Douglas S. Dreger

Determination of source parameters at regional distances with three-component sparse network data

We find remarkable similarities between regional body waves recorded by the TERRAscope network of broadband stations and synthetics constructed from a standard southern California velocity model. This model is shown to be effective for a variety of azimuths and ranges throughout southern California. At short periods some of the relative timing of the body waves are discordant, but at longer periods this becomes less of a factor. Thus we have developed a waveform inversion technique to rapidly determine source parameters using stored Greens functions for events out to 500 km, well outside the TERRAscope network. Often, only the three-component records of a single station are required because the ratio of SV to SH energy is dependent upon source orientation. Sensitivity analyses examining the effects of source mislocations and velocity model on the inversion results show that the long-period body waves appear relatively insensitive to lateral mislocations but are sensitive to source depth. However, the choice of velocity model can be a factor in obtaining reliable estimates of source depth. In this study the October 24, 1990, (Mw = 5.2) Lee Vining and the December 3, 1991, (Mw = 5.1) Baja California events are used to demonstrate the effectiveness of the inversion method. For the Baja event, we obtained unique results using a single station. For the Lee Vining event, inversions using a single station were not as stable. However, we found that using two stations with only a 24° aperture provided enough constraint to obtain unique results.

Seismic remote sensing for the earthquake source process and near-source strong shaking: A case study of the October 16, 1999 Hector Mine earthquake

A new methodology to automatically determine the earthquake source process and near-source strong ground motions using regional and local distance data was used to study the October 16, 1999 Hector Mine earthquake (MW7.1). Broadband displacement data were inverted for the seismic moment tensor and to resolve the fault plane ambiguity independent of aftershock location and surface faulting information. The identified NNW striking fault plane was used to invert for the distribution of fault slip over a planar surface. The slip model was used to predict near-fault peak ground velocity, which was found to agree well with the TriNet ShakeMap, and also with near-fault observations that were not available at the time of the initial analysis. Our method if automated is capable of providing near-source strong ground motion information within 30 minutes of the earthquake origin time even in cases where there are few near-fault recording instruments.

Empirical Green's function study of the January 17, 1994 Northridge, California earthquake

The January 17, 1994 Northridge main-shock was analyzed using an empirical Greens function approach to estimate the spatial and temporal distribution of fault slip. Far-field source time functions obtained from seismograms recorded by the Berkeley Digital Seismic Network (BDSN) and TERRAscope stations reveal the distributed nature of the source and the presence of a second large subevent. The location of the fault beneath the San Fernando valley as well as the updip directivity of the rupture contributed to the locally heavy damage due to this earthquake.

85.11 – TDMT_INV: Time Domain Seismic Moment Tensor INVersion

This chapter provides an overview of time domain seismic moment tensor inversion. It is a set of programs and shell scripts for determining the seismic moment tensor by inversion of complete, three-component seismic waveforms. Included are source code for programs to compute the required Greens function catalog, various utilities, and the moment tensor inverse routine. The frequency-wavenumber integration program, FKRPROG, written by Chandan Saikia of URS, is included with permission and is used to compute the Greens functions for appropriate one-dimensional seismic velocity models. In addition, an example “wrapper” program is included to illustrate how this package may be automated. The distribution has been built on PCs running Linux and Sun workstations running Solaris. It compiles using the GNU C (gcc) and GNU Fortran (g77) compilers. This software package has been in use at the University of California, Berkeley Seismological Laboratory (BSL) since 1993 and is employed to automatically investigate all M L > 3.5 events in northern California. The package has also been successfully implemented at the Japan National Research Institute for Earth Science and Disaster Prevention. In addition, it has been used by a number of individual researchers in the United States, Europe, and Asia.

