, P693–722

Specific barrier model for the quantitative description of inhomogeneous faulting and the prediction of strong ground motion. Part 2. Applications of the model : Papageorgiou, A S; Aki, K Bull Seismol Soc AmV73, N4, Aug 1983, P953–978

Specific barrier model for the quantitative description of inhomogeneous faulting and the prediction of strong ground motion. Part 2. Applications of the model : Papageorgiou, A S; Aki, K Bull Seismol Soc AmV73, N4, Aug 1983, P953–978

A specific barrier model for the quantitative description of inhomogeneous faulting and the prediction of strong ground motion. Part II. Applications of the model

abstractThe specific barrier model, which was described in detail by Papageorgiou and Aki (1983), is applied to a set of five moderate to strong Californian earthquakes: Kern County (1952); San Fernando (1971); Borrego Mountain (1968); Long Beach (1933); and Parkfield (1966). Source parameters such as barrier intervals, local stress drops, cohesive zone size, and cohesive stress are inferred. The analysis of the San Fernando earthquake of 1971 revealed a strong frequency dependence of Qβ, suggesting that the high frequencies may not be as strongly attenuated as initially thought. This suggests that the fall-off at high frequencies of observed spectra at a site may not be a propagation path effect but primarily a source effect. The cut-off frequency observed on the source acceleration power spectra of all the events analyzed in this paper, is interpreted in terms of the cohesive zone size and cohesive stress which arrest the localized fractures that occur on the fault plane. It was found that the barrier interval, as inferred by the specific barrier model, increases with the increase in maximum slip. The ratio of the latter to the former represents the local strain drop and was found to increase slightly (factor of 2) with earthquake size. This verifies the relation between barrier interval and maximum slip which has been observed by Aki et al. (1977) who inferred the barrier interval by different methods (i.e., surface measurement of fault slip, seismic measurement of rise time, scaling law of seismic spectrum). Striking similarities with respect to source parameters were found between the Fort Tejon (1857) and Kern County (1952) earthquakes as well as between the Long Beach (1933) and Parkfield (1966) earthquakes. The former are characterized by long barrier intervals and large slips while the latter are characterized by short barrier intervals and small slips. San Fernando (1971) and Borrego Mountain (1968) earthquakes lie in between these two extremes.