Southern California Earthquake Center (SCEC) 2014 Annual Meeting SDSU Abstract – Department of Earth & Environmental Sciences

The Bimaterial Effect on the Earthquake Cycle
Brittany A. Erickson and Steven M. Day
Both laboratory studies and dynamic rupture simulations on a bimaterial fault interface show that the material contrast can cause the rupture to take a “preferred” direction, where the rupture propagates in the direction of particle motion of the side with slower shear wave velocity. In addition, dynamic ruptures on bimaterial faults have lead to such phenomena as asymmetric rupture features, a preferred aftershock triggering, and asymmetric off-fault damage. However, many of these conclusions drawn from dynamic rupture simulations depend on the artificial initial conditions necessary to nucleate rupture. We have developed a computational method for studyinghow the bimaterial effect plays out over multiple earthquake cycles that develop spontaneously. The method is developed for the classical plane-strain equations capable of incorporating non-planar fault geometries and material heterogeneities. As a first step we consider the bimaterial problem on planar, vertical strike slip fault governed by rate-and-state friction. The system is loaded at the remote boundaries at the slow, tectonic plate rate. During the long interseismic period we solve the equations for static-equilibrium. Quasi-dynamic events nucleate spontaneously on the fault. We find that the preferred nucleation site migrates in the preferred direction. Ruptures propagate with asymmetric properties, the strength of which increases with greater material mismatch. We are currently studying the long term behavior of the earthquake cycle and under what conditions rupture in the non-preferred direction (periodically observed on natural faults) is possible.

Previous scenario simulations shared the assumption of a linear viscoelastic stress-strain relationship in crustal rocks, and methods that derive amplifications from ambient noise are also inherently assuming linear material behavior. Motivated by studies of dynamic rupture in nonelastic media, we simulate the wave propagation for selected San Andreas earthquake scenarios in a medium governed by Drucker-Prager elastoplasticity. For the ShakeOut earthquake scenario (based on the kinematic source description of Graves et al., 2008) plastic yielding in crustal rock reduces peak ground velocities in the Los Angeles basin between 30 and 70% as compared to viscoelastic solutions. Our simulations also show that the large strains induced by long-period surface waves in the San Gabriel and Los Angeles basins would trigger nonlinear behavior in cohesionless sediments, as suggested by theoretical studies (e.g., Sleep & Erickson, 2014). In the fault zone plasticity remains important even for conservative values of cohesions, suggesting that current simulations assuming a linear response of crustal rocks are overpredicting ground motions during future large earthquakes on the southern San Andreas fault. We analyze the sensitivity of ground motion levels to the choice of nonlinear material properties (cohesion, friction angle) and the initial stress field, and discuss how nonlinear material behavior may be included in future simulations of major earthquake scenarios and physics-based seismic hazard assessment.

Far-field seismic spectral response resulting from complex rupture behaviors
Yongfei Wang, Steven M. Day, and Peter M. Shearer
Many earthquake physical properties, such as seismic moment, rupture extent and stress drop, can be estimated from far-field seismic wave spectra. Corner frequency and the high-frequency fall-off rate of the spectra are often measured in order to make an assessment of dynamic stress drop and other source parameters such as radiated energy. Based on specific theoretic models, some quantitative relations have been established between far-field spectra and source properties. The most widely accepted model is described in Madariaga (1976), who performed a finite-difference simulation of a circular crack model. An important relation is f_c=kβ⁄a, where f_c is azimuthally averaged corner frequency, β is S-wave speed, a is circular radius and k is an empirical constant with different values for P and S wave spectra. Many other models have been proposed, including the recent dynamically realistic rupture simulations of Kaneko and Shearer (2014), all of which have yielded a variety of different values for k. However, models to date have been for relatively simple ruptures and the effect of rupture complexity, including heterogeneous stress and slip, on stress drop and scaled energy estimates has not been fully explored. Here, we consider complicated ruptures, which include fault roughness and complex pre-stress distributions, and compute the spectra that would be recorded by realistic distributions of surface stations. We then process the synthetic data using methods commonly applied to real data and attempt to quantify which fault properties can be reliably estimated from the observations and the most likely sources of errors in the analysis.

