Broadband Ground Research Articles

Teleseismic waves reveal anisotropic poroelastic response of wastewater disposal reservoir

Connecting earthquake nucleation in basement rock to fluid injection in basal, sedimentary reservoirs, depends heavily on choices related to the poroelastic properties of the fluid-rock system, thermo-chemical effects notwithstanding. Direct constraints on these parameters outside of laboratory settings are rare, and it is commonly assumed that the rock layers are isotropic. With the Arbuckle wastewater disposal reservoir in Osage County, Oklahoma, high-frequency formation pressure changes and collocated broadband ground velocities measured during the passing of large teleseismic waves show a poroelastic response of the reservoir that is both azimuthally variable and anisotropic; this includes evidence of static shifts in pressure that presumably relate to changes in local permeability. The azimuthal dependence in both the static response and shear coupling appears related to tectonic stress and strain indicators such as the orientations of the maximum horizontal stress and faults and fractures. Using dynamic strains from a nearby borehole strainmeter, we show that the ratio of shear to volumetric strain coupling is ~0.41 which implies a mean Skempton's coefficient of <inline-formula><tex-math id="M2">\begin{document}$ A = 0.24 $\end{document}</tex-math><alternatives><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="SI6250-Andrew_M2.jpg"/><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="SI6250-Andrew_M2.png"/></alternatives></inline-formula> over the plausible range of the undrained Poisson's ratio. Since these observations are made at relatively low confining pressure and differential stress, we suggest that the hydraulically conductive fracture network is a primary control on the coupling between pore pressure diffusion and elastic stresses in response to natural or anthropogenic sources.

Simulation of Broadband Ground Motion by Superposing High-Frequency Empirical Green’s Function Synthetics on Low-Frequency Spectral-Element Synthetics

Seismic wave-propagation simulations are limited in their frequency content by two main factors: (1) the resolution of the seismic wave-speed structure of the region in which the seismic waves are propagated through; and (2) the extent of our understanding of the rupture process, mainly on the short length scales. For this reason, high-frequency content in the ground motion must be simulated through other means. Toward this end, we adopt a variant of the classical empirical Green’s function (EGF) approach of summing, with suitable time shift, recorded seismograms from small earthquakes in the past to generate high-frequency seismograms (0.5–5.0 Hz) for engineering applications. We superimpose these seismograms on low-frequency seismograms, computed from kinematic source models using the spectral element method, to produce broadband seismograms. The non-uniform time- shift scheme used in this work alleviates the over-estimation of high-frequency content of the ground motions observed. We validate the methodology by simulating broadband motions from the 1999 Hector Mine and the 2006 Parkfield earthquakes and comparing them against recorded seismograms.

A Machine Learning Approach to Developing Ground Motion Models From Simulated Ground Motions

AbstractWe use a machine learning approach to build a ground motion model (GMM) from a synthetic database of ground motions extracted from the Southern California CyberShake study. An artificial neural network is used to find the optimal weights that best fit the target data (without overfitting), with input parameters chosen to match that of state‐of‐the‐art GMMs. We validate our synthetic‐based GMM with empirically based GMMs derived from the globally based Next Generation Attenuation West2 data set, finding near‐zero median residuals and similar amplitude and trends (with period) of total variability. Additionally, we find that the artificial neural network GMM has similar bias and variability to empirical GMMs from records of the recent Ridgecrest event, which neither GMM has included in its formulation. As simulations continue to better model broadband ground motions, machine learning provides a way to utilize the vast amount of synthetically generated data and guide future parameterization of GMMs.

Hybrid broadband ground motion simulation validation of small magnitude earthquakes in Canterbury, New Zealand

Ground motion simulation validation is an important and necessary task toward establishing the efficacy of physics-based ground motion simulations for seismic hazard analysis and earthquake engineering applications. This article presents a comprehensive validation of the commonly used Graves and Pitarka hybrid broadband ground motion simulation methodology with a recently developed three-dimensional (3D) Canterbury Velocity Model. This is done through simulation of 148 small magnitude earthquake events in the Canterbury, New Zealand, region in order to supplement prior validation efforts directed at several larger magnitude events. Recent empirical ground motion models are also considered to benchmark the simulation predictive capability, which is examined by partitioning the prediction residuals into the various components of ground motion variability. Biases identified in source, path, and site components suggest that improvements to the predictive capabilities of the simulation methodology can be made by using a longer high-frequency path duration model, reducing empirical V s30-based low-frequency site amplification, and utilizing site-specific velocity models in the high-frequency simulations.

