Yahoo! is not affiliated with the authors of this page or responsible for its content.
External Grant Award Number 05HQGR0067 IMPROVED THREE-DIMENSIONAL VELOCITY MODELS AND EARTHQUAKE LOCATIONS FOR CALIFORNIA External Grant Award Number 05HQGR0067
IMPROVED THREE-DIMENSIONAL VELOCITY MODELS AND
EARTHQUAKE LOCATIONS FOR CALIFORNIA
Clifford H. Thurber
University of Wisconsin-Madison
1215 W. Dayton St.
Madison, WI 53706
Telephone: (608) 262-6027
FAX: (608) 262-0693
Email: thurber@geology.wisc.edu 1
External Grant Award Number 05HQGR0067
IMPROVED THREE-DIMENSIONAL VELOCITY MODELS AND
EARTHQUAKE LOCATIONS FOR CALIFORNIA
Clifford H. Thurber
University of Wisconsin-Madison
1215 W. Dayton St.
Madison, WI 53706
Telephone: (608) 262-6027
FAX: (608) 262-0693
Email: thurber@geology.wisc.edu
Abstract
Our work focused on the development of a three-dimensional (3D) seismic wavespeed
model for the greater San Francisco (SF) Bay Area and for Northern California. The model has
been used to help validate the USGS 3D velocity model. That model in turn is being used to
compute hypothetical strong ground motions in a simulation of the great 1906 San Francisco
earthquake in preparation for the centennial of the 1906 event. Our models will also help to
characterize both well-known and hidden seismogenic structures, allowing the determination of
improved absolute locations and focal mechanisms for Northern California earthquakes. The
model itself provides new insights into the structure of the California crust and upper mantle.
The datasets combine P-wave arrival times from well-distributed sets of thousands of
earthquakes with a complete archive of available active-source (i.e., explosion) P-wave travel
times from the region. The wavespeed modeling was carried out with a combination of
conventional and "double-difference" seismic tomography. To date, only catalog differential data
have been incorporated.
A number of features are identified in the 3D model, some new and others confirming
previous results. Among these are: the subducting Gorda Plate in the northwest; high-Vp bodies
beneath the northern Great Valley and Sacramento River Delta areas, both interpreted to be
ophiolite bodies; and tentative identification of a detachment fault beneath the SF Bay Area. 2
Project Results
We have been working in collaboration with Tom Brocher of the USGS to assemble an
active-source dataset that is as complete as possible for the Northern California region, including
some data that have never been utilized previously for seismic tomography. We have also
extracted catalog phase arrivals for some of the larger shots recorded at the NCSN stations. That
effort included cross-checking of source and receiver coordinates, examination of individual and
merged travel time curves, and receiver-name rationalization across all experiments (plus the
NCSN). Maps showing the active-source lines for the Bay Area and Northern California are
presented in Figures 1 and 2, respectively.
Figure 1. Map of active-source (explosion and air-gun) surveys compiled by Tom Brocher for
the Bay Area model. Integration and reformatting of these data has been completed. 3
Figure 2. Map of active-source (explosion and air-gun) surveys compiled by Tom Brocher for
the Northern California model (sources in red, stations in black). The crosses indicate the nodes
of the grid used for tomographic inversion.
In parallel, we have extracted arrival-time data for thousands of earthquakes that are
optimally distributed throughout the two study regions (Figures 3 and 4). We have compiled and
merged all of these data and carried out a preliminary "conventional" tomography analysis using
simul2000 [Thurber and Eberhart-Phillips, 1999], followed by double-difference (DD)
tomography [Zhang and Thurber, 2003]. The final model grid for the Bay Area inversion uses 5
to 10 km spacing in the horizontal directions and gradually increasing grid intervals, from 1 to 5
km, in the vertical direction. We use a one-dimensional model from Hole et al. [2000] as our
starting model. We also experimented with using a geologically-based starting model. 4
Figure 3. Map of the earthquakes (circles) for the Bay Area inversion and locations of example
cross-sections through the 3D model (dashed lines) shown in Figure 5.
The Bay Area model covers the region west of the Great Valley from Hollister to Clear
Lake. The geometries of the major faults are clearly defined by the relocated seismicity, and
most of the seismogenic faults are marked by significant wavespeed features (Figure 5). The San
Andreas Fault is southwest-dipping (~70°) and the Calaveras Fault is northeast dipping (~75°) in
the southern part of the model. Bedrock blocks appear to control the distribution of seismicity in
many areas, including an apparent blind thrust fault beneath the edge of the Great Valley in the
northern section (Y = +40 km). The deep high-Vp body associated with these earthquakes is
consistent with Coast Range ophiolite, as proposed by Godfrey et al. (1997). 5
Figure 4. Map of the earthquakes (red circles) and network stations (black triangles) used for our
Northern California model, along with the inversion grid nodes (crosses).
