= $thisPage ?>
UNIVERSITY OF CALIFORNIA SAN DIEGO
Analysis of Data Recorded on the ANZA Seismic Network
A dissertation submitted in partial satisfaction of the requirements for the degree: Doctor of Philosophy in Earth Sciences
by Frank L. Vernon III
Committee in charge:
- Doctor Jonathan Berger, Co-Chairman
- Professor James N. Brune, Co-Chairman
- Doctor Thomas Hanks
- Professor Enrique Luco
- Professor John A. Orcutt
- Professor Robert L. Parker
- Adjunct Professor David J. Thomson
© University of California, San Diego, 1989
Table of Contents
- Signature Page
- Table of Contents
- List of Tables and Figures
- Vita and Publications
I. SEISMICITY OF THE SAN JACINTO FAULT ZONE
- 1. Introduction
- 2. Structure of the San Jacinto
- 3. Geologic Slip Rates
- 4. Historical Earthquakes
- 4.1. Paleoseismic Data
- 4.2. 1890 and 1892 Earthquakes
- 4.3. 1899, 1918, 1923 Earthquakes
- 4.4. 1937, 1942, 1954, 1968, 1969 Events
- 4.5. 1980 and 1987 Events
- 5. Seismic Slip, Seismic Gaps and Earthquake Probabilities
- 6. Conclusions
II. ANZA SEISMIC NETWORK: DESCRIPTION
- 1. Introduction
- 2. Field Instrumentation
- 3. The Detection and Recording System
- 4. Coherence Between Seismometers
- 5. Ground Noise
- 6. ANZA Data
- 7. Conclusions
- Appendix 1 - Station Descriptions
- Appendix 2 - Instrument Response of the ANZA Network
- Appendix 3 - Interstation Distances and Azimuths
III. INITIAL RESULTS FROM THE ANZA DATA
- 1. Introduction
- 2. Hypocenter Estimation
- 3. Seismicity Distribution
- 4. Earthquake Source Parameter Estimation Procedure
- 5. Equation Source Parameters
- 6. Borehole Data
- 7. Stress Drops
- 8. Conclusions
IV. MULTITAPER SPECTRAL ANALYSIS OF HIGH-FREQUENCY SEISMOGRAMS
- 1. Introduction
- 2. The Multitaper Algorithm
- 3. Spectral Comparisons Using Seismic Data
- 4. Statistical Comparisons
- 5. Conclusions
- Appendix: Calculating P ϖ Prolate Eigentapers
V. COHERENCE OF SEISMIC BODY WAVES
- 1. Introduction
- 2. Site Choice
- 3. Experimental Design
- 4. The Data Set
- 5. Data Analysis Techniques
- 6. Data Analysis
- 7. Earthquake Source Parameter Estimation
- 8. Discussion
- 9. Conclusions
- Appendix: Multiple Window Spectrum Analysis
VI. FREQUENCY DEPENDENT POLARIZATION ANALYSIS OF HIGH-FREQUENCY SEISMOGRAMS
- 1. Introduction
- 2. Polarization Analysis with the Multitaper Algorithm
- 3. A Synthetic Example
- 4. Data Examples
- 5. Conclusions
- Appendix: Formal Uncertainty of Polarization Estimates
LIST OF FIGURES AND TABLES
- CHAPTER I:
- Table 1. – Significant Right-Lateral Events Along San Jacinto Fault Zone
- Figure 1. – Fault map of southern California with the individual strands of the San Jacinto fault zone identified. The heavy dashed lines mark the limits of the surface outcrops of the Peninsular Ranges batholith.
- Figure 2. – Rupture areas of historical earthquakes and the existing seismic slip gaps along the San Jacinto fault zone. Rupture areas enclosed with a dashed line are not well constrained.
- Figure 3. – Fault segments on the San Jacinto fault zone defined by WGCEP . Percentages are the estimates of the probability of whether the segment filling earthquake will occur in the next 30 years.
- Figure 4. – The probability, probability density function, and 30-year condi- tional probability for the segment filling earthquake in the Anza-Coyote Mountain segment measured in years from the previous earthquake.