Large Stress Release During Normal-Faulting Earthquakes in Western Turkey Supported by Broadband Ground Motion Simulations

This article investigates the stress drop variability of shallow normal-faulting earthquakes in western Anatolia through strong-motion simulations. For this purpose, source characteristics of three moderate to large magnitude events are constrained by the empirical Green’s function simulation in a broadband frequency range. Recordings of ten strong-motion stations in 78-km epicentral distance range are utilized for the simulation. Estimated strong-motion generation areas (SMGAs) are 22 km2, 66 km2, and 110 km2 where rise times are 0.6 s, 0.7 s, and 0.6 s, respectively, for the 2011 Simav (Mw 5.8), 2017 Lesvos (Mw 6.3), and 2017 Bodrum-Kos (Mw 6.6) earthquakes. Those values are found to be consistent with global source-scaling relationships. One particular observation is that stress drop ratios between the mainshock and aftershock for all events are relatively large compared with those previously calculated for strike-slip events in Turkey. Stress drops of SMGAs for the Simav and Bodrum earthquakes are in the range of 25 MPa, and this value drops to 19 MPa for the Lesvos earthquake. To further investigate the stress drop variation of earthquakes in Western Anatolia, an earthquake source database (fc-Mo) offered by Yamanaka et al. (IAG-IASPEI 2017, S07-1-03, 2017) is utilized. Brune (Journal of Geophysical Research, 76:5002, 1971) stress drop values of > 360 small to moderate earthquakes (Mw 3.0–6.0) are calculated with the given corner frequency and seismic moment information assuming a constant shear wave velocity. Results indicate that the majority of the earthquakes have a stress drop value < 5 MPa. This value changes to between 5 and 57 MPa for the remaining earthquakes. These high stress drop values support the former findings stating that normal-faulting earthquakes may release higher stress than strike-slip earthquakes. This indicates that the regional stress regime in western Turkey may cause relatively larger stress release during the normal-faulting mainshocks.

Kinematic Source Modeling for the Synthesis of Broadband Ground Motion Using the f‐k Approach

Abstract 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.

Broadband Ground Motion Simulation Within the City of Düzce (Turkey) and Building Response Simulation

In areas with significant seismic activity and insufficient seismic networks, simulated ground motions are required for both seismic hazard and engineering analyses. One important task is evaluating the efficiency of the simulated records in building response estimation. In this study, a hybrid ground motion simulation framework is presented to obtain broadband ground motion time histories of a past major earthquake in Duzce (Turkey). The hybrid ground motion simulation framework presented herein is a combination of a discrete-wavenumber finite element method for simulating low frequencies and a stochastic finite-fault method for the higher frequencies. The proposed technique is first validated by the simulation of the 12 November 1999 Duzce earthquake (\(M_{\text{w}}\) = 7.1), which occurred in the North Anatolian Fault Zone, Turkey. Peak simulated ground motions at selected nodes are then compared against ground motion prediction equations derived with global and local data sets. Next, the spatial distribution of simulated peak ground motion intensities are obtained in the region and compared with the observed damage distribution. Results demonstrate that the earthquake is simulated effectively despite the limitations in number of stations and velocity models. Finally, to evaluate the simulated records in building response estimation, nonlinear time history analyses of three reinforced-concrete multi-degree-of-freedom structures are performed with real and simulated records of the 1999 Duzce earthquake in the OpenSees platform. The results reveal that reasonable predictions can be made regarding the dynamic response of structures using the records simulated with the approach presented herein.

