Crustal seismic-velocity models and datasets play a key role in regional 3D numerical earthquake ground-motion simulation, full waveform tomography, and modern physics-based probabilistic earthquake-hazard analysis, as well as in other related fields, including geophysics and earthquake engineering. Most of these models and datasets, often collectively identified as Community Velocity Models (CVMs), synthesize information from multiple sources and are delivered to users in variable formats, including computer applications that allow for interactive querying of material properties, namely P-and S-wave velocities and density.. Computational users often require massive and repetitive access to velocity models and datasets, and such access is often unpractical and difficult due to a lack of standardized methods and procedures. To overcome these issues and to facilitate access by the community to these models, the Southern California Earthquake Center developed the Unified CVM (UCVM) software framework, an open-source collection of tools that enables users to access one or more seismic-velocity models, while providing a standard query interface. Here, we describe the research challenges that motivated the development of UCVM, its software design, development approach, and basic capabilities, as well as a few examples of seismic-modeling applications that use UCVM.

We present a new, 3D description of crust and upper mantle velocity structure in southern California implemented as a Unified Structural Representation (USR). The USR is comprised of detailed basin velocity descriptions that are based on tens of thousands of direct velocity (Vp, Vs) measurements and incorporates the locations and displacement of major fault zones that influence basin structure. These basin descriptions were used to developed tomographic models of crust and upper mantle velocity and density structure, which were subsequently iterated and improved using 3D waveform adjoint tomography. A geotechnical layer (GTL) based on Vs30 measurements and consistent with the underlying velocity descriptions was also developed as an optional model component. The resulting model provides a detailed description of the structure of the southern California crust and upper mantle that reflects the complex tectonic history of the region. The crust thickens eastward as Moho depth varies from 10 to 40 km reflecting the transition from oceanic to continental crust. Deep sedimentary basins and underlying areas of thin crust reflect Neogene extensional tectonics overprinted by transpressional deformation and rapid sediment deposition since the late Pliocene. To illustrate the impact of this complex structure on strong ground motion forecasting, we simulate rupture of a proposed M 7.9 earthquake source in the Western Transverse Ranges. The results show distinct basin amplification and focusing of energy that reflects crustal structure described by the USR that is not captured by simpler velocity descriptions. We anticipate that the USR will be useful for a broad range of simulation and modeling efforts, including strong ground motion forecasting, dynamic rupture simulations, and fault system modeling. The USR is available through the Southern California Earthquake Center (SCEC) website (http://www.scec.org).

This project uses dynamic rupture simulations to investigate high-frequency seismic energy generation. The relevant phenomena (frictional breakdown, shear heating, effective normal-stress fluctuations, material damage, etc.) controlling rupture are strongly interacting and span many orders of magnitude in spatial scale, requiring highresolution simulations that couple disparate physical processes (e.g., elastodynamics, thermal weakening, pore-fluid transport, and heat conduction). Compounding the computational challenge, we know that natural faults are not planar, but instead have roughness that can be approximated by power laws potentially leading to large, multiscale fluctuations in normal stress. The capacity to perform 3D rupture simulations that couple these processes will provide guidance for constructing appropriate source models for high-frequency ground motion simulations. The improved rupture models from our multi-scale dynamic rupture simulations will be used to conduct physicsbased (3D waveform modeling-based) probabilistic seismic hazard analysis (PSHA) for California. These calculation will provide numerous important seismic hazard results, including a state-wide extended earthquake rupture forecast with rupture variations for all significant events, a synthetic seismogram catalog for thousands of scenario events and more than 5000 physics-based seismic hazard curves for California.

P>This paper presents a verification of three simulations of the ShakeOut scenario, an M-w 7.8 earthquake on a portion of the San Andreas fault in southern California, conducted by three different groups at the Southern California Earthquake Center using the SCEC Community Velocity Model for this region. We conducted two simulations using the finite difference method, and one by the finite element method, and performed qualitative and quantitative comparisons between the corresponding results. The results are in good agreement with each other; only small differences occur both in amplitude and phase between the various synthetics at ten observation points located near and away from the fault-as far as 150 km away from the fault. Using an available goodness-of-fit criterion all the comparisons scored above 8, with most above 9.2. This score would be regarded as excellent if the measurements were between recorded and synthetic seismograms. We also report results of comparisons based on time-frequency misfit criteria. Results from these two criteria can be used for calibrating the two methods for comparing seismograms. In those cases in which noticeable discrepancies occurred between the seismograms generated by the three groups, we found that they were the product of inherent characteristics of the various numerical methods used and their implementations. In particular, we found that the major source of discrepancy lies in the difference between mesh and grid representations of the same material model. Overall, however, even the largest differences in the synthetic seismograms are small. Thus, given the complexity of the simulations used in this verification, it appears that the three schemes are consistent, reliable and sufficiently accurate and robust for use in future large-scale simulations.