We expanded the active-source dataset (again thanks to the efforts of Tom Brocher;
Figure 2) and constructed a new earthquake dataset (emphasizing events with slightly larger
magnitudes; Figure 4) for our Northern California model. The horizontal gridding ranged from
10 to 20 km, with the vertical gridding comparable to the Bay Area model (1 km near the
surface, increasing to 5 km at depth). The grid extends from the vicinity of Parkfield in the south
to the Oregon border, and from the coast to Nevada except for the far northeastern corner of the
state (Figure 4). 6
Figure 5. Cross-sections through the Bay Area Vp model along the profiles indicated in Figure 3.
Earthquake hypocenters within 5 km of each section are indicated by circles.
San Andreas Fault,
Cienega Winery segment
Paicines
Fault
Ortigalita
Fault
San Andreas Fault,
Loma Prieta segment
Calaveras Fault,
Morgan Hill segment
San Andreas Fault,
San Francisco segment
Hayward Fault,
Berkeley segment
Vacaville-
Kirby Hills
Fault
Blind thrust
fault 7
Interpretation of the model is at a preliminary stage, but some intriguing features are
already evident. To the north, the subducting Gorda Plate is an obvious seismically active, high-
Vp anomaly (Figure 6). A presumed ophiolite body, also with high Vp, is present beneath the
northern Great Valley (Figure 7). Intriguing evidence for a seismically active detachment fault is
observed in the Bay Area (Figure 8).
Both the Bay Area and Northern California 3D models have been put to use by Tom
Brocher for validating his geologically based 3D models. Brocher's models are currently being
used by several research groups for strong-motion simulation and validation work. We are also
beginning to collaborate with Southern California scientists to prepare for the development of a
statewide 3D model.
Acknowledgements
We are indebted to Tom Brocher for his tireless efforts to assemble the most complete
active-source dataset possible, and to Doug Neuhauser for maintaining a superb earthquake data
archive at the Northern California Earthquake Data Center, from which data critical to this
project was obtained.
References
Godfrey, N.J., B.C. Beaudoin, S.L. Klemperer, and the Mendocino Working Group USA,
Ophiolitic basement to the Great Valley forearc basin, California, from seismic and gravity
data: Implications for crustal growth at the North American continental margin, Geol. Soc.
Am. Bull. 109, 1536-1562, 1997.
Hole, J. A., T.M. Brocher, S.L. Klemperer, T. Parsons, H.M. Benz, and K.P. Furlong, Three-
dimensional seismic velocity structure of the San Francisco Bay area, J. Geophys. Res. 105,
13,859, 2000.
Thurber, C., and D. Eberhart-Phillips, Local earthquake tomography with flexible gridding,
Comp. & Geosci., 25, 809, 1999.
Zhang, H., and C.H. Thurber, Double-difference tomography: method and application to the
Hayward fault, California, Bull. Seismol. Soc. Am., 93 1875, 2003.
Bibliography
Lin, G., P. Shearer, E. Hauksson, and C. Thurber, Earthquake locations in southern California
obtained from 3-D velocity inversion and waveform cross-correlation times (abstract),
Seismol. Res. Lett., 76, 267, 2005.
Thurber, C., H. Zhang, V. Langenheim, and T. Brocher, Geophysical characterization of
seismogenic structures in northern California (abstract), Eos Trans. AGU 85, Fall Meet.
Suppl., Abstract S51D-01, 2004.
Zhang, H., and C. Thurber, An overview of double-difference seismic tomography and its
development (abstract), Seismol. Res. Lett., 76, 209, 2005.
Zhang, H., and C. Thurber, Development and applications of double-difference seismic
tomography, "Advances in Studies of Heterogeneities in the Earth's Lithosphere: The Keiiti
Aki Volume II," Pure Appl. Geophys., in press.
Zhang, H., C. Thurber, T. Brocher, and C. Evangelidis A new regional seismic tomography
model for northern California (abstract), SSA Annual Meeting, 2006. 8
Figure 6. Northeast-southwest cross-sections
stepping from northwest to southeast (above)
through the northern portion of our 3D Northern
California model (left). The seismicity and high
Vp of the subducting Gorda Plate is clearly
evident c