Methodology and computational implementation of a New Zealand Velocity Model (NZVM2.0) for broadband ground motion simulation

ABSTRACTThis paper presents a generalised velocity model construction methodology, its computational implementation, and application in the construction of a New Zealand Velocity Model (NZVM2.0) for use in physics-based broadband ground motion simulation. The methodology utilises multiple datasets spanning different length scales, which is enabled via the use of modular subregions, geologic surfaces, and parametric representations of crustal velocity. A number of efficiency-related workflows to decrease the overall computational construction time are employed, while maintaining the flexibility and extensibility to incorporate additional datasets and refined velocity parameterisations as they become available. The methodology and computational implementation processes are then applied for development of a New Zealand Velocity Model (NZVM2.0) for use in broadband ground motion simulation. The model comprises explicit representations of the Canterbury, Wellington, Nelson-Tasman, Kaikōura, Marlborough, Waiau, Hanmer and Cheviot sedimentary basins embedded within an existing regional travel-time tomography-based velocity model for the shallow crust.

Kinematic Rupture Modeling of Ground Motion from the M7 Kumamoto, Japan Earthquake

We analyzed a kinematic earthquake rupture generator that combines the randomized spatial field approach of Graves and Pitarka (Bull Seismol Soc Am 106:2136–2153, 2016) (GP2016) with the multiple asperity characterization approach of Irikura and Miyake (Pure Appl Geophys 168:85–104, 2011) (IM2011, also known as Irikura recipe). The rupture generator uses a multi-scale hybrid approach that incorporates distinct features of both original approaches, such as small-scale stochastic rupture variability and depth-dependent scaling of rupture speed and slip rate, inherited from GP2016, and specification of discrete high slip rupture patches, inherited from IM2011. The performance of the proposed method is examined in simulations of broadband ground motion from the 2016 Kumamoto, Japan earthquake, as well as comparisons with ground motion prediction equations (GMPEs). We generated rupture models with multi-scale heterogeneity, including a hybrid one in which the slip is a combination of high- slip patches and stochastic small scale variations. We find that the ground motions simulated with these rupture models match the general characteristics of the recorded near-fault motion equally well, over a broad frequency range (0–10 Hz). Additionally, the simulated ground motion is in good agreement with the predictions from Ground Motion Prediction Equations (GMPEs). Nonetheless, due to sensitivity of the ground motion to the local fault rupture characteristics, the performance among the models at near-fault sites is slightly different, with the hybrid model producing a somewhat better fit to the recorded ground velocity waveforms. Sensitivity tests of simulated near-fault ground motion to variations in the prescribed kinematic rupture parameters show that average rupture speeds higher than the default value in GP2016 (average rupture speed = 80% of local shear wave speed), as well as slip rate durations shorter than the default value in GP2016 (rise time coefficient = 1.6), generate ground motions that are higher than the recorded ones at periods longer than 1 s. We found that these two parameters also affect the along strike and updip rupture directivity effects, as illustrated in comparisons with the Kumamoto observations.

Broadband (0–5 Hz) Fully Deterministic 3D Ground‐Motion Simulations of a Magnitude 7.0 Hayward Fault Earthquake: Comparison with Empirical Ground‐Motion Models and 3D Path and Site Effects from Source Normalized Intensities

Author(s): Rodgers, AJ; Pitarka, A; Anders Petersson, N; Sjogreen, B; McCallen, DB; Abrahamson, N | Abstract: We report on high-performance computing (HPC) fully deterministic simulation of ground motions for a moment magnitude (Mw) 7.0 scenario earthquake on the Hayward fault resolved to 5 Hz using the SW4 finite-difference code. We computed motions obeying physics-based 3D wave propagation at a regional scale with an Mw 7.0 kinematic rupture model generated following Graves and Pitarka (2016). Both plane-layered (1D) and 3D Earth models were considered, with 3D subsurface material properties and topography interpolated from a model of the U.S. Geological Survey (USGS). The resulting ground-motion intensities cover a broader frequency range than typically considered in regional-scale simulations, including higher frequencies relevant for engineering analysis of structures. Median intensities for sites across the domain are within the reported between-event uncertainties (τ) of ground-motion models (GMMs) across spectral periods 0.2-10 s (frequencies 0.1-5 Hz). The within-event standard deviation ϕ of ground-motion intensity measurement residuals range 0.2-0.5 natural log units with values consistently larger for the 3D model. Source-normalized ratios of intensities (3D/ 1D) reveal patterns of path and site effects that are correlated with known geologic structure. These results demonstrate that earthquake simulations with fully deterministic wave propagation in 3D Earth models on HPC platforms produce broadband ground motions with median and within-event aleatory variability consistent with empirical models. Systematic intensity variations for the 3D model caused by path and site effects suggest that these epistemic effects can be estimated and removed to reduce variation in site-specific hazard estimates. This study motivates future work to evaluate the validity of the USGS 3D model and investigate the development of path and site corrections by running more scenarios.

