S33C-01
Asymmetrical Behavior of Ground Motion in Extreme Shaking
The 2008 Iwate-Miyagi, Japan, earthquake (June 14, 2008, 8:43 JST; 39.0N, 130.9E, depth 10 km by the Japan Meteorological Agency; Mw 6.9) was a reverse-fault crustal earthquake with a source region extending 30 km in strike and 20 km in dip direction. Associate with this earthquake, the world-largest peak ground acceleration (PGA) on the free-surface, 4022 gal was recorded at the KiK-net IWTH25 station (West- Ichinoseki). The station was set in a stiffed soil (S-wave velocity = 450 m/s), and located on the hanging-wall site of the fault with 3 km southwest of the epicenter. The recorded PGA at the surface was extraordinarily high: 3866 gal in the vertical direction and 1434 gal in the horizontal direction. Contrary to the typical observations that horizontal ground motions are larger than vertical ground motions, vertical PGA at this site is more than twice than the horizontal one. The downhole (GL –260 m) site of this station set in a soft rock (S-wave velocity = 1810 m/s) also recorded a large amplitude of 640 and 1039 gals for the vertical and horizontal directions respectively. Thus the large PGA on the surface was caused by a large incident acceleration at the subsurface layer. The waveform and envelope of the up-down (U-D) component at the surface are strongly asymmetric with respect to the horizontal zero-axis, whereas the remaining surface and downhole (GL –260 m) components are broadly symmetric. It means that only the vertical accelerogram on the surface had distinct asymmetry; the positive (upward) envelope of the waveform was about 1.6 times larger than the negative (downward) envelope. The vertical ground motion on the surface was at a higher frequency than the horizontal motion, and the waveform asymmetry becomes visible above 8 Hz. Here we show this unusual asymmetry in ground surface acceleration, which is not explainable by any current model of linear or nonlinear site response, can be explained by a simple model of a mass bouncing during extreme ground shaking.
S33C-02
Simulation of a strong ground motion exceeding 4G during the 2008 Iwate-Miyagi Nairiku earthquake, Japan
The Iwate-Miyagi Nairiku earthquake, a powerful reverse fault event that occurred on the southern Iwate
prefecture Japan (2008/6/14), produced the largest peak ground acceleration recorded to date (4G), at the
West Ichinoseki, KiK-net/NIED strong motion station (IWTH25), which is located immediately above the
hypocenter. This KiK-net station which is equipped with surface and borehole accelerometers (GL -260m),
also recorded very high peak accelerations up to 1G at the borehole level, despite being located in a rock
site (Vs ~ 1800m/s). Preliminary analyses of these waveforms show a very strong content of high
frequencies (HF). To understand the generation process of this extreme shaking we simulated the three
components of ground motion at the IWTH25 borehole and surface levels, by applying a strong motion
simulation methodology based on a dynamic model of fault rupture (Pulido and Dalguer 2008, PD08).
Following PD08 we investigated the contribution of rupture velocity changes (Δ Vr) and stress drop
(Δσ) across the fault plane to the HF ground motion radiation of the earthquake at the borehole
level, and then calculated the non-linear response of the shallow layers to obtain the ground motion at the
surface.
To calculate the stress drop distribution of the earthquake we apply the methodology of Ripperger
and Mai (2004), by using the slip model obtained from an inversion of near-source strong motion recordings
(Suzuki et al. 2008), and a 1D velocity model for the Kanto region. The dislocation model of this earthquake
is characterized by two patches of large slip, the first one located at the hypocenter and the second at
~7 km south of the hypocenter. Our results for the calculation of stress drop follow a similar pattern,
namely a very large stress drop of ~80MPa concentrated at the hypocenter, as well as a large value of
~50MPa for the southern patch. To calculate the HF ground motion we follow PD08 who found that the
HF radiation from earthquakes is confined to regions in the fault plane where the product ΔσΔ
Vr is relatively large. Based on this result we calculate a stochastic distribution of Δ Vr values in
the ±3.0 km/s range for areas in the fault plane where Δσ > 15 MPa, and in the ±0.5
km/s range for the remaining areas. We calculate the ground motion at the IWTH25 station following the
methodology of PD08, by using the mapped HF radiation distribution and an average rupture velocity of 2
km/s. Our simulated waveforms incorporate the contribution of the P and S waves to the HF ground motion at
the horizontals and vertical component. Our simulation is able to reproduce the overall characteristics of the
observed waveform at IWTH25, including the observed ω -2 spectral radiation. Our results show a
strong contribution of the P wave radiation pattern to the relative amplitudes of ground motion for the three
components at IWTH25. In the present calculation our assumption of Δ Vr is entirely ad hoc. Future
research includes the calculation of a dynamic model of the earthquake to put physical constraints on the
fault rupture process and the simulation of near-source ground motion.