- CHAPTER II:
- Table 1. – Station Locations
- Table 2. – HS-10 Specifications
- Table 3. – Overall System Gain
- Figure 1. – All ML (Quality A and B) located since the upgrade of the southern California seismic network in 1975 are shown on the left with the Anza gap [Sanders and Kanamori, 1984] and the Anza seismic slip gap of Thatcher et al. . The expanded view gives the locations of the ANZA network stations (filled triangles), the USGS strong motion accelerograph network (open squares), and the epicenters of all ML > 4.0 earthquakes and their magnitudes since 1975.
- Figure 2. – Geological map of the Anza area copied from the 1:250,000 scale geological map of California, Santa Ana sheet. ANZA network sta- tions are identified by solid blue circles.
- Figure 3. – Map and schematic diagram of the ANZA telemetry network. Remote stations are identified by the open triangles and transmit data via VHF links (short dashed lines) to Toro Peak. The data is multiplexed on Toro and sent by microwave (dashed line) telemetry to Soledad Mt. where it is relayed to IGPP.
- Figure 4. – Block diagram of the functional elements at a remote field station.
- Figure 5. – Analog system response for the ANZA remote field station given in group delay, phase, and amplitude. Solid and dashed lines identify response using 62.5 and 100 Hz anti-alias filters respectively.
- Figure 6. – Block diagram of the functional elements at the Toro Peak relay station.
- Figure 7. – Power spectrum, phase, and coherence of ground noise measured on two vertical seismometers located at PFO.
- Figure 8 – Power spectrum, phase, and coherence of local earthquake measured on two vertical seismometers located at PFO.
- Figure 9. – Maximum, median, and minimum vertical acceleration ground noise spectra using data from all ANZA stations during a measurement in July, 1984.
- Figure 10. – Cumulative number of events recorded on the ANZA network as a function of time.
- Figure 11. – Epicenters of earthquakes recorded by the ANZA seismic network
- CHAPTER III:
- Table 1. – ANZA Velocity Models
- Table 2. – Source Parameter Stabilities
- Figure 1. – Ray paths for P and S waves through the one dimensional velocity model (Table 1) for the ANZA network generated by a joint hypocenter-velocity structure inversion program. Seismic sources were placed at 2.5, 5, 10, and 20 km depth in the model. At each depth the same take-off angles from the source were used for P and S waves.
- Figure 2. – Seismicity recorded on the ANZA network in map view and cross-sections. Four clusters of seismicity are identified: Anza (AN), Cahuilla (CA), Hot Springs (HS), and Table Mountain (TM). All depths are referenced to the top of the Toro Peak (elevation 2657 m). ANZA network stations are identified by the solid tri- angles.
- Figure 3. – Cumulative number of magnitude ML ≥ 2.0 events from the four clusters between 1975 and 1989. The seismicity from the Buck Ridge region has been merged with the TM cluster. The data is from the southern California seismic network.
- Figure 4. – Comparison of the source moment and radius estimates calculated from P and S waves for the ANZA dataset. Only parameters deter- mined with a minimum of six stations of data are shown. Dashed lines mark the ratios between the P and S wave values of 2 for source radius and 3 for scalar seismic moment.
- Figure 5. – Stress drop vs depth plots for the Brune and arms methods. The clusters from Figure 2 are marked by a different symbols. The values of stress drop tend to increase with depth.
- Figure 6. – Comparison of the arms and Brune stress drop estimates. The arms estimator gives larger values by a factor of 2 to 3 on average. The symbols designate the same clusters as is Figure 5.
- Figure 7. – Seismic moment vs source radius data for P and S waves. The solid lines mark values of constant stress drop. There appears to be a minimum bound for the size of about 50 meters for the source radius estimates from the ANZA network data.
- Figure 8. – Displacement amplitude spectra for the P wave of a local earth- quake measured at the KNW borehole site. Spectra from surface sensors and from borehole seismometers at 150 and 300 meter depth are displayed. The spectra from the surface measurements show large amplifications in the frequency band of 10 to 40 Hz as well as strong attenuation above 60 Hz when compared to the spectra from the boreholes.
- Figure 9. – Comparison of fc , Ω0 , and Δσ , determined from the surface, 150 meter, and 300 meter deep seis- mometers at the KNW borehole site. The values of Ω0 and Δσ calculated from data recorded on the surface are larger than the borehole data. The estimates of fc from the surface sensors are centered between 40 and 50 Hz for the P waves while the values from both borehole sensors range from 25 to 90 Hz. The estimates of fc from the surface sensors are scattered between 20 and 40 Hz for the S waves without an obvious relationship to the borehole measurements.
