Figure 2 - uploaded by Katrin Kieling
Content may be subject to copyright.
The locations of (a) the 2008 Wenchuan and (b) the 2003 Tokachi-Oki earthquakes. Triangles and circles indicate the strongmotion and GPS stations, respectively. Stations to which simulations are compared in Figures 3, 6, and 9 are marked by filled symbols. The black star marks the hypocenter. The color version of this figure is available only in the electronic edition.
Source publication
The simulation of realistic ground motions associated with large earthquakes is of utmost importance for engineering concerns, e. g. testing the seismic performance of buildings. It is also needed to estimate the level of ground shaking for future earthquakes. In this paper, we focus on two aspects of ground motion modeling. First, we use determini...
Context in source publication
Context 1
... the following subsections, we present applications of our methodology for the cases of the 2008 Wenchuan and the 2003 Tokachi-Oki earthquakes. Maps of the two earthquakes and the strong-motion and Global Positioning System (GPS) stations used are shown in Figure 2. We also aim to validate the rise-time scaling approach by comparing simulations of the Wenchuan earthquake with varied rupture velocity. ...
Similar publications
Physics-based earthquake ground-motion simulation, also referred to as deterministic earthquake ground-motion simulation, can be defined as the prediction of the ground motion generated by earthquakes by means of numerical methods and models that incorporate explicitly the physics of the earthquake source and the resulting propagation of seismic wa...
This paper develops a simplified method to quantify the effects of liquefaction-induced lateral spreading on bridge responses, expressed as the response modification factor of the column drift ratio under seismic shaking. Using the newly developed global dynamic p-y analysis procedure, nonlinear time history responses were obtained for a benchmark...
Citations
... This relation has been used for pseudo dynamic rupture simulations [70]. This result also allows us to compare energy-based kinematic relations (equations (2) and (5)- (7)) to the relation ∝ τ u τ Δ / R 2 , where u is the slip and τ R the rise time [71]. From equation (7), the rupture velocity is . ...
The lack of clarity regarding slip distribution within heterogeneous rupture areas has a significant impact on characterizing the seismic source and the role of heterogeneities in determining ground motion. One approach to understand the rupture process is through dynamic simulations, which require substantial computational resources, thereby limiting our comprehension of seismic rupture processes. Consequently, there is a need for methods that efficiently describe the spatial complexities of seismic rupture in a realistic manner. To address this, the statistics of real self-arrested ruptures that conform to the asperity criterion are investigated. This research demonstrates that power law distributions can describe the final slip statistics. Regarding the computational efficiency, a simple heterogeneous energy-based (HE-B) method is proposed. The HE-B method is characterized by the spatial correlation between the rupture parameters, such as the final slip or the rupture velocity, and the distribution of residual energy which determines the zones where the rupture could occur. In addition, the HE-B method defines the rupture area in those zones of the fault where the coupling function exceeds the energy required for rupture initiation. Therefore, the size of the earthquake is directly influenced by the distribution of coupling within faults. This method also leads to the successful reproduction of the statistical characteristics of final slip and generates slip rates that match the kinematic behavior of seismic sources. Notably, this kinematic rupture simulation produces seismic moment rates characterized by f − 1 {f}^{-1} and f − 2 {f}^{-2} spectra with a double corner frequency. Finally, it is observed that the maximum fracture energy value within the ruptured area is strongly correlated with both the magnitude and peak seismic moment rate. Thus, by employing this method, realistic rupture scenarios can be generated efficiently, enabling the study of spatial correlations among rupture parameters, ground motion simulations, and quantification of seismic hazard.
... A more realistic hybrid source model is valuable for the prediction of ground motion from strong earthquakes, especially in the absence of a sufficient understanding of small-scale inhomogeneities within the regional crust. Similar modelling approaches have been developed and used by many researchers [45][46][47] to obtain broadband synthetics of recent earthquakes. ...
... The ground motion synthesis approach based on the frequency-wavenumber Green's function (FK approach) can provide three-dimensional motions [1][2][3][4], and thus it is a promising technique for the multidimensional broadband ground motion synthesis. The FK approach obtains broadband ground motion by convolving the deterministic full-waveform Green's function with the appropriate kinematic rupture process. ...