Simulation of strong ground motion from the 1995 Mw 6.5 Kozani-Grevena, Greece, earthquake using a hybrid deterministic-stochastic approach

The 1995 Mw 6.5 Kozani-Grevena earthquake struck an area in northwestern Greece characterized by relatively low seismicity based on historical information and instrumental data. The ground shaking in the near-fault region was recorded by a strong-motion accelerograph in the city of Kozani, at a distance of about 23 km from the epicenter. In this article, broadband ground motions are generated at selected locations and at a dense grid of observation points extending over the causative fault of the 1995 Kozani-Grevena earthquake using a hybrid deterministic-stochastic method. Based on a multi-segment fault model with curved geometry proposed in the literature, the low-frequency components of the synthetic ground motion are simulated using the discrete wavenumber representation method and the generalized transmission and reflection coefficient technique. The high-frequency components of the synthetic ground motion are generated using the stochastic modeling approach and the specific barrier model. The two independently derived ground-motion components are then combined using matched filtering at a crossover frequency of 1.2 Hz to generate broadband ground-motion time histories and response spectra. The simulation results are validated by comparing the synthetic ground motions with the recorded ones in the city of Kozani and with estimates of peak ground-motion quantities inferred from ground-motion prediction equations.

Validation of Deterministic Broadband Ground Motion and Variability from Dynamic Rupture Simulations of Buried Thrust Earthquakes

Validation of Deterministic Broadband Ground Motion and Variability from Dynamic Rupture Simulations of Buried Thrust Earthquakes

Sensitivity Analysis of Reinforced Concrete Frame Structures Under Near-Fault Pulse-Like Ground Motions Using a Broadband Simulation Method

ABSTRACT The effect of pulse characteristic of near-fault ground motions on multistory reinforced concrete frame structures is investigated by means of a synthetic broadband ground motion model (SBGM). A simplified analytical pulse model is used to describe the low-frequency components of ground motions, while a stochastic simulation method based on wavelet packet is adopted to simulate the high-frequency components of ground motions. First, the time histories, response spectrum, and structural response of synthetic ground motions are compared with those corresponding to recorded near-fault ground motions. The good agreement indicates that the SBGM model used in this article is a suitable and efficient tool for evaluating the structural response. Sensitivity analysis on structural response of 2-, 4-, 8-, and 12-story building models to the SBGM model input parameters is then performed to identify the key ground-motion parameters which control the main structural response indicators, that is, interstory drift ratio and maximum interstory drift ratio. Final results indicate that the amplitude of velocity pulse in the SBGM model has a significant influence on structural drift demands. Nevertheless, for low-rise buildings, the maximum drift demands would effectively decrease with increasing proportion of high-frequency components. Moreover, the ratio of pulse period to structural fundamental period and number of oscillations impose a large drift demands to the structure under near-fault pulse-like ground motions.

Broadband Ground Motions from 3D Physics‐Based Numerical Simulations Using Artificial Neural Networks

Broadband Ground Motions from 3D Physics‐Based Numerical Simulations Using Artificial Neural Networks

Probabilistic Estimates of Ground Motion in the Los Angeles Basin from Scenario Earthquakes on the San Andreas Fault

