Publication Date

5-2013

Date of Final Oral Examination (Defense)

3-8-2013

Type of Culminating Activity

Thesis

Degree Title

Master of Science in Geophysics

Department

Geosciences

Major Advisor

Paul Michaels, Ph.D.

Advisor

Matt Haney, Ph.D.

Advisor

Craig White, Ph.D.

Abstract

Earthquake swarms at volcanoes are common indicators of unrest and can be used to predict eruptions. However, not all earthquake swarms lead to an eruption but may die off instead. Variabilities in characteristics of swarms can lead to false predictions of an eruption. During the 2009 eruption of Redoubt Volcano in Alaska, there were five earthquake swarms, three of which preceded explosive eruptions and two that did not. These data were used to explore the variable characteristics that may be diagnostic of whether or not an eruption is imminent.

Data were recorded by the Alaska Volcano Observatory throughout the eruption. Band-pass filtering removed unwanted frequencies outside the long-period earthquake range of about 0.5-5.0 Hz. The onset of long-period earthquakes were cataloged and used to find features that varied between swarms. Duration times of individual events were calculated using the Arias Intensity. The power spectrum of the autocorrelation was used to determine central frequencies and shape factor values for each swarm. Earthquake swarms that preceded eruptions had short duration times. There was a small correlation between central frequency and shape factor values and eruption outcome and no correlation with time between earthquake swarm events.

Download

Share

COinS