References
Ripperger, J., and P.M., Mai (2004), Geophys. Res. Lett., 31, L18610.
Pulido, N., and L.A. Dalguer (2008), Estimation of the high-frequency radiation of the 2000 Tottori (Japan)
earthquake based on a dynamic model of fault rupture: Application to the strong ground motion simulation,
Bull. Seism. Soc. Am., 2008 (in review).
Suzuki, W., S. Aoi, and H. Sekiguchi (2008), 2008 AGU fall meeting.
S33C-03
Synthetic Ground Motions for Engineering Applications and the Role of Nonlinear Site Response
Quantitative criteria are being developed for the efficient integration of site response models in broadband ground motion simulations. For this purpose, downhole array observations and broadband synthetics are combined, and the sensitivity of ground motion and nonlinear structural performance attributed to bias and uncertainty in nonlinear site response models is investigated. Results from medium-to-soft soil sites in Southern California are here presented, subjected to synthetic ground motions estimated for finite-source dynamic rupture scenarios of weak, medium and large magnitude events (M = 3.5~7.5), on a surface station grid of epicentral distances 2km~75km. For each site, elastic and nonlinear site response analyses are evaluated using multiple soil models, and the modeling ground motion variability is estimated by means of the COV (coefficient of variation) of site amplification. For each soil model, the parametric uncertainty of ground motion predictions is next estimated by systematically randomizing selected model parameters. Quantitative measures are developed that may describe the site properties and ground motion characteristics where the nonlinear models show large prediction COV, namely where incremental nonlinear analyses significantly deviate from empirical methodologies. Finally, the role of nonlinear soil response in physics-based seismic hazard predictions is illustrated by subjecting a series of inelastic SDOF (single-degree-of-freedom) oscillators to the ensemble of ground motion predictions, and evaluating the bias and uncertainty introduced as a result, in the structural response predictions. It is shown that the bias and uncertainty introduced in structural performance analyses when nonlinear site effects are not accounted for, strongly correlates with the deviation observed when the assessment of structural response is evaluated using synthetic seismograms from existing methodologies as opposed to real motions. It is concluded that soil nonlinearity, when efficiently integrated in ground motion models, can substantially improve existing physics-based earthquake predictions intended for use by the engineering community in the structural design practice.