We model the region surrounding the fault with and without small-scale medium complexity, with varying statistical parameters. Furthermore, we analyze the effect of varying both the power-law exponent of the attenuation relation (n) and the reference Q (Q0) (assumed to be proportional to the S-wave velocity), and compare our synthetic ground motions with several Ground Motion Prediction Equations (GMPEs) as well as observed accelerograms. We find that the spectral acceleration at various periods from our models are within 1 interevent standard deviation from the median GMPEs and compare well with that of recordings from strong ground motion stations at both short and long periods. At periods below 1 second, Q(f) is needed to match the decay of spectral acceleration seen in the GMPEs as a function of distance from the fault (n~0.6-0.8).We find when binning stations with a common distance metric (such as Joyner-Boore or rrup) that the effect of media heterogeneity is canceled out, however, the similarity between the intra-event variability of our simulations and observations increases when small-scale heterogeneity is included.

Dynamic Compaction as a Simple Mechanism for Fault Zone Weakening
Evan T. Hirakawa and Shuo Ma
Elevated pore fluid pressures have long been thought to contribute to the apparent weakness of large plate bounding faults such as the San Andreas Fault. Sleep and Blanpied (1992) propose a mechanism in which compaction during interseismic creep reduces available pore space and hence increases fluid pressure. Assuming that the fault zone is not fully compacted during the interseismic period, we invoke a similar concept in this study, however in this case the compaction process occurs dynamically via the stresses associated with earthquake rupture. Microstructural analysis of fault zone rocks reveals intragranular fragmentation that is developed dynamically (e.g. Rempe et al., 2009) and is consistent with the style of fracture observed in individual quartz grains during bulk compaction (Chester et al., 2004), which thus lends some justification to our proposed mechanism.
We incorporate undrained compaction into a dynamic rupture model of a strike-slip fault by using an end-cap failure criterion (e.g. Wong et al., 1997). We show that dynamic stresses associated with rupture propagation cause the fault zone to compact, leading to elevated pore pressure in the undrained fault zone and low effective shear stress on the fault, consistent with the heat-flow constraints. An important result is that radiated S-waves propagating ahead of the rupture front can cause compactant failure and consequently lower the static friction as well as the dynamic friction, leading to a low strength drop on the fault. This may present a possible advantage over other dynamic weakening mechanisms such as thermal pressurization, which involves large strength drops accompanying rupture propagation.

Optimizing CyberShake Seismic Hazard Workflows for Large HPC Resources
Scott Callaghan, Philip Maechling, Gideon Juve, Karan Vahi, Robert W. Graves, Kim B. Olsen, Kevin Milner, David Gill, and Thomas H. Jordan
The CyberShake computational platform is a well-integrated collection of scientific software and middleware that calculates 3D simulation-based probabilistic seismic hazard curves and hazard maps for the Los Angeles region. Currently each CyberShake model comprises about 235 million synthetic seismograms from about 415,000 rupture variations computed at 286 sites. CyberShake integrates large-scale parallel and high-throughput serial seismological research codes into a processing framework in which early stages produce files used as inputs by later stages. Scientific workflow tools are used to manage the jobs, data, and metadata. The Southern California Earthquake Center (SCEC) developed the CyberShake platform using USC High Performance Computing and Communications systems and open-science NSF resources. CyberShake calculations were migrated to the NSF Track 1 system NCSA Blue Waters when it became operational in 2013, via an interdisciplinary team approach including domain scientists, computer scientists, and middleware developers. Due to the excellent performance of Blue Waters and CyberShake software optimizations, we reduced the makespan (a measure of wallclock time-to-solution) of a CyberShake study from 1467 to 342 hours. This improvement enabled the calculation of hazard maps for 4 models, including the new community velocity model CVM-S4.26. We will describe the technical enhancements behind this improvement, including judicious introduction of new GPU software, improved scientific software components, increased workflow-based automation, and Blue Waters-specific workflow optimizations. Our CyberShake performance improvements highlight the benefits of scientific workflow tools. The CyberShake workflow software stack includes the Pegasus Workflow Management System (Pegasus-WMS, which includes Condor DAGMan), HTCondor, and Globus GRAM, with Pegasus-mpi-cluster managing the high-throughput tasks on the HPC resources. The workflow tools handle data management, automatically transferring about 13 TB back to SCEC storage. We will present performance metrics from the most recent CyberShake study, executed on Blue Waters. We will compare the performance of CPU and GPU versions of our large-scale parallel wave propagation code, AWP-ODC-SGT. Finally, we will discuss how these enhancements have enabled SCEC to move forward with plans to increase the CyberShake simulation frequency to 1.0 Hz.

