I. Rupture Properties of Large Subduction Earthquakes. II. Broadband Upper Mantle Structure of Western North America

Citation

Melbourne, Timothy Ian (1999) I. Rupture Properties of Large Subduction Earthquakes. II. Broadband Upper Mantle Structure of Western North America. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/87nd-p040. https://resolver.caltech.edu/CaltechTHESIS:09122016-121754817

Abstract

This thesis contains two studies, one of which employs geodetic data bearing on large subduction earthquakes to infer complexity of rupture duration, and the other is a high frequency seismological study of the upper mantle discontinuity structure under western North America and the East Pacific Rise. In the first part, we present Global Positioning System and tide gauge data which record the co-seismic deformation which accompanied the 1995 M_w8.0 Jalisco event offshore central Mexico, the 1994 M_w7.5 Sanriku event offshore Northern Honshu, Japan, and the 1995 M_w8.1 Antofagasta earthquake offshore Northern Chile. In two of the three cases we find that the mainshocks were followed by significant amounts of rapid, post-seismic deformation which is best and most easily explained by continued slip near the co-seismic rupture patch. In Jalisco, we find that the post-seismic deformation which occurred during the two weeks following the mainshock amounted to as much 70% of the co-seismic deformation, from which we estimate an additional moment release of 40%, while in the Sanriku event an additional 30% moment release followed in the 10 days after the mainshock. Because of the favorable geometry of the Jalisco network, we infer that the post-seismic faulting occurred predominantly down-dip of the co-seismic rupture plane. This is the first documented case of rapid slip migration following a large earthquake, and is pertinent to earthquake prediction based on precursory deformation. Following the Antofagasta mainshock there was no rapid post-seismic displacement within the resolution of the GPS measurements, which equals roughly 1% of the co-seismic displacement. As the three GPS data sets represent the best observations of large subduction earthquakes to date and two of them show significant amounts of aseismic energy release, they strongly suggest silent faulting may be common in certain types of subduction zones. This, in turn, bears on estimates of global moment release, seismic coupling, and our understanding of the natural hazards associated with convergent margins.

The second part of this dissertation utilizes high frequency body waves to infer the upper mantle structure of western North America and the East Pacific Rise. An uncharacteristically large M_w5.9 earthquake located in Western Texas provided a vivid topside reflection off the 410 Km velocity discontinuity ("410"), which we model to infer the fine details of this structure. We find that, contrary to conventional wisdom, the 410 is not sharp, and our results help reconcile seismic observations of 410 structure with laboratory predictions. By analyzing differences between our structure and seismic 410 structure estimates under the nearby Gulf of California, we attempt to extract differences in temperature and mineralogy between subcontinental and suboceanic 410 structures.

Extending this analysis, we utilize teleseismic events from East Pacific Rise transform faults to model multiple S upper mantle triplications. We find that for raypaths traversing the rise crest the 1-D model TNA [Grand and Helmberger (1984)] derived for the western US accurately predicts differential SnS-S travel times and triplication waveform structure, implying that there is little velocity heterogeneity along the ridge crest along nearly its entire length. We find that for energy traversing paths increasingly away from the ridge axis there is no discernible change in the apparent depth of the 410 and 670 Km discontinuities. In the shallowest mantle (uppermost 75 Km), there is a strong lateral shear velocity gradient amounting to 3% over roughly 150 Km. The LID, nonexistent at the ridge crest, grows slowly in thickness beyond 150 Km from the axis. The compatible geodynamic model of these two results is that the East Pacific Rise is not fed from the local lower mantle, rather, upper mantle material must be transported laterally to supply the ridge axis spreading center, and the LID reflects the source region of the East Pacific Rise magma supply.

Thesis Files