S33C-04
Distribution of Ground Motions for the 2008 Mw5.4 Chino Hills Earthquake
The 29 July 2008 Mw5.4 Chino Hills earthquake was widely felt throughout the greater Los Angeles region. Over 40,000 people filled out the Community Internet Intensity Map (CIIM), also known as the "Did You Feel It?" questionnaire, and the ShakeMap for the event is constrained by amplitudes from over 500 stations. The CIIM responses provide Modified Mercalli Intensity (MMI) values for 802 Zip codes. Due to the dense population of the greater Los Angeles region, intensity values averaged within Zip codes provide good spatial correlation to the location of the actual reporting sites. We use a least-squares approach to fit the average MMI to a standard functional form, MMI = A - Br - C log r (where r represents distance from the observation site to the epicenter and A, B and C are constants), and calculate the residuals. The distribution of average intensities clusters closely around the curve of the equation, corroborating the conclusion (e.g., Atkinson and Wald, SRL, 2006) that CIIM intensities provide a consistent measure of earthquake ground motions. A comparison of CIIM intensities and instrumentally determined ShakeMap intensities reveals generally good correspondence, but suggests a tendency for ShakeMap intensities to be higher at basin sites. We conclude that the ShakeMap intensities reflect long-period amplification, whereas observed intensities were more controlled by higher frequency accelerations. Where both the CIIM map and ShakeMap show systematic variations, the distributions show a strong correlation with near-surface geological structure, as well as with basin depth. Our preliminary conclusions are: 1) as expected, intensities are lower at hard rock sites; 2) intensities are systematically higher in the deepest parts of the basin and significantly lower in shallow regions. Intensities are systematically lower to the west of the Newport-Inglewood fault correlating with subtle topographic relief and shallower basin depths; 3) there is a suggestion that intensities are systematically higher in the western half of the San Fernando Valley, consistent with a shoaling effect at basin edges; 4) intensities are systematically higher to the north, northwest, and west, and lower towards the east, at sites in the San Bernardino basin. The distribution suggests that source radiation and or focusing was favorable to the northwest or southwest, but not the northeast/southeast.
S33C-05
Simulation of Long-Period Ground Motion From a Hypothetical Nankai Subduction Earthquake in Western Japan
The long-term occurrence potentials of the megathrust earthquakes in Nankai trough of the subducting
Philippine Sea Plate are from 50% and 70% within 30 years from 2008 (The Headquarters for Earthquakes
Research Promotion, 2008). The hypothetical source region of Nankai earthquake extends as wide as 300
km in the east-west direction; therefore it is likely to cause large long-period ground motion to many
sedimentary basins in wide area of Japan. We study the propagation and generation of the long-period (3 -
20 s) ground motion in two sedimentary basins in western Japan: the Osaka basin and the Oita basin,
located northeast and west of the source region, respectively. Among the sedimentary basins in western
Japan, these are two deep basins in which long-period structures such as high rise buildings and oil storage
tanks are at high risk of seismic hazard by long-period ground motion.
We carried out a ground motion simulation of a hypothetical Nankai earthquake by a 3D finite-difference
method (Pitarka, 1999) using 3D basin and crustal velocity structure models (Iwata et al., 2008; Oita
Prefecture, 2008). The basin velocity structure models are validated by long-period ground motion
simulations of observed earthquake records. The source model is based on Sekiguchi et al. (2008) that
consists of five asperities and a background area, distributed on the top surface of Philippine Sea Plate of
the crustal velocity model. We tried two source models with different hypocenter, one in which rupture
propagates from east to west (east-hypo model), which is generally accepted as hypothetical source models
of the Nankai earthquake, and another from west to east (west-hypo model). The total seismic moment is
7.4× 1021 Nm (MW8.5) for both the source models.
In the Oita basin, the maximum peak ground velocity (PGV) reaches 100 cm/s in the east-hypo model
simulation while it is roughly one fifth smaller in the west-hypo model simulation. The east-hypo model causes
the directivity effect that enlarges the seismic waves generated by the asperities near to the Oita basin. On
the other hand, the maximum PGV exceeds 120 cm/s for both east- and west-hypo models in the Osaka
basin, and the difference between the two models is not as explicit as in the Oita basin because one asperity
that is located southward of Osaka is the most responsible for the large ground motion in the Osaka basin for
both source models. The simulated waveforms in the east-hypo model are comparable to the observed
seismograms of the 1946 Nankai earthquake (MJMA8.0) recorded at Japan Meteorological Agency
stations in western Japan, which indicates that the simulated ground motion in the basins in this study at least
not excessively overestimating.
Seismograms provided by Japan Meteorological Agency were used in this study.
S33C-06
Quantitative comparison of 3D numerical predictions of ground motion in the alpine valley of Grenoble, France.