UCVM: Open Source Software for Understanding and Delivering 3D Velocity Models
David Gill, Patrick Small, Philip Maechling, Thomas H. Jordan, John Shaw, Andreas Plesch, Po Chen, En-Jui Lee, Ricardo Taborda, Kim Olsen, and Scott Callaghan
Physics-based ground motion simulations can calculate the propagation of earthquake waves through 3D velocity models of the Earth. The Southern California Earthquake Center (SCEC) has developed the Unified Community Velocity Model (UCVM) framework to help researchers build structured or unstructured velocity meshes from 3D velocity models for use in wave propagation simulations. The UCVM software framework makes it easy to extract P and S wave propagation speeds and other material properties from 3D velocity models by providing a common interface through which researchers can query earth models for a given location and depth. Currently, the platform supports multiple California models, including SCEC CVM-S4 and CVM-H 11.9.1, and has been designed to support models from any region on earth. UCVM is currently being use to generate velocity meshes for many SCEC wave propagation codes, including AWP-ODC-SGT and Hercules. In this presentation, we describe improvements to the UCVM software. The current version, UCVM 14.3.0, released in March of 2014, supports the newest Southern California velocity model, CVM-S4.26, which was derived from 26 full-3D tomographic iterations using CVM-S4 as the starting model (Lee et al., this meeting), and the Broadband 1D velocity model used in the CyberShake 14.2 study. We have ported UCVM to multiple Linux distributions and OS X. Also included in this release is the ability to add small-scale stochastic heterogeneities to extract Cartesian meshes for use in high-frequency ground motion simulations. This tool was built using the C language open-source FFT library, FFTW. The stochastic parameters (Hurst exponent, correlation length, and the horizontal/vertical aspect ratio) can be customized by the user. UCVM v14.3.0 also provides visualization scripts for constructing cross-sections, horizontal slices, basin depths, and Vs30 maps. The interface allows researchers to visually review velocity models . Also, UCVM v14.3.0 can extract isosurfaces of shear-wave velocities equal to 1 km/s (Z1.0) and 2.5 km/s (Z2.5) for any of the registered velocity models. We have also improved our open source distribution by including a user’s guide, an advanced user’s guide, and a developer’s guide so that users of all levels can get started using and extending the UCVM platform.

SCEC Community Modeling Environment SI2 and Geoinformatics Research Projects
Philip J. Maechling, Thomas H. Jordan, Jacobo Bielak, Yifeng Cui, Kim B. Olsen, and Jeroen Tromp
The SCEC Community Modeling Environment (SCEC/CME) collaboration performs a broad range of computational research with support from NSF awards SI2-SSI: A Sustainable Community Software Framework for Petascale Earthquake Modeling (SEISM) (OCI-1148493), and Geoinformatics: Community Computational Platforms for Developing Three Dimensional Models of Earth Structure (EAR-1226343). On the SEISM project, SCEC researchers are integrating high-level and middle-level scientific software elements (SSEs) into a Software Environment for Integrated Seismic Modeling (SEISM), a sustainable software ecosystem for physics-based seismic hazard analysis. The SCEC SEISM software supports the use of petascale computers by earthquake scientists to generate and manage the large suites of earthquake simulations needed for physics-based PSHA, and advances basic research on rupture dynamics, an elastic wave scattering, and Earth structure. The SEISM software framework includes high-level SSEs for developing and managing unified community velocity models, codes for dynamic and pseudo-dynamic rupture generation, deterministic and stochastic earthquake simulation engines, and the applications necessary to employ forward simulations in two types of inverse problems: seismic source imaging and full-3D tomography. On the Geoinformatics project, SCEC researchers have established an interoperable set of community computational platforms that allow investigators to employ the techniques of full-3D tomography to refine Earth structures. Two tomographic platforms have been built on highly scalable codes for solving the forward problem: the AWP-ODC 4th-order, staggered-grid, finite-difference code, which has been widely used for regional earthquake simulation and physics-based seismic hazard analysis and the SPECFEM3D spectral element code, which is capable of modeling wave propagation through aspherical structures of essentially arbitrary complexity on scales ranging from local to global. A third platform, based on the Unified Community Velocity Model (UCVM) software developed by the Southern California Earthquake Center (SCEC), provides a common framework for comparing and synthesizing Earth models and delivering model products to a wide community of geoscientists.