We recently propose a method modeling kinematic source of a scenario earthquake for synthesis of broadband ground motion based on the frequency-wavenumber Green's function (FK approach). The kinematic source model describes the spatiotemporal rupture process with a set of source parameters, and the parameters may have different influences on the low-and high-frequency components of the synthetic ground motion. To explore the main influencing parameters at low and high frequencies respectively, this study quantitatively analyzes the sensitivity of the synthetic motion to five types of source parameters by case studies: (1) those on the occurrence of the rupture plane, including the depth and dip of the fault; (2) on the spatial variation of slip, including the size of the rupture plane, and the slip distribution; (3) on the temporal evolution of slip, including the rise time, rupture velocity and their correlation with slip, and the source time function; (4) on the randomness, including the perturbation of the rupture time, and the fault roughness expressed by the spatial randomness of slip direction, dip and strike; and (5) others, including the stress drop, the subsource size, and the constraints over the entire rupture. The results show that the low-and high-frequency ground motions are controlled by different source parameters. Generally, the temporal evolution of slip is the dominant influencing factor at high frequency, whereas the spatial variation of slip is the main influencing factor at low frequency. This study contributes to the application of the FK approach in the synthesis of broadband ground motion.
... Recent theoretical studies show that geometrical variations of the fault surface lead to ensembles of complex rupture with varying directionalities and magnitudes (e.g., Fang & Dunham, 2013), modifying the radiated seismic spectrum and radiation pattern. In particular, those complexities boost high-frequency ground motions (Dunham et al., 2011b), bringing them into general accord with the aggregated statistics of recorded motions, as reflected in GMPEs (e.g., Kieling et al., 2014;Shi & Day, 2013;. The previous rough fault modeling studies have not focused on the dynamics of low-frequency pulses, nor on the degree to which those pulses are sensitive to off-fault plasticity. ...
Near‐fault motion is often dominated by long‐period, pulse‐like particle velocities with fault‐normal polarization that, when enhanced by directivity, may strongly excite middle‐ to high‐rise structures. We assess the extent to which plastic yielding may affect amplitude, frequency content, and distance scaling of near‐fault directivity pulses. Dynamic simulations of 3‐D strike‐slip ruptures reveal significant plasticity effects, and these persist when geometrical fault roughness is added. With and without off‐fault yielding, these models (~M 7) predict fault‐normal pulse behavior similar to that of observed pulses (periods of 2–5 s, amplitudes increasing with rupture distance until approaching a limit), but yielding systematically reduces pulse amplitude and increases the dominant period. Yielding causes near‐fault (< ~2 km) peak ground velocity (PGV) to saturate with respect to increases in both stress drop and epicentral distance, and, in that distance range, yielding may contribute significantly to the observed magnitude saturation of PGV. The results support the following elements for functional forms in empirical pulse models: (i) a fault‐normal distance saturation factor, (ii) a period‐dependent and along‐strike distance‐dependent factor representing directivity, and (iii) an along‐strike saturation factor to truncate growth of the directivity factor. In addition to the foregoing effects on long‐period fault‐normal pulses, the model with off‐fault plasticity is very efficient in suppressing the high‐frequency fault‐parallel acceleration pulses that otherwise develop when rupture breaks free surface. This effect is likely to inhibit the initiation of a sustained supershear rupture triggered by a strong free surface breakage.
... The deterministic and stochastic methods usually describe the spatiotemporal rupture process in different ways, and some source parameters of the two methods may be inconsistent. The synthesis approach based on the frequency-wavenumber (f-k) Green's function provides 3D motions (Hartzell et al., 2005;Kieling et al., 2014;Crempien and Archuleta, 2015) by convolving the deterministic full-waveform Green's function with the appropriate kinematic rupture process. A kinematic source model specifies the rupture process of an earthquake scenario as a function of both space and time while incorporating many aspects of the rupture dynamics (Guatteri et al., 2003) and significantly influences the effective bandwidth of the synthetics (Cao and Tao, 2018). ...
... To determine the spatial variation of slip, Irikura and Miyake (2011) developed a fully deterministic asperitybased rupture modeling approach, and Graves and Pitarka (2016) adopted a semistochastic approach to realize a random slip field that was filtered to have a roughly k-squared fall-off (Mai and Beroza, 2002). Many other researchers (e.g., Kieling et al., 2014;Crempien and Archuleta, 2015;Frankel, 2017) applied similar approaches. Deterministic modeling emphasizes the effect of asperity, which represents the main area on the rupture plane that radiates seismic-wave energy, whereas semistochastic modeling emphasizes the effect of random scattering. ...
... STFs compatible with slip-rate histories obtained with rupture dynamics models are generally preferable (Dreger et al., 2007). In addition, most studies independently determine the rise time and rupture velocity (e.g., Pitarka, 2010, 2016;Kieling et al., 2014;Frankel, 2017). However, an allowable combination of rise time, rupture velocity, and slip still constitutes an issue in kinematic source modeling because the source parameters in the rupture process must be coordinated (Konca et al., 2007). ...
