Thrust 1: Earthquake Potential and Seismic Hazard Estimation

Thrust 2: Earthquake Geology

Thrust 3: Subsurface Imaging, Seismicity and Tectonics

Thrust 4: Crustal Deformation

Thrust 5: Earthquake Physics

References

The eighth year of the Southern California Earthquake Center has seen major progress toward completion of principal integrative scientific products in three of the center's major thrust areas, and important headway in these and other thrust areas. Highlights include:

• Final assembly of the comprehensive SCEC "Phase III" report "Accounting for Site Effects in Probabilistic Seismic Hazard Analysis" and preparation for external scientific review (Integrative product of the Earthquake Potential and Seismic Hazard Estimation Thrust Area).
• Completion of the second (revised) version of the SCEC crustal deformation velocity model for southern California (Integrative product of the Crustal Deformation Thrust Area).
• Completion of the first version of the SCEC crustal seismic velocity model for southern California (Integrative product of the Subsurface Imaging, Seismicity, and Tectonics Thrust Area).
• Modeling ground motions (periods >3 seconds) in the Los Angeles basin from nine scenario earthquakes using the new SCEC 3-D seismic velocity model.
• A re-assessment of the rather provocative inference in our 1995 "Phase II" report of the existence of an "earthquake deficit" in southern California (i.e., that model predictions of seismicity were a factor of two higher [in the M6-7 range] than the historical record), with the conclusion that within the limits of uncertainty, model predictions are not inconsistent with past seismicity.
• Further evaluation of active faults in the greater Los Angeles basin including: a) the discovery that M7+ earthquakes have occurred on at least one local thrust fault in the Los Angeles metropolitan region, b) a determination that one major thrust fault beneath the region is apparently inactive, and c) a suggestion that north-south convergence across the region may be partially accommodated by east-west crustal extrusion.
• Mounting evidence that earthquake locations (including aftershocks) are strongly influenced by the evolution and distribution of stress from previous large events.

It has been long known that neighboring sites can experience significantly different levels of shaking during earthquakes. This behavior is referred to as a site effect. Site effects influence earthquake shaking in the form of sediment amplification, resonance, focusing and defocusing, basin-edge induced surface waves, and nonlinear response. An important issue addressed by SCEC in its Phase III report is whether accounting for site effects can significantly improve probabilistic seismic hazard analyses. In a series of papers that will be published together in the near future, SCEC has carried out both empirical and theoretical investigations of how attenuation relations might be evaluated and modified to account for site effects. Our findings show that certain site attributes are correlated with ground motion amplitudes. Phase III will quantify how correcting for these systematic differences influence seismic hazard assessment. While many of these effects can be treated well for past earthquakes, their predictability on a site by site basis in future earthquakes is uncertain.

An equally important finding is that accounting for observed site effects does not appreciably reduce the prediction uncertainty of attenuation relations. This can be understood by the fact that basin-edge induced surface waves, focusing and defocusing, and scattering, in general, are highly sensitive to source location. These effects produce a high degree of variability between earthquakes at any given site, making the systematic site effects seem relatively small. However, even small corrections may have a significant impact on hazard assessment.
Lastly, Phase III explores various ways of handing uncertainties in hazard calculations. For example, it is customary to use the spatial variability of ground motion from a limited number of earthquakes as a proxy for the temporal variability from many earthquakes (ergodic assumption), which may lead to overstated hazard levels under certain conditions.

The SCEC Crustal Deformation Working Group recently released Version 2 of its Crustal Deformation Velocity Map for southern California (Figure 1). Nearly all the EDM, VLBI, and SCEC archived GPS data acquired for southern California between 1970 and 1997 were included. The map and accompanying table contain data from 363 stations for which the uncertainty in horizontal velocity is less than 5 mm/yr. For stations close to the Landers epicenter, we have provided estimates of both pre- and post-Landers velocities. The most important differences between Version 2.0 and Version 1.0 (October 1996) are: a) the direct use of the VLBI data and the addition of GPS data from continuous observations since 1992, b) post-Landers surveys of the epicentral region, c) a 1992 survey of about 60 stations in and around the Los Angeles basin, and d) a 1997 re-survey of about 60 stations along the southern San Andreas fault system. These additions have added 76 stations to the map and reduced 2/3 of the station velocity uncertainties to about 1 mm/yr.
SCEC has completed Version 1.0 of a standard 3-D velocity model for the coastal basins of southern California, including the Ventura, San Fernando, Los Angeles, San Gabriel, Chino, and San Bernardino basins. The 3-D basin velocity structures are embedded in a smoothly varying crust over a flat Moho. The impetus for developing the model has been the calculation of accurate strong motion seismograms. Version 1 of the model fits a range of geological and geophysical observations and can be used for waveform modeling at periods of 3 seconds and greater. Future versions of the 3-D velocity model will: a) increase the region of 3-D coverage, b) increase the spatial resolution to accommodate higher frequency waveform modeling, c) add constraints from other data sources including oil wells and seismic reflection interval velocities, and d) allow for varying depth to Moho. The model is available on the SCEC Data Center at http://www.data.scec.org.

