Allen CV
Seismo Lab
Earth & Planetary
UC Berkeley

MSc Thesis:
A new way to look at fault rupture:
Understanding and characterizing earthquake predominant period

Erik L. Olson
Master of Science
Geology and Geophysics
University of Wisconsin - Madison

Advisor: Richard M. Allen


Download the thesis: OlsonMScThesis2005.pdf

Earthquakes are a fascinating phenomenon that over the course of millions of years have changed the face of our planet. Earthquakes remain an enigma, defying scientists' best attempts to characterize their behavior. Until the mid-1900's we could not even fully characterize the size of an earthquake; only through assessing the destruction wrought by the event could we guess at the size of the force. In the late 1930's the magnitude scale arose as a new method of understanding the strength of an earthquake [Stein and Wysession, 2003]. Even after the magnitude scale was developed, another 30 years passed before magnitude was tied to earthquake moment, and therefore related to slip on the fault [Stein and Wysession, 2003].

Throughout history, even before the size of an earthquake was well understood, people attempted to predict these events. However, calculating the magnitude of an earthquake was not possible, even during the earthquake, let alone prior to rupture initiation. As rupture theory developed, and more earthquakes were studied, scientists began to postulate that the very nature of rupture prevented any fore-warning of an earthquake, and that an earthquake's magnitude could only be determined post-rupture.

Several other theories were proposed contradicting the non-deterministic model of earthquake rupture, however they often failed to explain earthquakes in more than one region, or fell into disfavor after additional research weakened supporting arguments. Then, a few years ago, new empirical evidence appeared which hinted at the possibility of a deterministic rupture process, i.e. evidence that the size of an earthquake may be determined prior to rupture termination. Working with a dataset from California, Allen and Kanamori [2003] found that the predominant period of an earthquake scales with the magnitude of the earthquake. The maximum predominant period of an earthquake, a value calculated from the frequency content of the P-wave, could be measured in 5 seconds, which is faster than a large earthquake takes to rupture. However, the study of Allen and Kanamori [2003] was limited to a small region, and only included 3 events greater than magnitude 6.

Lockman and Allen [In Review] took the analysis of maximum predominant period a step further and added earthquakes from Japan and the Pacific Northwest to the data set. These additional events remained consistent with the original observations that the maximum predominant period of an earthquake is related to the magnitude of the earthquake. Unfortunately the additional data set did not substantially increase the number of large magnitude events, and the source of predominant period remained an enigma.

The study presented in the following chapters was initiated to pick up the next step in the development of our understanding of predominant period. Despite several years of study, and the publication of many papers on the subject, no one has yet pinned down what causes the earthquake predominant period; nor what process occurs during earthquake rupture initiation that affects the event's final magnitude. This study represents the beginning of this characterization.

© Richard M Allen