[Thrust 1] [Thrust 2] [Thrust 4] [Thrust 5]

Faculty and Collaborators: Kei Aki (USC), Yehuda Ben-Zion (USC), James Brune (Nevada-Reno), Rob Clayton (Caltech), Paul Davis (UCLA), Gary Fuis (USGS), Mike Gurnis (Caltech), Egill Hauksson (Caltech), Don Helmberger (Caltech), Tom Henyey (USC), Sue Hough (USGS), Gene Humphreys (Oregon), Lucy Jones (USGS), Hiroo Kanamori (Caltech), John Louie (Nevada-Reno), Jim Mori (USGS), Craig Nicholson (UCSB), Mousumi Roy (New Mexico), Nano Seeber (Columbia), John Shaw (Harvard), Peter Shearer (UCSD), Joann Stock (Caltech), John Vidale (UCLA), David Wald (USGS), Lisa Wald (USGS)

Postdocs: John Armbruster (Columbia), Eli Baker (Maxwell Labs), Jishu Deng (Caltech), Nikki Godfrey (USC), Liz Hearn (Oregon), Marc Kamerling (UCSB), Monica Kohler (UCLA), Yong-Gang Li (USC), Harold Magistrale (SDSU), Andrew Meigs (Caltech), David Okaya (USC), Chris Sorlien (Columbia), Jamie Steidl (UCSB)

Students: Sara Carena (Princeton), Ji Chen (Caltech), Jennifer Eakins (UCSD), Jeanne Hardebeck (Caltech), DeAnn Icenogle (UCLA), Jennifer Lewis (SDSU), Peter Maechling (Caltech), John Marquis (Caltech), Julie Nazareth (Caltech), Tracy Pattelana (PCC), Zhi-Gang Peng (USC), Jascha Polet (Caltech), Keith Richards-Dinger (UCSD), Justin Rubenstein (UCLA), Michael Suess (Harvard), Fei Xu (UCLA)

Our primary focus this year has been on developing a 3D velocity model for Southern California, and on preparing for the upcoming LARSE II seismic survey. This year we will have developed the first version of a velocity model that incorporates the complexities of the sedimentary basins in the greater Los Angeles area. It is hoped that this velocity model will serve as a reference standard for tomographic, strong motion, waveform modeling and even geologic studies.
This year the analysis of the LARSE I data have been largely completed. Preparations are now under way for the LARSE II project which is scheduled to start in October of 1998, with the main part of the experiment occurring in the Fall of 1999. A considerable amount of effort has gone into the site locations and permits.

Other research this year includes precise earthquake locations, fault plane imaging and determining the properties of fault zones, the state of stress in the upper mantle, and the effect of subsurface structure on seismic waveforms.

Harold Magistrale, Rob Graves and Rob Clayton have developed a Standard Three-Dimensional Seismic Velocity Model For Southern California. Version 1 of the model fits a range of geological and geophysical observations. It consists of the major populated basins (Los Angeles basin, Ventura basin, San Gabriel Valley, San Fernando Valley, and San Bernardino Valley) embedded in a crust with velocity smoothly varying with depth, over a depth varying Moho. The model is parameterized as a set of objects and rules that are used to generate any 3D mesh of seismic velocity and density values (at length scales appropriate for different uses). The detailed basin portion of the model has over 70 defined surfaces that generally represent lithological interfaces. This parameterization is convenient to store and update as new information (e.g., oil well sonic logs) and verification results become available. Version 1 will be tested by a variety of waveform modeling, tomographic studies, and new data releases that will provide constraints to be incorporated into Version 2. The model is available on the SCEC Data Center at http://www.data.scec.org.