Access to high performance computers such as those at Los Alamos National Laboratory, together with construction of the SCEC 3-D velocity model, have provided us (Kim Olsen and Steven Day) with the capability to examine 3-D wave propagation in the Los Angeles basin and strong ground motion from local earthquakes. Comparisons between surrogate 1-D amplification factors using separate 1-D calculations at each site (top figure), and the averages of complete 3-D calculations (at frequencies of 3 Hz and below) for nine hypothetical earthquakes (including a replay of the 1994 Northridge event) are shown in Figure 2. Shaking is most intense and the duration greatest above the deepest parts of the basin. Directivity effects are clearly seen for the two scenarios involving the San Andreas fault. Simulations of this type, especially for frequencies greater than 0.3 Hz (requiring improved resolution of the velocity model), are being requested by structural engineers to supplement actual recordings of earthquake time-histories for application to performance-based seismic design.

Three groups explored the apparent discrepancy between the historical earthquake record and the magnitude-frequency relationships predicted by the Phase II hazard model. The factor of two disparity between the predicted rate and the historical rate of M6-7+ earthquakes (the predicted rate being greater), if true, has important implications for southern California. Mark Stirling and Steven Wesnousky concluded that the discrepancy could be explained partly by random errors in the geologic data underlying the forecasts, and the randomness of the earthquake process itself. Ross Stein and Tom Hanks argued that the earthquake catalog used in the Phase II report was not complete, and that rates inferred from the more complete recent data were higher and more nearly in agreement with the Phase II model. Ned Field, David Jackson, and James Dolan show that theoretical models based on geology can be constructed that agree well with the historic earthquake catalog. Work continues toward a consensus-based California-wide model in this important area.
Paleoseismic studies in the Los Angeles basin continue to add important new information on fault behavior and earthquake hazard. Charlie Rubin, Scott Lindvall and Tom Rockwell infer that The Sierra Madre fault at the base of the San Gabriel Mountains has slipped about 10.5 meters in two earthquakes during the past 15,000 years, consistent with earthquake magnitudes above 7. This result supports the hypothesis that large earthquakes on the large, composite fault systems in the region relieve most of the north-south strain. Karl Mueller and Tom Rockwell showed that the Compton-Los Alamitos thrust ramp has been inactive for 300,000 years, implying that low-angle faulting below the Los Angeles basin may not be as threatening as perceived in a 1995 report. Nevertheless, the estimated 7 to 9 mm per year of shortening across the Los Angeles basin from geodetic data still implies that strain is accumulating.

Chris Walls, Yehuda Bock, and Tom Rockwell have developed an alternative model (to pure thrust faulting) that integrates GPS data with geological slip rate estimates (Figure 3) to explain the response to crustal shortening. They argue that much of the shortening can be accommodated by eastward and westward extrusion of portions of the southern California crust along known east-west trending strike-slip faults. Below latitude 34.7, geological slip rates and those inferred from GPS data are in very close agreement. Above that latitude, rates inferred from GPS are substantially lower than those implied by GPS. Possible explanations are that geological slip rates are underestimated, that additional faults contribute to the geodetic slip rates, or that more sophisticated modeling is required for adequate comparison of the geodetic and geologic slip rates. Direct measurements of the strain distribution from more sites of the fixed GPS sites of the SCIGN array should help substantially to resolve this question.

Finally, SCEC has long emphasized stress accumulation as a promising criterion to recognize faults or regions most likely to experience earthquakes. A major SCEC contribution in this area has been to produce the important ingredients of stress evolution models for southern California – earthquake catalogs, fault maps, and crustal deformation data. Through the efforts of Ruth Harris, Joan Gomberg and numerous authors, 13 papers resulting from a 1997 SCEC conference on this subject will soon be published in a special edition of JGR. This past year SCEC sponsored another such conference hosted by Ross Stein. Global and regional studies show that both main shock and aftershock occurrence are favored in regions where shear stress has been increased and/or normal stress decreased by previous events. Other investigators have combined stress increment calculations with Jim Dieterich's "rate-state" friction models to calculate the influence of Coulomb stress increments on individual fault segments. Many are working on ways to assess the predictive power of stress evolution calculations for future earthquakes.

Thrust 1: Earthquake Potential and Seismic Hazard Estimation

Thrust 2: Earthquake Geology

Thrust 3: Subsurface Imaging, Seismicity and Tectonics

Thrust 4: Crustal Deformation

Thrust 5: Earthquake Physics

References