The SDSU Broadband Ground Motion Generation Module Version 1.5
Kim B. Olsen and Rumi Takedatsu
SCEC has completed Phase 1 of its Broadband Platform (BBP) ground motion simulation results, evaluating the potential applications for engineering of the resulting PSAs generated by 5 different methods. The evaluation included part A, where the methods were evaluated based on the bias of simulation results to observations for 7 well-recorded historical earthquakes with source-station distances between 1 and 193 km, and part B, where simulation results for Mw 6.2 and Mw 6.6 strike-slip and reverse-slip scenarios were evaluated at 20 km and 50 km from the fault. The methods were assessed based on the bias of the median PSA for the 7 events in part A, and on a specified acceptance criterion compared to NGA-West2 GMPEs in part B. One of the 5 methods evaluated was BBtoolbox, a hybrid method combining deterministic low-frequency (LF) synthetics with high-frequency (HF) scatterograms (Mai et al., 2010; Mena et al., 2010; V1.4). In the validation exercise, the LFs are generated using 1-D Green’s Functions and 50 source realizations from the kinematic source generator module by Graves and Pitarka (2010, ‘GP’). However, the results from BBtoolbox V1.4 did not pass the SCEC validation phase 1. In order to obtain more accurate BB synthetics, we generated BBtoolbox V1.5 which scales the HFs to a theoretical spectral level at the merging frequency (GP, Eq. 10) fixed at 1Hz, rather than the level of the LFs in V1.4. This modification has generated much improved spectral levels at higher frequencies, while we have seen little evidence of artifacts from this technique. In addition, V1.5 introduced a new source time function with rise-time scaled as a function of moment. With these modifications, BBtoolbox V.15 became one of three methods passing the SCEC validation exercise Phase I. Here, we describe the details of BBtoolbox V1.5, and show comparisons between BBtoolbox V1.5 synthetics and observations from the valiation exercise.

Can We Rely on Linear Elasticity to Predict Long-period Ground Motions?
Daniel Roten, Kim B. Olsen, Steven M. Day, and Yifeng Cui
A major challenge in seismic hazard assessment consists in the prediction of near-source ground motions resulting from large, rare earthquakes, which are not well represented in observed data. In southern California a basis for assessing the economical and societal impact of a major (M>7.8) earthquake rupturing the San Andreas fault has been provided by numerical simulations of various rupture scenarios. Simulations of SE-NW propagating sources (e.g., ShakeOut) have predicted strong long-period (> 2s) ground motions in the densely populated Los Angeles basin due to channeling of seismic waves through a series of contiguous sedimentary basins. Recently such a waveguide amplification effect has also been identified from virtual earthquakes based on Green’s function derived from the ambient seismic field (Denolle et al., 2014). Previous scenario simulations shared the assumption of a linear viscoelastic stress-strain relationship in crustal rocks, and methods that derive amplifications from ambient noise are also inherently assuming linear material behavior. Motivated by studies of dynamic rupture in nonelastic media, we simulate the wave propagation for selected San Andreas earthquake scenarios in a medium governed by Drucker-Prager elastoplasticity. For the ShakeOut earthquake scenario (based on the kinematic source description of Graves et al., 2008) plastic yielding in crustal rock reduces peak ground velocities in the Los Angeles basin between 30 and 70% as compared to viscoelastic solutions. Our simulations also show that the large strains induced by long-period surface waves in the San Gabriel and Los Angeles basins would trigger nonlinear behavior in cohesionless sediments, as suggested by theoretical studies (e.g., Sleep & Erickson, 2014). In the fault zone plasticity remains important even for conservative values of cohesions, suggesting that current simulations assuming a linear response of crustal rocks are overpredicting ground motions during future large earthquakes on the southern San Andreas fault. We analyze the sensitivity of ground motion levels to the choice of nonlinear material properties (cohesion, friction angle) and the initial stress field, and discuss how nonlinear material behavior may be included in future simulations of major earthquake scenarios and physics-based seismic hazard assessment.