Kinematic source inversions of past earthquakes in the magnitude range of 6–8 are used to simulate 60 scenario earthquakes on the San Andreas fault. The unilateral rupture scenario earthquakes are hypothetically located at 6 locations spread out uniformly along the southern section of the fault, each associated with two hypocenters and rupture directions. Probabilities of occurrence over the next 30 years are assigned to each of these earthquakes by mapping the probabilities of 10,445 plausible earthquakes postulated for this section of the fault by the Uniform California Earthquake Rupture Forecast. Three-component broadband ground motion histories are computed at 636 sites in the greater Los Angeles metropolitan area by superposing short-period (0.2–2.0 s) empirical Green’s function synthetics on top of long-period (&gt;2.0 s) spectral element synthetics. The earthquake probabilities and the computed ground motions are combined to develop probabilistic estimates of ground shaking in the region from San Andreas fault earthquakes over the next 30 years. The results could be useful in city planning, emergency management, and building code enhancement.

Broadband (0–4 Hz) Ground Motions for a Magnitude 7.0 Hayward Fault Earthquake With Three‐Dimensional Structure and Topography

AbstractWe performed fully deterministic broadband (0–4 Hz) high‐performance computing ground motion simulations of a magnitude 7.0 scenario earthquake on the Hayward Fault (HF) in the San Francisco Bay Area of Northern California. Simulations consider average one‐dimensional (1‐D) and three‐dimensional (3‐D) anelastic structure with flat and topographic free surfaces. Ground motion intensity measures (GMIMs) for the 3‐D model display dramatic differences across the HF due to geologic heterogeneity, with low wave speeds east of the HF amplifying motions. The median GMIMs agree well with Ground Motion Prediction Equations (GMPEs); however, the 3‐D model generates more scatter than the 1‐D model. Ratios of 3‐D/1‐D GMIMs from the same source allow isolation of path and site effects for the 3‐D model. These ratios show remarkably similar trends as site‐specific factors for the GMPE predictions, suggesting that wave propagation effects in our 3‐D simulations are on average consistent with empirical data.

The 2015 Gorkha (Nepal) earthquake sequence: I. Source modeling and deterministic 3D ground shaking

To better quantify the relatively long period (<0.3Hz) shaking experienced during the 2015 Gorkha (Nepal) earthquake sequence, we study the finite rupture processes and the associated 3D ground motion of the Mw7.8 mainshock and the Mw7.2 aftershock. The 3D synthetics are then used in the broadband ground shaking in Kathmandu with a hybrid approach, summarized in a companion paper (Chen and Wei, 2017, submitted together). We determined the coseismic rupture process of the mainshock by joint inversion of InSAR/SAR, GPS (static and high-rate), strong motion and teleseismic waveforms. Our inversion for the mainshock indicates unilateral rupture towards the ESE, with an average rupture speed of 3.0km/s and a total duration of ~60s. Additionally, we find that the beginning part of the rupture (5–18s) has about 40% longer rise time than the rest of the rupture, as well as slower rupture velocity. Our model shows two strong asperities occurring ~24s and ~36s after the origin and located ~30km to the northwest and northeast of the Kathmandu valley, respectively. In contrast, the Mw7.2 aftershock is more compact both in time and space, as revealed by joint inversion of teleseismic body waves and InSAR data. The different rupture features between the mainshock and the aftershock could be related to difference in fault zone structure. The mainshock and aftershock ground motions in the Kathmandu valley, recorded by both strong motion and high-rate GPS stations, exhibited strong amplification around 0.2Hz. A simplified 3D basin model, calibrated by an Mw5.2 aftershock, can match the observed waveforms reasonably well at 0.3Hz and lower frequency. The 3D simulations indicate that the basin structure trapped the wavefield and produced an extensive ground vibration. Our study suggests that the combination of rupture characteristics and propagational complexity are required to understand the ground shaking produced by hazardous earthquakes such as the Gorkha event.