- Figure 10. – Corner frequency picks arranged by seismometer orientation for all stations in the ANZA network. fmax values for each station are indicated by a diamond for the horizontal components.
- Figure 11. – Seismic moment vs source radius values determined from the ANZA seismic network which are averages from six or more station esti- mates. The arrows indicate the correction to the surface mea- surements by using the data from the 300 meter deep seismometers at KNW.
- Figure 12. – Seismic moment vs source radius values determined from the ANZA seismic network which are averages from six or more station esti- mates. Crosses identify the individual KNW station source parameter estimates.
- CHAPTER IV:
- Table 1. – Fractional Containment of Eigentapers
- Table 2. – Statistical Comparison
- Figure 1. – Comparison plot of boxcar, Hann, and 20% cosine tapers
- Figure 2. – The five lowest-order 4ϖ prolate eigentapers. The zeroth-order eigentaper v(0) is plotted with a solid line, and the higher-order tapers are plotted with dashed lines.
- Figure 3. – Fourier transform amplitudes of the five 4ϖ prolate tapers shown in Figure 2, using the same conventions for dashed and solid lines.
- Figure 4. – (Top) Comparison of unsmoothed and smoothed estimates of the spectrum of a high-frequency S wave. The spectra are plotted on a log-linear scale and are offset to facilitate comparison. The boxcar spectral estimates are graphed with a solid line. The dashed lines at the top of each of the lower figures are spectral estimates employing a Hann taper. The middle curves are spectral estimates obtained using a 20% cosine taper.
- Figure 5. – A multitaper spectral estimate (solid line, labeled a) of the frequency content of an SH wave (top) is compared with direct spectral estimates using the boxcar taper (fine dashed line, labeled b), 20% cosine taper (coarse dashed line, labeled c), and Hann taper (asterisks, labeled d). The spectra are plotted using linear scales for the horizontal and vertical axes. The boxcar, 20% cosine, and multitaper estimates of the S wave spectrum are almost identical, but the Hann taper estimate is substantially different. This is because the Hann-tapered spectra overemphasize the data in the center of the time series and downweight data points toward the ends of the record. The section of the time series which was analyzed is bracketed by dashed lines in the seismogram at the top.
- Figure 6. – Comparison of the leakage of various estimates of the spectrum of a vertical seismogram recorded 412 km away from a Nevada Test Site explosion. The spectral estimate using a cosine taper (asterisks, labeled d) and the multitaper spectral estimate (solid line, labeled a) give good representations of the spectra of the seismic signal (0 -20 Hz) and ground noise (20 -60 Hz). The spectra are plotted using a log-linear scale.
- Figure 7. – Comparison of the variance and broadband bias of several single-taper spectral estimates (solid circles) and the multitaper estimates (solid triangles).
- CHAPTER V:
- Table 1. – Source Parameters
- Table 2. – P Wave Spectral Estimates
- Table 3. – S Wave Spectral Estimates
- Figure 1. – Geology and topography of Pinyon Flat from Wyatt . The heavy circle shows the radius of this seismic array.
- Figure 2. – Location of magnitude 3 or greater events (circles) from the Cal- tech catalog for the years 1980-1984. The events which were recorded on this are marked with asterisks.
- Figure 3. – Layout of coherence array.
- Figure 4. – Seismograms recorded on the transverse component from event 8. The section of the seismograms used for the shear wave analysis is bounded by the dashed lines.
- Figure 5. – (a). Mean log power spectrum with 95% confidence limits for the P wave on the vertical components of event 1. (b). Individ- ual spectra from which the log average spectrum was calculated. Average error bars for individual spectrum are given.
- Figure 6. – The data from station AZA is the common time series and the MSC estimates of the common series with each of the other stations' data are shown in each successive plot. The ±95% confidence limits (short dashed lines) of the Γ2( f ) are plotted as well as the theoretical 95% confidence limit (for Gaussian data) that MSC ≠ 0 (dashed line).
- Figure 7. – Contour plots of average magnitude squared coherence for the com- ponents of the P wave plotted against distance and frequency. Black signifies region where coherence has less than 95% confidence of being not equal to 0. The right side plots show the average signal to noise ratio for all event for each phase. All data where the signal to noise ratio Σ2 ≤ 20 was not used for the contour plots.