The validation of numerical simulations of 3D seismic wave propagation is an important step for the prediction of strong ground motion. Here we report the results of benchmarking numerical simulations of ground motion in the alpine valley of Grenoble, a region of moderate seismicity but strong site effects. We compare different predictions of the seismic response of the Grenoble valley to two local earthquakes: a well-recorded event with magnitude M=3, and a hypothetical strong motion event with M=6. Four different codes are evaluated: one based on the Discontinuous Galerkin Method, another based on the Finite-Difference Method, while the other two are different implementations of the Spectral Element Method. The comparison starts with a visual inspection of ground acceleration at selected receivers and of global peak velocity maps. A quantitative analysis is then presented, based on two different measures introduced recently: the goodness-of-fit score [1], which consists of an average of ground motion indicators commonly used in engineering seismology; and the time-frequency (TF) misfit measure [2] which relies on a TF analysis of the difference between seismograms. By analysing the TF amplitude and phase misfits in different frequency bands we gain some insight about the origin of the discrepancies observed between the predictions of the different codes. Finally, we present the results obtained for the strong motion case when the effects of surface topography are taken into account. The different codes are found to provide consistent predictions of the effect of surface topography, regarding the location of the zones where amplification and deamplification occur, but differences in the level of (de-)amplification are observed, which will require further work to be fully understood. References: [1] Anderson J., Quantitative measure of the goodness-of-fit of synthetic seismograms, proceedings of the 13th World Conference on Earthquake Engineering, Vancouver, paper #243, 2004. [2] Kristekova M. et al., Misfit Criteria for Quantitative Comparison of Seismograms, Bull. Seism. Soc. Am., 96 (5), p. 1836-1850.
S33C-07
Searching for 3D wave-propagation effects in southern California
Ground motions in southern California can be significantly influenced by three-dimensional basin wave-propagation effects. Simulations of large events on the southern San Andreas fault within the SCEC Community Velocity Model (CVM) find large amplifications at Whittier Narrows due to basin-guided waves. We search for similar basin effects in southern California via a series of 3D wave-propagation simulations over a suite of potential source scenarios. Selection of scenarios is guided by the Uniform California Earthquake Rupture Forecast, as well as results from the CyberShake project. CyberShake provides an exhaustive set of synthetic seismograms at selected sites, calculated using a 3D model, that can be interrogated to identify source/site combinations with high basin-wave excitation. Simulations are performed up to 0.5 Hz using the Support Operator Rupture Dynamics (SORD) code and kinematic finite-fault sources, with fault geometry (often non-planar) derived the SCEC Community Fault Model. Selected scenarios are recomputed with dynamic rupture sources. Waves are propagated through the SCEC-CVM as well as the SCEC-CVMH (Harvard version) with comparisons helping to address uncertainty in basin-effect results.
S33C-08
Effects of realistic topography on seismic wave propagation: Large- and small-scale topography effects in northern Taiwan
Topography influences ground motion as is observed from data recorded during and after real earthquakes and from numerical simulations. However, the effects of realistic topography on ground motion have not been clearly characterized in numerical simulations. Furthermore, recent publications have mainly focused on implications for ground motion in the mountainous regions themselves, whereas the impact on surrounding low-lying areas has received less attention. Here we develop a new spectral-element mesh implementation to accommodate realistic topography in northern Taiwan. Spectral-element numerical simulations indicate that high-resolution topography can change peak ground velocity (PGV) values in mountainous areas by 50% compared to a half-space response. We further demonstrate that large-scale topography can affect the propagation of seismic waves in nearby areas. For example, if a shallow earthquake occurs in the I-Lan region of Taiwan, the Central Mountain Range will significantly scatter the surface waves and in turn reduce the amplitude of ground motion in the Taipei basin. We also use LiDAR DTM data, which provide two-meter resolution, to simulate 3-D seismic wave propagation in the Yangminshan region. We show that seismic shaking in mountainous areas is strongly affected by topography and source frequency content. Interactions between small-scale topographic features and high-frequency surface waves can produce unusually strong shaking. These studies suggest that high-resolution, realistic topographic features should be taken into account in seismic hazard analyses, especially in densely-populated metropolitan areas.