A procedure for building a kinematic source model is proposed in this article for the synthesis of broadband ground motion based on the frequency–wavenumber Green’s function. The spatial distribution of slip on the rupture plane is generated by combining asperity slip with random slip. A set of scaling laws recently updated for the global and local parameters of seismic sources is adopted. To characterize the temporal evolution of slip on the rupture plane, different rupture velocities, and rise times are first generated by considering the correlation with slip, and a source time function obtained by rupture dynamics is selected for each subsource. Then, the entire rupture process is set as the object to jointly determine the rise time and rupture velocity for a given slip distribution under the selection criterion that the entire rupture process should radiate the closest seismic energy to the expected energy. To reduce uncertainty, 30 spatiotemporal rupture processes for an earthquake scenario are realized to select a mean source model. To demonstrate the feasibility of the proposed source modeling approach, two California earthquakes, the Whittier Narrows earthquake and the Loma Prieta earthquake, are chosen as case studies. The performance of the obtained source models shows that our modeling approach is advantageous for estimating the size of the rupture plane, emphasizing the effect of asperity, and considering the correlation between temporal rupture parameters and slip. The bias values between the observed and synthetic pseudospectral accelerations are relatively small compared to those for the methods on the Southern California Earthquake Center broadband platform. The synthetics are further compared with the estimates from regional ground‐motion prediction equations for four scenario earthquakes with moment magnitudes of 6.0, 6.5, 7.0, and 7.5. Finally, the sensitivity of the synthetic motion to various rupture parameters is analyzed.
... The dynamic simulations produce a strong anticorrelation of roughness-induced fluctuations in final slip, rupture velocity, and peak slip velocity with the local gradient of the fault surface for strike-slip ruptures. The incorporation of stochastic strike and dip perturbations related to fault roughness has been done for kinematic simulations of the 2003 Tokachi-Oki, Japan and the 2008 Wenchuan, China, earthquakes (Kieling et al., 2014). The simulations produce a reasonable match to the geometric mean ground-motion levels up to 8 Hz; however, near-fault radiation-pattern effects at high frequencies were not addressed in this study. ...
We describe a methodology for generating kinematic earthquake ruptures for use in 3D ground-motion simulations over the 0–5 Hz frequency band. Our approach begins by specifying a spatially random slip distribution that has a roughly wavenumber-squared fall-off. Given a hypocenter, the rupture speed is specified to average about 75%–80% of the local shear wavespeed and the prescribed slip-rate function has a Kostrov-like shape with a fault-averaged rise time that scales selfsimilarly with the seismic moment. Both the rupture time and rise time include significant local perturbations across the fault surface specified by spatially random fields that are partially correlated with the underlying slip distribution. We represent velocity-strengthening fault zones in the shallow (<5 km) and deep (>15 km) crust by decreasing rupture speed and increasing rise time in these regions. Additional refinements to this approach include the incorporation of geometric perturbations to the fault surface, 3D stochastic correlated perturbations to the P- and S-wave velocity structure, and a damage zone surrounding the shallow fault surface characterized by a 30% reduction in seismic velocity. We demonstrate the approach using a suite of simulations for a hypothetical Mw 6.45 strike-slip earthquake embedded in a generalized hard-rock velocity structure. The simulation results are compared with the median predictions from the 2014 Next Generation Attenuation-West2 Project ground-motion prediction equations and show very good agreement over the frequency band 0.1–5 Hz for distances out to 25 km from the fault. Additionally, the newly added features act to reduce the coherency of the radiated higher frequency (f >1 Hz) ground motions, and homogenize radiation-pattern effects in this same bandwidth, which move the simulations closer to the statistical characteristics of observed motions as illustrated by comparison with recordings from the 1979 Imperial Valley earthquake.
... Soc. Kieling et al. (2014). The software Genso , which I implemented for the generation of random or refined slip distributions, has also been used by other researchers recently, e.g by Cattania et al. (2014) and Bach (2013). ...
... Simulations of the 2008 Wenchuan earthquake and the 2003 Tokachi earthquake have been published in Kieling et al. (2014). However, here, I additionally implement the high-frequency site attenuation with parameter κ. ...
In many procedures of seismic risk mitigation, ground motion simulations are needed to test systems or improve their effectiveness. For example they may be used to estimate the level of ground shaking caused by future earthquakes. Good physical models for ground motion simulation are also thought to be important for hazard assessment, as they could close gaps in the existing datasets. Since the observed ground motion in nature shows a certain variability, part of which cannot be explained by macroscopic parameters such as magnitude or position of an earthquake, it would be desirable that a good physical model is not only able to produce one single seismogram, but also to reveal this natural variability.