- Figure 8. – Contour plots of average magnitude squared coherence for the com- ponents of the S wave plotted against distance and frequency. Black signifies region where coherence has less than 95% confidence of being not equal to 0. The right side plots show the average signal to noise ratio for all event for each phase. All data where the signal to noise ratio Σ2 ≤ 20 was not used for the contour plots.
- Figure 9. – Displacement amplitude spectrum and the ±95% confidence lim- its for station AZE, Event 8, from the transverse component of the S wave. The solid curve is the best fitting model using equation 7.1 where fc = 10 Hz, Ω0 = 5.6 x 10-5 cm-sec, N = 3.5. The point ( Ω0 , fc ) is shown by the shaded triangle. The dashed curve uses equation 7.1 except the corner frequency is calculated by Snoke's method and N = 2. The point ( Ω0 , fc ) is marked with the filled circle.
- Figure 10. – (a). Average amplitude spectral ratios between all station pairs for each event on the P wave, vertical component. (b). Maximum amplitude spectral ratio at each frequency between any station pair for each event on the P wave, vertical component.
- Figure 11. – a. Average amplitude spectral ratios between all station pairs for each event on the S wave, vertical component. b. Maximum amplitude spectral ratio at each frequency between any station pair for each event on the S wave, vertical component.
- Figure 12. – Comparison of the averaged magnitude squared coherence estimates between sets of sensors spaced approximately 150 meters apart. Three pairs of sensors are used, as shown in the top diagram, with two measurements in the vertical direction and one surface horizontal pair. The borehole sensors, KN1 and KN2 are placed at 300 and 150 meters depth, respectively, while KN3 is located on the surface. The selected PFO data used station pairs which are 127 meters apart. At least 10 events were used to form the aver- age MSC for each pair of sensors. At frequencies above 30 Hz the coherence between KN1 and KN2 is much greater than the PFO sur- face values. The apparent holes in the MSC associated with KN1 and KN2 sensors is probably due to interference effects caused by the free surface reflection.
- CHAPTER VI:
- Figure 1. – Diagram to illustrate the definitions of the polarization angles ΘH and ΘV. The azimuth ΘH is restricted to [-180, 180] and is measured counterclockwise from ê2. The angle ΘH is chosen by determining the maximum horizontal displacement of the particle motion for which ΘV will fall in the range 0 ≤ ΘV ≤ 90. The ellipticity of the particle motion is defined by the amplitudes |z1|, |z2|, |z3| and the phase angles φHH and φVH (defined in text).
- Figure 2. – Polarization test series.
- Figure 3. – (a) Amplitude spectra and polarization angles calculated from the test series. Spectra for components 1 (solid line), 2 (coarse dashed line), and 3 (fine dashed line). (b) The singular value associated with principal polarization is plotted against fre- quency (solid line), and the secondary singular values (dashed lines). (c) Horizontal azimuth of particle motion. (d) Phase angle defined by the major and minor axis of the horizontal par- ticle motion ellipse. (e) Angle of incidence of particle motion measured from nadir. (f) Phase angle defined by major and minor axis of the vertical particle motion ellipse.
- Figure 4. – Anza data used in polarization example. Range in kilometers and expected ΘH are given in right-hand columns. Maximum amplitude in counts is given in left-hand column, along with station code and component number.
- Figure 5. – (a) Plots of precursory waveform observed on station FRD. (b) Plots of precursory waveform observed on station CRY. The portion used for spectrum analysis is bounded by dashed lines. Both horizontal components at station FRD exhibit visible 60-Hz power line noise. The spectral leakage resistance of the 4ϖ prolate eigentapers used in the analysis guards against bias in the frequency band of interest.
- Figure 6. – Amplitude spectra and polarization angles for precursory waveform observed at station FRD. Solid/dashed line conventions are identical to those of Figure 3.
- Figure 7. – Amplitude spectra and polarization angles for precursory waveform observed at station CRY. Solid/dashed line conventions are identical to those of Figure 3.
- Figure 8. – Three-component seismogram for Superstition Mountain event observed at Anza station KNW. The 14-s segment chosen for polarization analysis is within the dashed lines.
- Figure 9. – Amplitude spectra and polarization angles for the 14-s P coda segment shown in Figure 8. Solid/dashed line conventions are identical to those of Figure 3.
