S21F-03:
Volcanic Tremor from Alaska to Hawai`i - What Can We Learn from Self-Organizing Maps?

Tuesday, 16 December 2014: 8:30 AM
Kathi Unglert and Mark Jellinek, University of British Columbia, Vancouver, BC, Canada
Abstract:
Volcanic tremor accompanies eruptions at basaltic ocean islands such as Kilauea, Hawai`i as well as at more silicic arc volcanoes such as Pavlof, Alaska. Tremor can be characterized by its source location, by statistically stationary frequency content, and by time-dependent features including frequency gliding. Tremor properties are governed by the underlying tremor source mechanisms, which, in turn, depend on distinct magmatic and tectonic controls such as magma composition, the volcanic plumbing system, and the regional crustal stress regime.
Key physical relationships among various tremor properties and their main controls are, however, poorly understood. We use a machine learning method called Self-Organizing Maps (SOMs) to identify characteristic tremor spectra and patterns related to when they occur during multiple eruptions in Hawai`i and in Alaska. Preliminary results show that the type and location of eruptive activity partly controls spectral properties and their temporal evolution. On Kilauea, broadband tremor (0.5-6 Hz) during eruptions associated with East Rift Zone intrusions in 2007 and 2011 differed from tremor during explosive activity at the summit in 2008. Narrowband tremor (0.5-3 Hz) on Pavlof during an eruption with a Volcanic Explosivity Index (VEI) of 2 in 2007 was slightly weaker than during a VEI 3 eruption in 2013, but did not show strong changes in character between the two eruptions.
SOMs are a powerful tool to analyze long timeseries such as continuous seismic data from multiple eruptions, which are becoming increasingly common in volcano seismology. We show that SOMs can reliably detect changes in spectral properties that constrain characteristic time scales for plausible tremor source mechanisms. By applying SOMs to tremor from different tectonic settings we identify chief similarities and differences, which ultimately help to determine the influence of various tectonic and magmatic controlling factors on volcanic tremor.