Broadband ground-motion simulation of the 24 May 2014 Gokceada (North Aegean Sea) earthquake (Mw 6.9) in NW Turkey considering local soil effects

On 24 May 2014, a Mw 6.9 earthquake occurred in the west of Gokceada Island, northern Aegean Sea. The earthquake was close to Canakkale, Enez, Tekirdag cities, and damaged 300 buildings in the Marmara Region, NW Turkey. We simulated its broadband (0.1–10 Hz) ground motions including 1D deep and shallow structures soil amplification effects at the 12 strong ground motion stations in the western Marmara Region. The 1D deep velocity structures from the focal layer to the engineering bedrock with an S-wave velocity of 0.78 km/s in different azimuthal directions were tuned by comparing the observed group-velocity dispersion curves of Rayleigh and Love waves from the mainshock with theoretical ones. We also added the shallow parts from previous surveys into the 1D models. Synthetic seismograms on the engineering bedrock were generated using the discrete wave number method with a source model and the 1D deep velocity structures. Then the surface motion was generated considering shallow soil amplification. The synthetic seismograms are generally in good agreement with the observed low and high-frequency parts at most of the stations indicating an appropriateness of the source model and the 1D structural model.

Broadband Ground Motion Simulation of the 2004 and 1977 Vrancea, Romania, Earthquakes Using Empirical Green’s Function Method

We estimated the parameters of strong motion generation areas and simulated broadband ground motions for the moderate October 27, 2004 (M w 5.8) and damaging March 4, 1977 (M w 7.4) Vrancea (Romania) intermediate-depth subduction earthquakes using the empirical Green’s function method. The method allows the simulation of ground motions in a broadband frequency range by summing up the subevent records, corresponding to small magnitude events in the near-source areas, which are assumed to follow the source-scaling relationship and the omega-square source spectral model. We first estimated the strong motion generation area that reproduces near-source ground motions in a broadband frequency range of 0.3–10 Hz for the 2004 earthquake, by fitting the synthetic acceleration, velocity, and displacement waveforms to the observed data. The source properties of the obtained strong motion generation area are in agreement with the predictions made using an empirical source scaling relationship for crustal earthquakes, implying a stress drop of approximately 10 MPa for the 2004 earthquake. We then modeled the strong motion generation area for the 1977 damaging earthquake using the 2004 earthquake as an empirical Green’s function and constructing a source model based on its estimated source parameters. To simulate the unique record at Bucharest, capital city of Romania, the rupture was assumed to have propagated from the northeast bottom of the strong motion generation area having a stress drop of 50 MPa. Broadband ground motion simulations were further compared in terms of the modified Mercalli intensity values, calculated from the peak ground accelerations and peak ground velocities of synthetic waveforms, with the observed Medvedev–Sponheuer–Karnik intensity values. Our estimates of the source properties for the 2004 and 1977 Vrancea intermediate-depth earthquakes support the size-dependent stress drop.

Physics-Based Hazard Assessment for Critical Structures Near Large Earthquake Sources

We argue that for critical structures near large earthquake sources: (1) the ergodic assumption, recent history, and simplified descriptions of the hazard are not appropriate to rely on for earthquake ground motion prediction and can lead to a mis-estimation of the hazard and risk to structures; (2) a physics-based approach can address these issues; (3) a physics-based source model must be provided to generate realistic phasing effects from finite rupture and model near-source ground motion correctly; (4) wave propagations and site response should be site specific; (5) a much wider search of possible sources of ground motion can be achieved computationally with a physics-based approach; (6) unless one utilizes a physics-based approach, the hazard and risk to structures has unknown uncertainties; (7) uncertainties can be reduced with a physics-based approach, but not with an ergodic approach; (8) computational power and computer codes have advanced to the point that risk to structures can be calculated directly from source and site-specific ground motions. Spanning the variability of potential ground motion in a predictive situation is especially difficult for near-source areas, but that is the distance at which the hazard is the greatest. The basis of a “physical-based” approach is ground-motion syntheses derived from physics and an understanding of the earthquake process. This is an overview paper and results from previous studies are used to make the case for these conclusions. Our premise is that 50 years of strong motion records is insufficient to capture all possible ranges of site and propagation path conditions, rupture processes, and spatial geometric relationships between source and site. Predicting future earthquake scenarios is necessary; models that have little or no physical basis but have been tested and adjusted to fit available observations can only “predict” what happened in the past, which should be considered description as opposed to prediction. We have developed a methodology for synthesizing physics-based broadband ground motion that incorporates the effects of realistic earthquake rupture along specific faults and the actual geology between the source and site.