Rupture characteristics of California earthquakes

Citation

Wald, David Jay (1993) Rupture characteristics of California earthquakes. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/0efb-ty71. https://resolver.caltech.edu/CaltechTHESIS:01152013-161612475

Abstract

The rupture characteristics of the 1987 Superstition Hills (M_s = 6.6), the 1989 Loma Prieta (M_s = 7.1), and the 1991 Sierra Madre (M_L = 5.8) earthquakes were determined using a constrained, damped, least-squares inversion of strong motion and teleseismic waveforms. Extension of the modeling procedure to employ teleseismic, empirical Green's functions allowed determination of faulting details of a fourth earthquake, the great 1906 San Francisco event.

The 1987 Superstition Hills earthquake was the second and larger of two significant earthquakes that occurred on conjugate faults in the western Imperial Valley. The first event (M_s = 6.2), located on the Elmore Ranch Fault, had a geometry and mechanism favorable for triggering the larger event on the Superstition Hills Fault some 12 hours later. The Superstition Hills event was modeled as three independent subevents, each nucleating from a common location near the intersection of the two faults. This required rerupturing of one fault region on the time scale of several seconds. Slip was quite heterogeneous along strike, but fairly systematic as a function of depth. Substantial differences between the source process as observed from strong motion data and from teleseismic data were observed.

The 1989 Loma Prieta Earthquake began with a small (magnitude 4.5 to 5.0) precursor, which preceded the main part of the rupture by about 2 sec. Rupture was bilateral, with the overall radiation greater from the northwest portion of the fault. Separate inversions of the teleseismic data (periods 3-30 sec) and strong motion data (periods 1-5 sec) resulted in similar models, indicating a close correspondence of long- and short-period radiation. Forward predictions of the local strong motions from the teleseismic rupture model matched the distribution, duration and overall frequency content of the recordings, suggesting that constraints on strong motions can be made with teleseismic broadband recordings.

Short period and broadband teleseismic waveform data and three-component strong-motion records were analyzed to obtain the source rupture history of the 1991 Sierra Madre earthquake. The near-field, shear-wave displacement pulse from this event had a relatively short duration (about 1 sec) for the magnitude of the event, requiring a particularly high-average stress drop (175 bars). The ground-motion variations in the Los Angeles region were controlled predominantly by source directivity. Rupture was updip and southwestward, resulting in strong motions and heavier damage in regions to the southwest of the epicenter and near the updip fault projection.

The rupture process of the 1906 San Francisco earthquake was analyzed, using all high-quality, teleseismic recordings archived in the 1908 Carnegie Report of the State Earthquake Investigation Commission. The recordings are relatively simple considering the great rupture length in 1906, requiring that substantial portions of the fault, while having large slips, radiated little 5-25 sec energy. Two regions of the fault, one near the epicenter south of San Francisco, and one between Point Reyes and Fort Ross were responsible for generating the greater part of the energy observed on the teleseismic recordings. By comparison of our model for 1906 with modern, well-studied, large strike-slip events, we found similarities in rupture style with the relatively simple 1990 Philippines earthquake (M_s = 7.8), but contrasts with the complexity of the 1976 Guatemala earthquake (M_s = 7.5).

The rupture characteristics of these events when analyzed with previous finite-fault studies over the past decade indicate several common features. Variations in slip are more pronounced along strike than downdip. Vertical strike-slip faults show a systematic slip variation with depth, consistent with both shallow and deep zones with velocity-strengthening frictional resistance; nucleation is usually at the base of the seismogenic zone. Oblique and dip-slip events show much more depth variation in slip, indicative of thicker, more complex seismogenic zones associated with tectonic regimes involving crustal thickening or extension. The Superstition Hills, Loma Prieta and Sierra Madre strong-motion data sets all require short rise times, so only a small portion of the fault is slipping at a particular time, in agreement with the "self-healing" model described by Heaton [1990] and in conflict with long slip durations required by many crack-like models of dynamic rupture. With the exception of the Superstition Hills earthquake, seismic moments and slip distributions determined from the strong-motion data concur with moments and slips derived from geodetic and longer-period waveforms. This indicates that the higher-frequency data are sufficient for estimating the total slip, and therefore, the rupture durations inferred represent the entire coseismic slip duration. The agreement between longand short-period source models makes it possible to estimate ground motions for important historical events from source models determined using longer-period (5-15 sec), teleseismic body waves.

Thesis Files