John Shaw and Peter Shearer have produced the first compelling image of an active blind-thrust fault beneath metropolitan Los Angeles, linking precisely located earthquakes, a direct fault image, and near-surface folds that characterize deformation above blind thrusts. The Puente Hills blind-thrust system, which is mapped using industry seismic reflection profiles, extends more than 40 kilometers along strike beneath downtown Los Angeles and northern Orange County. A small segment of this blind thrust apparently caused the 1987 Whittier Narrows (M 6.0) earthquake, the origin of which has remained controversial due to the lack of a surface fault trace and uncertainties in earthquake locations. To precisely relocate the earthquake mainshock and aftershocks, we use the L1-norm, waveform cross-correlation technique, with newly derived station terms and direct velocity control from petroleum wells. The relocated aftershock cluster is coincident with both the mainshock nodal plane and the projection to depth of the fault plane imaged in the seismic reflection profiles. The clear linkage between the earthquake and the imaged fault segment indicates that the Puente Hills blind-thrust system, which is not considered in current hazard models, is active and capable of damaging earthquakes. The Whittier Narrows event ruptured less than ten percent of this blind-thrust system, and thus remaining fault segments could generate much larger earthquakes (M 6.5-7.0+) beneath Los Angeles. In addition to these implications for regional hazards assessment, our results illustrate new methods to identify concealed faults in other metropolitan areas.

Luciana Astiz, Keith Richards-Dinger and Peter Shearer have successfully used waveform cross-correlation techniques to providing precise relative locations of similar events. These detailed relocations are useful for imaging fault structure and may help in understanding the rupture dynamics of earthquakes. We are using this approach in an ongoing project to improve earthquake locations for a number of aftershock sequences in southern California. We use seismograms recorded by the Southern California Seismic Network (SCSN), which are available through the Southern California Earthquake Center (SCEC) Data Center. We relocated 300,000 events occurring between 1981 and 1997 by applying an L1-norm, grid-search algorithm that uses station terms to account for three-dimensional velocity structure outside each aftershock region. This procedure greatly reduces the scatter in the relative locations (compared to the catalog hypocenters) producing sharper images of faults at depth.

Fundamental to the earthquake hazard problem is how the brittle crust is loaded by relative plate motion, whether faults are loaded by deep fault plastic creep, or viscous shear of a broad region of flow. Finite strain of the lithospheric mantle associated with viscous flow, is thought to align olivine crystals and cause seismic anisotropy. If the brittle crust is driven by plastic flow in the underlying lithosphere, a fabric may develop, that could be detected seismically, and used to quantify the flow. As part of the goals of LARSE II to probe the regional velocity variation, Paul Davis and colleagues seek to determine the orientation of seismic anisotropy in the mantle beneath Southern California and to compare it with adjacent regions and other tectonic regions of the world. SKS splitting can be used to determine the azimuthal variation of anisotropy. As we found earlier, the majority of stations show east-west splitting. This observation is in contrast to the vast majority of stations further east in the continental US which show NE-SW splitting perhaps related to the direction of absolute plate motion. If the contrast in southern California is due to flow in the mantle it may extend well into the continent, perhaps reaching as far as the Basin and Range province. We have searched for but have been unable to find evidence for a San-Andreas parallel second layer of splitting as has been suggested by others.

Monica Kohler, Dave Wald, and Rob Graves have studied the effect of localized sediments and subsurface structure on teleseismic waveforms in the Los Angeles Basin. Waveform data from a recent, high-density seismic array show a large variation in teleseismic P- and S-wave amplitudes across the Los Angeles basin. Since the relatively long-period (>1 sec) waveforms have nearly identical ray paths, the amplitude variations can be directly attributed to basin amplification. The recordings, made by the 18 stations of the 1997 Los Angeles Basin Passive Seismic Experiment (LABPSE), have an average spacing of 3 km, and span the San Gabriel Valley and deeper central Los Angeles sedimentary basins from Azusa to Seal Beach. Teleseisms from events with M>5.5 exhibit a P-wave amplitude ratio of up to 4:1 for stations near the deepest portion of the Los Angeles basin (where sediments are ~7 km thick) relative to the northernmost crystalline rock San Gabriel Mountain station. S-wave amplitude ratios are as much as 6:1. Some of the largest ratios, however, are seen at the southernmost edge of the basin (e.g., Seal Beach), even though sediments have thinned to about 3 km. The observed P waveforms also show a clear high-frequency shift at sites with large amplifications, indicating that while the overall sedimentary thickness is important, the majority of the amplitude variations are controlled by the nature of the shallowest structure. This will be more fully explored with synthetic waveform modeling. Initial two-dimensional, finite-difference modeling suggests that recent high-resolution upper crustal velocity models may not adequately predict the observed large amplification or frequency shift. The sensitivity of these waveforms to the shallow velocity profile will allow improvement to existing 2- and 3-D models of the Los Angeles Basin. This experiment illustrates that high-density arrays intended for local/regional wave propagation and deeper mantle studies, are also quite suitable for independent validation of the structural elements of previously obtained 2- and 3-D crustal velocity models intended for estimation of strong ground motions in hazards analysis.

