Publication Date

8-2015

Date of Final Oral Examination (Defense)

12-5-2014

Type of Culminating Activity

Thesis

Degree Title

Master of Science in Geophysics

Department

Geosciences

Major Advisor

Jeffrey B. Johnson, Ph.D.

Advisor

Hans-Peter Marshall, Ph.D.

Advisor

Lee M. Liberty, Ph.D.

Advisor

Brittany D. Brand, Ph.D.

Abstract

Active volcanic processes produce large amounts of acoustic energy within the infrasound band (0-20 Hz). Infrasound-sensitive microphones are often installed in addition to other forms of volcano monitoring equipment to increase the ability to remotely detect volcanic activity. In this study, an array of microphones was deployed without any additional sensor types for 36 hours at Santiaguito, Guatemala, to test the detection capabilities of a standalone microphone array. Array processing was applied to the recorded data, through frequency-domain beamforming and calculating a Fisher statistic (F). A changing F-threshold value was applied to differentiate between desired detections, or signal, and acoustic energy not originating from desired sources, or noise. Through determination of signal backazimuth, and knowing the azimuthal ranges of expected events, detections were categorized into three potential sources: volcanic explosion, volcanic rock fall, or non-volcanic rock fall.

After characterizing a signal as one of the three event types, determination of relative occurrence of each event type showed that volcanic rock falls were the most common event through the deployment. Progression of signal backazimuth through time indicated movement of rock fall events. Explosion events were demonstrated to be non-moving as expected. Calculation of a spectral median frequency for these events, and comparison to signal backazimuth validates results of previous studies, that rock fall infrasound is characterized by higher frequencies than explosions. This method was shown to be effective for remotely categorizing activity at a volcano during time periods of low wind.

Share

COinS