In this thesis, I develop a method to model realistic ground motions in a way that is computationally simple to handle, permitting multiple scenario simulations. I focus on two aspects of ground motion modelling. First, I use deterministic wave propagation for the whole frequency range – from static deformation to approximately 10 Hz – but account for source variability by implementing self-similar slip distributions and rough fault interfaces. Second, I scale the source spectrum so that the modelled waveforms represent the correct radiated seismic energy. With this scaling I verify whether the energy magnitude is suitable as an explanatory variable, which characterises the amount of energy radiated at high frequencies – the advantage of the energy magnitude being that it can be deduced from observations, even in real-time.
Applications of the developed method for the 2008 Wenchuan (China) earthquake, the 2003 Tokachi-Oki (Japan) earthquake and the 1994 Northridge (California, USA) earthquake show that the fine source discretisations combined with the small scale source variability ensure that high frequencies are satisfactorily introduced, justifying the deterministic wave propagation approach even at high frequencies. I demonstrate that the energy magnitude can be used to calibrate the high-frequency content in ground motion simulations.
Because deterministic wave propagation is applied to the whole frequency range, the simulation method permits the quantification of the variability in ground motion due to parametric uncertainties in the source description. A large number of scenario simulations for an M=6 earthquake show that the roughness of the source as well as the distribution of fault dislocations have a minor effect on the simulated variability by diminishing directivity effects, while hypocenter location and rupture velocity more strongly influence the variability. The uncertainty in energy magnitude, however, leads to the largest differences of ground motion amplitude between different events, resulting in a variability which is larger than the one observed.
For the presented approach, this dissertation shows (i) the verification of the computational correctness of the code, (ii) the ability to reproduce observed ground motions and (iii) the validation of the simulated ground motion variability. Those three steps are essential to evaluate the suitability of the method for means of seismic risk mitigation.
This paper aims at obtaining a semi‐analytical and semi‐numerical 3D model of source‐to‐site seismic wave propagation due to kinematic finite‐fault sources. To this end, a two‐step procedure integrating the frequency‐wavenumber (FK) approach with the spectral element method (SEM) is proposed based on the concept of domain reduction. First, the broadband responses of a stratified crust are accurately calculated by using a novel FK approach and are converted into effective seismic inputs around the region of interest. After that, the seismic wavefields at local and regional scales arising from complex geological and topographical conditions are finely simulated using the SEM, and a perfectly matched layer absorbing boundary condition is simultaneously applied to realize the absorption of outgoing waves. In this procedure, a hybrid source modelling scheme that combines the low‐wavenumber deterministic and high‐wavenumber stochastic components on the fault plane is introduced, effectively addressing the high‐frequency motion radiated from the source rupture process. Subsequently, the proposed FK‐SEM procedure is verified step‐by‐step using the point source and finite‐fault source models. To illustrate the feasibility of the procedure, 3D physics‐based numerical simulations (PBSs) of two seismic events, including a historical Sanhe‐Pinggu earthquake and a well‐recorded Yangbi earthquake, are performed. The case studies validate that the proposed FK‐SEM procedure allows a significant reduction in computational effort and a substantial improvement in modelling resolution and can be applied to the source‐to‐site broadband synthetics of earthquake scenarios with limited resources. In addition, this coupled geophysics‐engineering simulation meets the requirements of time‐history analysis for engineering structures, which facilitates the study of soil‐structure interactions and regional‐scale building damage assessment.
Stress transfer between earthquakes is recognized as a fundamental mechanism governing aftershock sequences. A common approach to relate stress changes to seismicity rate changes is the rate-and-state constitutive law developed by Dieterich: these elements are the foundation of Coulomb-rate-and-state (CRS) models. Despite the successes of Coulomb hypothesis and of the rate-and-state formulation, such models perform worse than statistical models in an operational forecasting context: one reason is thatCoulomb stress is subject to large uncertainties and intrinsic spatial heterogeneity. In this study, we characterize the uncertainties in Coulomb stress inherited from different physical quantities, and assess their effect on CRS models. We use aMonte Carlo method, and focus on the following aspects: the existence of multiple receiver faults; the stress heterogeneity within grid cells, due to their finite size; and errors inherited from the coseismic slip model. We study two well recorded sequences from different tectonic settings: the Mw=6.0 Parkfield and the Mw=9.0 Tohoku earthquakes. We find that the existence of multiple receiver faults is the most important source of intrinsic stress heterogeneity, and CRS models perform significantly better when this variability is taken into account. The choice of slip model also generates large uncertainties. We construct an ensemble model based on published slip models, and find that it outperforms individual models. Our findings highlight the importance of identifying sources of errors and quantifying confidence boundaries in the forecasts; moreover, we demonstrate that consideration of stress heterogeneity and epistemic uncertainty has the potential to improve the performance of operational forecasting models.