Validation Exercise for Two Southern California Earthquakes
William H. Savran and Kim B. Olsen
We are in the process of a comprehensive validation exercise constraining parameters describing Q(f) and statistical models of small-scale heterogeneities for finite-difference earthquake simulations in Los Angeles basin. The parameters for our model of shallow crustal heterogeneities are constrained by inverting 38 deep borehole logs in Los Angeles basin (Hurst number 0.0-0.1, correlation length 50-150 m, ~5% σ), which are superimposed onto the SCEC CVM-SI 4.26. We simulate viscoelastic waves for the 2008 Mw 5.4 Chino Hills Event (0-2.5 Hz, modified finite fault source from Shao et al., 2012) and the 2014 Mw 5.1 La Habra event (0-1 Hz, point source), and compare our simulations to strong-motion records. The optimal linear Qs-Vs relation derived from our results is Qs=0.1Vsf^0.6 (Vs in m/s). We find excellent goodness-of-fit (GOF) scores for f<1.0 Hz, in particular at deep-basin sites, that degrade when frequencies increase to 2.5Hz, suggesting an inadequate description of the finite-fault source for f > 1Hz. Poor fits (under prediction of metrics) are often found at hard-rock sites even for f < 1 Hz, likely due to too large values of the shallow Vs in the CVM. The statistical distributions of small-scale heterogeneities generate localized 2x amplifications and de-amplifications, and tend to improve GOF scores by 5-10%. Our simulations suggest that the majority of the scattering recorded in ground motions originates as a path effect as waves propagate through the basins, while local site-specific scattering in the immediate vicinity of a station at the earth’s surface tends to play a smaller role. We find almost equal contributions from scattering in the sedimentary basins and deeper parts of the shallow crust. Our modeling demonstrates unique amplification patterns caused by scattering due to the heterogeneous structure of the shallow crust. In particular, we find that shallow sources located on the boundary to a sedimentary basin generate bands of strong amplification aligned in the direction of the ray paths. The nature of these bands depends strongly on the incidence angle of the waves into the sediments. Moreover, this banded amplification pattern is absent for sources deeper than 1-2 km. Our results imply that surface rupture on a range-bound fault (e.g, the San Andreas fault by the San Bernardino Basin) may generate a different patterns of ground motion shaking along lines parallel to the fault as compared to profiles perpendicular to the fault.

High-Complexity Deterministic Q(f) Simulation of the 1994 Northridge Mw 6.7 Earthquake
Kyle B. Withers, Kim B. Olsen, Zheqiang Shi, and Steve Day
With the recent addition of realistic fault topography in 3D simulations of earthquake source models, ground motion can be deterministically generated more accurately up to higher frequencies. The synthetic ground motions have been shown to match the characteristics of real data, having a flat power spectrum up to some cutoff frequency (Shi and Day, 2013). However, the earthquake source is not the only source of complexity in the high-frequency ground motion; there are also scattering effects caused by small-scale velocity and density heterogeneities in the medium that can affect the ground motion intensity. Here, we dynamically model the source of the 1994 Mw 6.7 Northridge earthquake using the Support Operator Rupture Dynamics (SORD) code up to 8 Hz. Our fault model was input into a layered velocity structure characteristic of the southern California region characterized by self-similar roughness from scales of 80 m up to the total length of the fault. We extend the ground motion to further distances by converting the output from SORD to a kinematic source for the finite difference anelastic wave propagation code AWP-ODC. This code incorporates frequency-dependent attenuation via a power law above a reference frequency in the form Q0fn. We model the region surrounding the fault with and without small-scale medium complexity, with varying statistical parameters. Furthermore, we analyze the effect of varying both the power-law exponent of the attenuation relation (n) and the reference Q (Q0) (assumed to be proportional to the S-wave velocity), and compare our synthetic ground motions with several Ground Motion Prediction Equations (GMPEs) as well as observed accelerograms. We find that the spectral acceleration at various periods from our models are within 1 interevent standard deviation from the median GMPEs and compare well with that of recordings from strong ground motion stations at both short and long periods. At periods below 1 second, Q(f) is needed to match the decay of spectral acceleration seen in the GMPEs as a function of distance from the fault (n~0.6-0.8).We find when binning stations with a common distance metric (such as Joyner-Boore or rrup) that the effect of media heterogeneity is canceled out, however, the similarity between the intra-event variability of our simulations and observations increases when small-scale heterogeneity is included.