Dynamic rupture, and resultant ground motions up to 0.25 Hz, are simulated for an M-w 7.6 earthquake on the southern San Andreas fault. Spontaneous rupture is modeled with slip-weakening friction, and 3D viscoelastic wave solutions are computed with a support-operator numerical method. The initial traction model is derived from inversions of the M-w 7.3 1992 Landers strong ground-motion records, and borrows heavily from that used for the TeraShake2 simulations by Olsen et al. (2008). Heterogeneity in the traction model leads to focusing of the rupture front, and the focusing produces cases of supershear rupture velocity in asperities (areas of high initial traction), as well as cases of high peak slip rate and cohesive zone contraction in antiasperities. Separate solutions are computed for version 3.0 and 4.0, respectively, of the Southern California Earthquake Center Community Velocity Model (SCEC-CVM). We also compare the case of a flat ground surface (a common simplification made for finite-difference simulations) to the case of the ground surface conformed to regional topography. The overall distribution of simulated ground motion intensity is consistent with that derived from the empirical model of Campbell and Bozorgnia (2008), in the sense that the bulk of simulated pseudospectral velocity (PSV) values are within the 68% confidence intervals of the empirical model. Simulated PSVs corresponding to low probability in the empirical model are principally associated with basin waveguide and directivity effects. An important example, first identified by the TeraShake1 simulations (Olsen et al., 2006), is the stronger than expected ground motions at the site of Montebello due to a basin wave-guide effect. We find that this effect is lessened for version 4.0 of the SCEC-CVM, relative to version 3.0, due to a shallower model for the Chino basin.

Petascale simulations are needed to understand the rupture and wave dynamics of the largest earthquakes at shaking frequencies required to engineer safe structures (> 1 Hz). Toward this goal, we have developed a highly scalable, parallel application (AWP-ODC) that has achieved “M8”: a full dynamical simulation of a magnitude-8 earthquake on the southern San Andreas fault up to 2 Hz. M8 was calculated using a uniform mesh of 436 billion 40-m 3 cubes to represent the three-dimensional crustal structure of Southern California, in a 800 km by 400 km area, home to over 20 million people. This production run producing 360 sec of wave propagation sustained 220 Tflop/s for 24 hours on NCCS Jaguar using 223,074 cores. As the largest-ever earthquake simulation, M8 opens new territory for earthquake science and engineering - the physics-based modeling of the largest seismic hazards with the goal of reducing their potential for loss of life and property.

SUMMARY We present a numerical method to simulate spontaneous shear crack propagation within a heterogeneous, 3-D, viscoelastic medium. Wave motions are computed on a logically rectangular hexahedral mesh, using the generalized finite-difference method of Support Operators (SOM). This approach enables modelling of non-planar surfaces and non-planar fault ruptures. Our implementation, the Support Operator Rupture Dynamics (SORD) code, is highly scalable, enabling large-scale, multiprocessors calculations. The fault surface is modelled by coupled double nodes, where rupture occurs as dictated by the local stress conditions and a frictional failure law. The method successfully performs test problems developed for the Southern California Earthquake Center (SCEC)/U.S. Geological Survey (USGS) dynamic earthquake rupture code validation exercise, showing good agreement with semi-analytical boundary integral method results. We undertake further dynamic rupture tests to quantify numerical errors introduced by shear deformations to the hexahedral mesh. We generate a family of meshes distorted by simple shearing, in the along-strike direction, up to a maximum of 73 ◦ . For SCEC/USGS validation problem number 3, grid-induced errors increase with mesh shear angle, with the logarithm of error approximately proportional to angle over the range tested. At 73 ◦ , rms misfits are about 10 per cent for peak slip rate, and 0.5 per cent for both rupture time and total slip, indicating that the method (which, up to now, we have applied mainly to near-vertical strike-slip faulting) is also capable of handling geometries appropriate to low-angle surface-rupturing thrust earthquakes. Additionally, we demonstrate non-planar rupture effects, by modifying the test geometry to include, respectively, cylindrical curvature and sharp kinks.

We apply the method of support operators (SOM) to solve the 3-D, viscoelastic equations of motion for use in earthquake simulations. SOM is a generalized finite-difference method that can utilize meshes of arbitrary structure and incorporate irregular geometry. Our implementation uses a 3-D, logically rectangular, hexahedral mesh. Calculations are second-order in space and time. A correction term is employed for suppression of spurious zero-energy modes (hourglass oscillations). We develop a free surface boundary condition, and an absorbing boundary condition using the method of perfectly matched layers (PML). Numerical tests using a layered material model in a highly deformed mesh show good agreement with the frequency-wavenumber method, for resolutions greater than 10 nodes per wavelength. We also test a vertically incident P wave on a semi-circular canyon, for which results match boundary integral solutions at resolutions greater that 20 nodes per wavelength. We also demonstrate excellent parallel scalability of our code.