The Combined Inversion of Seismic and Geodetic Data for the Source Process of the 16 October 1999 Mw 7.1 Hector Mine, California, Earthquake

We investigate the source process of the M w 7.1 Hector Mine earthquake by inverting broadband regional and local seismic displacement waveforms com- bined with Global Positioning System (GPS) and synthetic aperture radar interfer- ometry (InSAR) geodetic measurements. We find that the three data sets individually produce remarkably similar slip distributions over a multisegment fault. A simulta- neous inversion of the three data sets is presented, and the sensitivity of the combined inversion to the weighting of the three independent data sets is examined. The results indicate that the overall length of the fault that slipped is 42 km with a peak slip of 5.50 m and total scalar seismic moment of 6.8 10 19 N m. The majority of slip is located on the western branch of the Lavic Lake fault to the northwest of the hypo- center, although the overall rupture is bilateral with appreciable slip on the Bullion fault to the southeast. Some slip is located on an eastern branch of the Lavic Lake fault, an apparent bifurcation, which is also evident from aftershock locations. The average slip and stress drops are 2.70, 1.62, 1.60 m and 78, 36 and 55 bars for the western Lavic Lake, the eastern Lavic Lake, and Bullion faults, respectively. The results also indicate that the overall rupture process is relatively slow, which is char- acterized by a several-second delay before the onset of significant slip, a 1.8 km/sec rupture velocity of the primary asperity, and long, spatially variable, dislocation rise times.

Rupture Process of the 26 January 2001 Mw 7.6 Bhuj, India, Earthquake from Teleseismic Broadband Data

We investigate the rupture process of the 26 January 2001 Bhuj, India, M w 7.6 earthquake through inversion of teleseismic broadband body waves. This earthquake ranks as one of the most important recent events due to its occurrence within a stable continental interior, where such events are rare. The Bhuj earthquake occurred on a moderately dipping blind thrust fault within an ancient failed rift. About 70% of the seismic moment released in the earthquake was confined to a very small area (∼375 km2) surrounding the hypocenter and at depths below 12 km. The static stress drop of the Bhuj earthquake is anomalously high (∼20 MPa). The source time history of the event indicates very rapid onset to the moment release and most likely high slip velocities within the deep asperity. This suggests that some of the damage near the epicenter may have been caused by anomalously high-frequency ground motions. The teleseismic data also indicate the presence of a second area of large slip in the shallow part of the Bhuj fault, although the depth extent of this shallow large-slip area is not resolved. Comparisons of the predicted ground motions with observed intensities suggest that substantial slip occurred in the upper 10 km of the fault in order to explain the distribution of high intensities to the west and northwest of the fault. The upper surface layers near the Bhuj fault consist of unconsolidated, low-rigidity sediments and alluvium. The upper ∼ 10 km of the Bhuj fault may therefore be in a conditionally stable region that normally deforms through aseismic creep and can sustain seismic rupture only when dynamically stressed by rupture of the high-strength deep asperity. We suggest that this deep asperity may be related to a lithologic anomaly of ultramafic composition.

Slip of the 2004 Sumatra-Andaman Earthquake from Joint Inversion of Long-Period Global Seismic Waveforms and GPS Static Offsets

The 26 December 2004 Great Sumatra-Andaman earthquake opened a new era for seismologists to understand the complex source process of a great earth- quake. This is the first event with moment magnitude greater than 9 since the de- ployment of high-dynamic-range broadband seismic and Global Positioning System (GPS) sensors around the globe. This study presents an analysis of the ruptured fault- plane geometry and slip distribution using long-period teleseismic data and GPS- measured static surface displacements near the fault plane. We employ a rupture geometry with six along-strike segments with and without a steeper down-dip ex- tension. The fault segments are further subdivided into a total of 201 � 30 30 km fault patches. Sensitivity tests of fault-plane geometry and the variation in rupture velocity indicate that the dip and curvature of the fault plane are not well resolved from the given data set and the rupture velocity is constrained to be between 1.8 and 2.6 km/sec. Error estimations of the slip distribution using a random selection of seismic and GPS station subsets (50% of all stations) illustrate that slip is well re- solved along the whole rupture and the mean slip uncertainty is less than 1.5 m (about 11%). Although it is possible that near-field GPS data include contributions from additional postseismic transient deformation, our preferred model suggests that the Sumatra-Andaman earthquake had a magnitude of Mw 9.20 0.05/0.06. Online material: Comparison of slip models, GPS modeling, waveform fit, fault geometry, and inversion parameters.