Holocene geologic slip rate for the Banning strand of the southern San Andreas Fault near San Gorgonio Pass
Peter O. Gold, Whitney M. Behr, Dylan Rood, Katherine Kendrick, Thomas K. Rockwell, and Warren D. Sharp
We present the first Holocene geologic slip rate for the Banning strand of the southern San Andreas Fault in southern California. Our new slip rate measurement, critically located at the northwestern end of the Banning strand, overlaps within errors with the published rate for the southern San Andreas Fault measured to the south at Biskra Palms Oasis. This indicates that the majority of southern San Andreas Fault displacement transfers from the southeastern Mission Creek strand northwest to the Banning strand and into San Gorgonio Pass. Our result corroborates the UCERF3 hazard model, and is consistent with most previous interpretations of how slip is partitioned between the Banning and Mission Creek fault strands. To measure this slip rate, we used B4 airborne LiDAR to identify the apex of an alluvial fan offset laterally 30 ± 5 m from its source. We calculated the depositional age of the fan using in-situ Be-10 cosmogenic exposure dating of 5 cobbles and a depth profile. We calculated a most probable fan age of 4.0 +2.0/-1.6 ka (1σ) by combining the inheritance-corrected cobble ages assuming Gaussian uncertainty. However, the probability density function yielded a multi-peaked distribution, which we attribute to variable 10Be inheritance in the cobbles, so we favor the depth profile age of 2.2-3.6 ka. Combined, these measurements yield a late Holocene slip rate for the Banning strand of the southern San Andreas Fault of 11.1 +3.1/-3.3 mm/yr. This slip rate does not preclude possibility that some slip transfers north along the Mission Creek strand and the Garnet Hill fault, but it does confirm that the Banning strand has been the most probable rupture path for earthquakes nucleated on the southern San Andreas Fault over the past few thousand years, and is likely to remain so in the near future. This clarification of slip partitioning within the northwest Coachella Valley is timely given that the southern San Andreas Fault is considered overdue for a large earthquake.

Long-term uplift of the southern California coast between San Diego and Newport Beach resolved with new dGPS survey data: Testing blind thrust models in the offshore Borderlands
Erik C. Haaker, Thomas K. Rockwell, George L. Kennedy, Lisa B. Grant Ludwig, Justin A. Zumbro, and S. Thomas Freeman
Marine terrace shorelines provide information on vertical tectonic motions, thereby yielding constraints on rates and styles of deformation for underlying structures, such as blind thrust faults. In coastal southern California, the Oceanside Blind Thrust (OBT) has been inferred from offshore seismic reflection data, and its intersection with the coast has been inferred to be the source of uplift of the San Joaquin Hills (SJH). The OBT has been interpreted to be the result of a tectonically inverted Miocene detachment fault, and has been hypothesized as a late Quaternary seismic source underlying coastal San Diego and southern Orange counties. Late Quaternary motion on the OBT should deform and uplift Quaternary marine terraces. To test OBT seismic source models, we collected over 3000 high-resolution GPS elevation data points for flights of Pleistocene marine terraces spanning the southern California coastal zone from central San Diego County through the city of Newport Beach in Orange County. We mapped the terraces by geomorphically tracing out and correlating individual shoreline exposures. In addition, we compiled subsurface geotechnical borehole data that supplemented our survey data where urban development had obscured or obliterated the original geomorphic relationships. From these new data, the shorelines for terraces below 140 m elevation are observed to remain at nearly constant elevation from San Diego northward through Camp Pendleton. The lowest two terraces that date to MIS 5.1 and 5.5 show minor variation in San Clemente, and then gently decrease a few meters in elevation towards the north in the vicinity of Newport Bay, consistent with the observation that terraces below about 60 m elevation show no significant variation along the coast from San Diego into Newport Beach. These observations do not appear to support late Quaternary activity of the OBT, and raise questions about how to reconcile recent seismicity in the SJH with published models of an underlying blind thrust