- Figure 10. – Amplitude spectra and polarizaton angles for the first 2 s of the 14-s P coda segment shown in Figure 8. Solid/dashed line conven- tions are identical to those of Figure 3.
- Figure 11. – Amplitude spectra and polarization angles for the seventh and eighth seconds of the 14-s P coda segment shown in Figure 8. Solid/dashed line conventions are identical to those of Figure 3.
- Figure 12. – Amplitude spectra and polarization angles for the eleventh and twelfth seconds of the 14-s P coda segment shown in Figure 8. Solid/dashed line conventions are identical to those of Figure 3.
The completion of this thesis is the result of a tremendous amount of support and effort made by many colleagues, fellow graduate students, and friends. It is not possible to properly acknowledge the contributions of each individual towards this final product, and in recognizing this I ask the forgiveness and understanding of those who are not mentioned by name.
The design, the installation, and operation of the ANZA seismic network, required the expertise of many different people. The pri- mary cast of characters on the ANZA project included Larry Baker, Joe Fletcher, Linda Haar, and Tom Hanks from the USGS, and James Batti, Jon Berger, Jim Brune, and Jennifer Scott from IGPP. The data logging equipment was designed and built by the people at Refraction Technology, particularly Paul Passmore, Bill Witkowski, and Charles Garcera. They were very responsive to suggestions for improving various sections of the telemetry sys- tem. Equally important to the success of this project is the generosity of the landowners at each site for giving us a place to operate our equipment. The ANZA project has benefited from the services of Mert Ingram, who helped design and build seismometer and antenna mounts for the stations, and Gary VanCouver- ing, who spent many hot summer days pouring concrete, digging trenches, and installing cabling at every site. Of singular importance are
Any project of this magnitude can not exist without the unsung heroes who can find a way to make things work within a bureaucracy. During my graduate student career I have been under the wings of five masters: Kate Harps, Shelley Marquez, and Susie Mueller at IGPP, Betty Stover in the Scripps Graduate office, and Patty Cuneo at the USGS. Elaine Blackmore, Don Betts (computer graphics repairman), and Kate Harps guided me through the thesis preparation gauntlet.
I would like to thank the members of my committee for giving me guidance when needed, and the freedom to explore and develop new ideas. I have benefited from many stimulating discussions with Duncan Agnew, Alan Chave, Craig Lindberg, and Jeff Park. Parts of the data analysis required hypocentral locations and arrival times recorded on the southern California seismic network, which were generously provided by Lucy Jones from the Pasadena office of the USGS.
Finally, I want to thank my parents. They have blessed me with their love and support and have stimulated my interests in science and in life.
April 6, 1954 -- Born -- Pasadena, California
1989 - PhD, University of California, San Diego.
1980 - Staff Research Associate, Institute of Geophysics and Planetary Physics, SIO.
1977-1980 Development Technician, Institute of Geophysics and Planetary Physics, SIO.
1976-1977 Laboratory Assistant, Physics Department, University of California at San Diego.
Summer 1976 Geophysicist, Shell Oil Company, Houston, Texas.
Summer 1975 Geophysicist Shell Oil Company, Houston, Texas.
Summer 1974 Field Geophysicist, Shell Oil Company, Gaylord, Michigan.
Summer 1973 Science Aide, U.S. Geological Survey, Los Angeles, California.
Sales, B. C., M. B. Maple, and F. L. Vernon III (1978). Initial oxidation kinetics near the Curie temperature of nickel, Phys. Rev. Bull. 18, 486-491.
Brune, J. N., R. S. Simons, F. Vernon, L. Canales, and A. Reyes (1980). Digital seismic event recorders: description and exam- ples from the San Jacinto fault, the Imperial fault, the Cerro Prieto fault, and the Oaxaca, Mexico subduction fault, Bull. Seism. Soc. Am. 70, 1395-1408.
Munguia, L., J. N. Brune, A. Reyes, J. Gonzalez, R. Simons, and F. Vernon (1978). Digital seismic event recorder records and spectra for aftershocks of the November 29, 1978 Oaxaca earth- quake, Geofis. Int. 17, 359-366.
Brune, J. N., F. L. Vernon III, R. Simons, J. Prince, and E. Mena (1982). Strong motion data recorded in Mexico, in U.S.G.S. Prof. Paper 1254, The Imperial Valley, California, earthquake of Octo- ber 15, 1979, U.S. Geol. Survey, Menlo Park, Calif.