Andy Michael and Yehuda Ben-Zion have used genetic inversion algorithms to study fault zone trapped waves. Fault zone (FZ) trapped waves can help constrain the seismic velocities and attenuation of the rocks in and around the FZ as well as the width of the FZ. We are developing a genetic algorithm inverse approach to extract this information from the waveforms along with realistic confidence bounds. Strong trade-offs between parameters such as FZ width, propagation distance, velocity contrasts, attenuation, and source and receiver location have been observed in sets of synthetic seismograms. This creates problems for both forward modeling and traditional optimization methods which genetic algorithms can overcome.

Forward modeling results suggest that these waves can accurately constrain various parameters because small changes in single parameters can make large changes in the synthetic waveforms. With the strong trade-offs between various parameters there may be multiple sets of parameters that yield similar synthetic seismograms, but without a systematic search this will not be completely explored.

The inverse method allows us to test a wider variety of models in an objective manner and develop confidence bounds on the best result. We quantify the fit function as the correlation coefficient between the observed and synthetic waveforms. Synthetics are computed with a generalized version of the solution of Ben-Zion and Aki [1990] for a scalar wave field in a structure consisting of two FZ layers between two quarter spaces. A grid search inversion shows that good fits can be obtained for a wide range of values in any one parameter by varying the other parameters. Thus, the strong trade-offs between parameters do undermine the forward modeling results as expected. The grid search results also reveal that the correlation coefficient is a complicated function of the fault zone parameters. This creates problems for traditional optimization techniques.

A. Venkataraman, Jim Mori and Hiroo Kanamori have determined fault planes of the April 26th and 27th, 1997 earthquakes. The low angle fault planes that they found have implications for the importance of low-angle detachment faults elsewhere. Using the directivity effects of source time functions we determined the rupture planes of the April 26th and 27th Mw5, 1997 earthquakes, close to Northridge, California. The two events occurred in the western part of the aftershock zone of the 1994 Northridge earthquake. Both aftershocks have thrust mechanisms, with one shallow dipping nodal plane (30 degrees NE) and one steeply dipping (60 degrees SW) nodal plane. Smaller aftershocks were used as Empirical Green's functions to deconvolve the path, site and instrument effects from the mainshock to obtain the source time functions for the events. A waveform inversion method was used to generate synthetics for the steep and shallow dipping planes for each event to determine the nodal plane which best fit the data. The results of our study show that the shallow dipping plane ruptured in the April 26th event, while the steeply dipping plane ruptured in the event on April 27th. We conclude that conjugate fault planes ruptured in the two events. Earlier studies of the Northridge aftershock sequence show that there have been Mw3.5 to Mw5 earthquakes with similar mechanisms in the region. The rupture planes for these events were not determined in these studies. The existence of low-angle detachment planes at depths of 12-19 km in southern California has long been debated. In the earlier studies of earthquakes in the region, there was no direct determination of the existence of low-angle planes. Here, we suggest that the April 26th event, which has a shallow dipping fault plane, ruptured on the detachment fault.


[Thrust 1] [Thrust 2] [Thrust 4] [Thrust 5]

Back to Executive Summary