Dynamic Rupture Models of the Historic and Recent Paleoseismic Rupture Sequence of the Northern and Central San Jacinto Fault
Julian C. Lozos, Thomas K. Rockwell, and Nathan W. Onderdonk
We use the 3D finite element method to conduct dynamic rupture models on the Claremont, Casa Loma, and Clark strands of the San Jacinto Fault, in order replicate rupture extents and slip distributions for historic and recent paleoseismic earthquakes. The recent historic behavior of the northern and central San Jacinto Fault has been limited to moderate events, but paleoseismic data indicates that all three strands have experienced multiple large ruptures, with an average of 2.5 to 3 m slip on the Claremont (Onderdonk et al., 2014) and 3 to 3.5 m slip on the Clark (Rockwell et al., 2014). We vary initial stresses on the fault to see which conditions are necessary to replicate these larger events, as well as which changes to those conditions are required to also accommodate smaller historic earthquakes. A wide range of initial stress conditions produces many ruptures that span the entire Claremont strand, with an average surface slip of ~2.5 m, regardless of where we nucleate along strike. These Claremont ruptures continue through the extensional stepover at Mystic Lake, but produce lower slip on the adjacent Casa Loma strand; this may imply incomplete stress drop on the Casa Loma even in a large rupture, leaving some stored stress for a smaller event, such as the ~M6.4 1899 earthquake that caused damage in this area. Our model does produce 1899-like events when we use a lower stress drop. These same initial conditions also produce ruptures similar to the ~M6.9 1918 earthquake centered on the Clark strand near Anza. We find that the major barrier to rupture on the northern San Jacinto Fault is the compressional stepover between the Casa Loma and Clark strands at Park Hill. The persistence of this barrier suggests that early-1800s paleoseismic records of high-slip events on both the Claremont and Clark may represent separate earthquakes, rather than a single rupture. If the 22 November 1800 earthquake noted at Missions San Diego and San Juan Capistrano occurred on the Clark (Salisbury et al., 2012), then the Claremont may have participated in the 8 December 1812 earthquake, which is traditionally interpreted as a San Andreas-only rupture. We also find that a lower fault strength and a higher stress drop are required to produce a rupture of the full Clark strand than of the full Claremont. This suggests that the 1800 event, if it did rupture the Clark, must have been a very energetic rupture, which is consistent with observations of significant damage at large distances from the fault.

Slip rate, slip history, and slip per event along the northern San Jacinto fault zone during the past 2000 years: a summary of work at the Mystic Lake and Quincy sites on the Claremont fault.
Nathan Onderdonk, Sally McGill, and Tom Rockwell
The main objective of the SCEC Southern San Andreas Fault Evaluation project was to “increase our understanding of the slip rate, paleo-earthquake chronology, and slip distribution in recent earthquakes on the southern San Andreas fault system during the past 2000 years”. Work along the Claremont strand of the San Jacinto fault zone during the past 6 years has done this and we summarize the results here. Four field seasons of paleoseismic trenching at the Mystic Lake site have produced a record of 15 ground-rupturing earthquakes in the past 3500 years. During the last 2000 years there were 12 earthquakes, with the most recent occurring around AD 1800. The average recurrence interval is 175 ± 20 years, and the 200+ years since the most recent event is longer than any gap in the past 2000 years. Trenching and dating of offset streams and one buried channel at a second site (the Quincy site) has resulted in four independent calculations of slip rate and average slip per event. The slip rate for the last 2000 years is 12 to 19 mm/yr and the average slip per event is 2.3 m. Short-term slip rates calculated from an offset stream and the buried channel, which were offset in the past three events, are higher by about 10 mm/yr due to a cluster of earthquakes on the fault between AD 1400 and AD 1800 and higher than average displacements of 2.7 to 3 m in one or more of these earthquakes. This spurt of activity may explain the longer than normal time between earthquakes that we are now experiencing. Comparison of the Mystic Lake earthquake history with the Hog Lake paleoseismic record on the Clark fault to the south, and the Wrightwood paleoseismic record on the Mojave segment of the San Andreas fault to the north, shows overlaps in the timing of some events. Five possible correlations between the Mystic Lake record and the Hog Lake record, and another five possible correlations between Mystic Lake and Wrightwood suggest that some events may have ruptured through the step-overs at the ends of the Claremont fault or that events on one fault trigger closely-timed events on the adjacent fault. None of the events in the past 2000 years overlap at all three sites, indicating that an earthquake rupturing the San Andreas fault all the way to the central San Jacinto fault (or vice-versa) is unlikely.