Chavez, D., J. Gonzalez, A. Reyes, M. Medina, C. Duarte, J. N. Brune, F. Vernon, R. Simons, L. K. Hutton, P. T. German, and C. E. Johnson (1982). Mainshock location and magnitude determina- tion using combined U.S. and Mexican data, in U.S.G.S. Prof. Paper 1254, The Imperial Valley, California, earthquake of Octo- ber 15, 1979, U.S. Geol. Survey, Menlo Park, Calif.
Reyes, A., J. Gonzalez, L. Munguia, A. Nava, F. Vernon, and J. N. Brune (1978). Locations of aftershocks of the Oaxaca earthquake using smoked paper recorders and digital event recorders, Geofis. Int. 17, 341-357.
Prince, J., E. Mena, I. Mora, J. Brune, L. Alonso, F. Vernon (1982). Observations of damage and intensity, in J.G. Anderson and R.S. Simons, eds., The Mexicali Valley Earthquake of 9 June 1980, EERI Newsletter 16, 87-93.
Anderson, J. G., J. N. Brune, J. Prince, and F. L. Vernon III (1983). Preliminary report on the use of digital strong motion recorders in the Mexicali Valley, Baja California, Bull. Seism. Soc. Am. 73, 1451-1468.
Berger, J., L. N. Baker, J. N. Brune, J. B. Fletcher, T. C. Hanks, and F. L. Vernon (1984). The Anza array: a high dynamic-range, broad-band, digitally radio-telemetered seismic array, Bull. Seism. Soc. Am. 74, 1469-1682.
Fletcher, J., L. Baker, T. C. Hanks, F. Vernon III, J. Berger and J. Brune (1984). Seismicity and source parameters from the digi- tal array at Anza, California, in Proc. 8th World Conf. Earth- quake Engineering, July 21-28, San Francisco, Calif., 215-222.
Fletcher, J. B., L. C. Haar, F. L. Vernon, J. N. Brune, T. C. Hanks, and J. Berger (1986). The effects of attenuation on the scaling of source parameters for earthquakes at Anza, California, in Earthquake Source Mechanics, Geophys. Monog. 37 (M. Ewing V. 6), Amer. Geophys. Union, Washington, D.C., 331-338.
Brune, J. N., J. Fletcher, F. Vernon, L. Haar, T. Hanks, and J. Berger (1986). Low stress-drop earthquakes in the light of new data from the Anza, California telemetered digital array, in Earthquake Source Mechanics, Geophys. Monog. 37 (M. Ewing V. 6), Amer. Geophys. Union, Washington, D.C., 237-246.
Frankel, A., J. Fletcher, F. Vernon, L. Haar, J. Berger, T. Hanks, and J. Brune (1986). Rupture characteristics and tomo- graphic source imaging of $M sub L$3 earthquakes near Anza, Southern California, J. Geophys. Res. 91, 12,633-12,651.
Fletcher, J., L. Haar, T. Hanks, L. Baker, F. Vernon, J. Berger, and J. Brune (1987). The digital array at Anza, California: Pro- cessing and initial interpretation of source parameters, J. Geo- phys. Res. 92, 369-382.
Park, J., F. L. Vernon III and C.R. Lindberg (1987). Frequency dependent polarization analysis of high frequency seismograms J. Geophys. Res. 92, 12,664-12,674.
Park, J., C. R. Lindberg, and F. L. Vernon III (1987). Multita- per spectral analysis of high frequency seismograms. J. Geophys. Res. 92, 12,675-12,684.
Berger, J., H. Eissler, F. L. Vernon, I. L. Nersesov, M. B. Gokhberg, O. A. Stolyrov, and N. D. Tarasov (1988). Studies of high-frequency seismic noise in Eastern Kazakhstan. Bull. Seis- mol. Soc. Amer. 78, 1744-1758.
Hough, S., J. G. Anderson, J. Brune, F. Vernon III, J. Berger, J. Fletcher, L. Haar, T. Hanks, and L. Baker (1988). Attenuation near Anza, California. Bull. Seism. Soc. Am., 78, 676-691.
URL: http://eqinfo.ucsd.edu/faq/thesis_vernon/index.php [Last updated: 2010-11-23 (327) 23:01:44 UTC]