Finite-Source Modeling of the 1999 Taiwan (Chi-Chi) Earthquake Derived from a Dense Strong-Motion Network

The 1999 Chi-Chi earthquake (M W 7.6) (20 September 1999, 17:47:15.9 UTC) (located at 23.853 N, 120.816 E, and depth of 7.5 km) inicted severe re- gional scale damage to Taiwan. The strong-motion waveeld was captured by a dense network of stations (with average station spacing of 5 km), which represents the most complete strong-motion dataset to date to use to study the kinematic source process of an earthquake. We inverted velocity waveforms recorded by 21 stations for the spatial variation in slip on a planar fault model composed of 416 subfaults, each with a dimension of 3.5 km by 3.5 km. The planar model has a strike of N5 E and a dip of 30 E, and the inversion solves for the direction and magnitude of the slip. To account for possible temporal source complexity we allowed each subfault to slip within 10 overlapping time windows, each with a duration of 3 sec. The results show that the source is composed primarily of three major asperities, the rst of which is mainly dip slip, extending from the hypocenter to the northern end of the surface rupture. In this asperity, slip occurred in two pulses separated in time by 5 sec. The dislocation rise time for each pulse is short (3n4 sec), yielding an approximate av- erage slip velocity of 80 cm/sec. The second asperity is located at shallow depth near the northern end of the rupture where very large ground velocities were observed. This asperity is on average oblique and shows a temporal rake rotation from pure dip-slip to strike-slip. The rotating rake suggests a low initial shear-stress on the northern end of the fault. Slip in this asperity is dominated by a large pulse with a dislocation rise time of 8 sec. A station near the northern end of the surface rupture recorded a peak velocity of 390 cm/sec, which we nd to be due to the constructive interference of energy radiated from the rst two asperities. The third asperity is located south of the epicenter. The total moment from the three asperities is 4.1 10 27 dyne cm, which was released over a period of 30n35 sec within an area of 900 km 2 . Synthetics calculated from the three-asperity model explain 85% of the data and represent 98% of the total variance reduction. Our results indicate that slip is conned to the shallow region of the fault, deep slip patches are less constrained, and that the slip distribution may be representative of fault segmentation along the Chelungpu fault system.

Monitoring of strain release in central and northern California using broadband data

Systematic cataloguing of seismic moment, depth and mechanism of regional earthquakes down to magnitude M∼4 can now be achieved in close to real-time using data from sparse networks of digital broadband stations. The procedure we have developed relies on 3 independent methods for the determination of moment tensor, providing confidence limits on the results. We show the results of application to the seismicity in central and northern California for a one year period starting in March 1992 and illustrate how patterns of stress and strain release can now be monitored systematically and reliably in a timely fashion.

Source parameters of the Sierra Madre Earthquake from regional and local body waves

We inverted the three-component, long-period data recorded by the TERRAscope array for the June 28, 1991 Sierra Madre, California event to determine the seismic moment and source orientation of the mainshock (ML=5.8). Remarkably four of the six stations were located 159.2±0.7 km from the epicenter. Variations in absolute traveltime were smaller than 2% and 3% for Pn and Sn, whereas the surface waves displayed greater variation. Similarities in the waveforms and the small variance in traveltime suggest that a common Greens function can be used to invert the data. We used the whole waveforms excluding the fundamental surface waves at the more distant stations to invert for source parameters. The results of the inversion indicated that this event was predominately a thrust type earthquake, where the strike, rake and dip were determined to be 235°, 74° and 50°, respectively. The seismic moment was determined to be 2.5 × 1024 dyne-cm. The source duration was found to be about 1.0 seconds by direct measurement of the S-wave recorded at the Pasadena station (located 21 km from the epicenter). Considering circular rupture we obtain a stress drop of 460 bars.