Evidence of Subsidence Events along the Rincon Creek Fault in Carpinteria Marsh
Laura C. Reynolds, Alexander R. Simms, Thomas K. Rockwell, and Robert Peters
Recent marine terrace work indicates that the Ventura Avenue Anticline (VAA) and associated Pitas Point thrust fault have produced at least four Holocene uplift events between Ventura and Carpinteria, with a maximum uplift of 7-8m each, implying a significant earthquake and tsunami risk for the southern California coast. Previous work has also suggested that Carpinteria Marsh, located on the northern, downgoing side of the Rincon Creek Fault (RCF), has subsided nearly 1 km during the Quaternary. What is less clear is whether the nature of this subsidence is slow and steady or episodic, and to what degree the RCF is linked to motion on the VAA and Pitas Point thrust. A comparison between 10 new radiocarbon dates obtained from Carpinteria Marsh and regional records of Holocene relative sea level indicates that Carpinteria Marsh has experienced ~1m/ky of subsidence over the past 7ky. A preliminary stratigraphy based on 39 vibracores that are up to 4m in length, and 7 Geoprobe cores that are up to 13m in length, shows three candidate surfaces for subsidence events at 4m, 5.5m and 8m bmsl, respectively. The candidate event at 4m is characterized by a marsh surface that is sharply overlain by marine-influenced estuarine sand that correlates across the front of the marsh. The two other candidate surfaces are characterized by sharp-based marsh surface deposits overlying fluvial deposits, also possibly indicating a change in the elevation of the marsh surface at times when mean sea level was lower. Natural marsh processes, such as changes in sediment supply or mouth closure are other explanations for these candidate surfaces, although some degree of subsidence is likely required to produce the observed relationships. Future work will focus on using microfossil analysis to address whether the candidate surfaces represent abrupt elevation changes, as well as improving the chronology to test whether marsh surfaces are synchronous with the uplifted Holocene marine terraces.

Geologic and structural controls on rupture zone fabric: A field-based study of the 2010 Mw 7.2 El Mayor–Cucapah earthquake surface rupture
Orlando J. Teran, John M. Fletcher, Michael E. Oskin, Thomas K. Rockwell, Kenneth W. Hudnut, Ronald M. Spelz, Sinan O. Akciz, Ana P. Hernandez, and Alexander E. Morelan
We systematically mapped (scales >1:500) the surface rupture of the 4 April 2010 Mw 7.2 El Mayor-Cucapah earthquake through Sierra Cucapah to understand how faults with similar structural and lithologic characteristics control rupture zone fabric, which is here defined by the thickness, distribution and internal configuration of shearing in a rupture zone. Fault zone architecture and master fault dip showed the strongest controls on rupture zone fabric. Highly localized slip was observed along simple narrow fault cores (<20 m), whereas wide cores (>>50 m) composed of multiple zones of high shear strain had wider and more complex rupture zones that generally lacked principal-displacement scarps. Rupture zone thickness also increases systematically with decreasing fault zone dip. We observed that coseismic slip along faults that dip >40° was mostly confined to the fault core, whereas faults that dip as low as 20° had surface rupture entirely developed entirely outside of the fault zone. The lack of large off-fault strain along faults dipping >40° is contrary to predictions by dynamic stress modeling (e.g., Ma, 2009). We show that static tectonic loading, which varies significantly with fault orientation, has a significant effect not only on rupture zone fabric but also on the evolution of fault zone architecture in this transtensional setting. Rupture zones in undeformed alluvium are dominated by secondary fractures associated with fault-tip propagation, and arrays of fault scarps become wider and more complex with oblique slip compared to pure normal dip-slip or pure strike-slip. Field relations show that as magnitude of coseismic slip increases from 0 to 60 cm, the linkages between kinematically distinct fracture sets increases systematically to the point of forming a through going principal scarp, which is contrary to many analogue models (e.g., Tchalencko, 1970; Naylor et al., 1995). Our data indicate that secondary faults and penetrative off-fault strain continue to accommodate the oblique kinematics of coseismic slip after the formation of a through going principal scarp.