Wavelet Analysis of Seismic Signals at Mt. Etna

Wavelet methods are useful to study the time-varying properties of seismic signals at basaltic volcanoes, where unrest may lack obvious seismic precursors. Here we demonstrate the use of the DWPT, best basis algorithm, and Ljung-Box-Pierce portmanteau test to isolate frequency bands of strong signal in continuous volcanic tremor, denoise tremor data, and determine polarization and centroid locations for volcanic tremor, using methods described in Jones et al. (2005) and Jones et al. (2006). We present results from sample data recorded by 10 stations at Mt. Etna, Italy, during 2002 and 2005. Our findings suggest that the tremor has a small source region during December 2005, when the centroid of the seismic energy distribution varied by as little as 200 meters. Tremor is typically clustered at 2000-2800m a.s.l. and is centered roughly underneath the summit, in general agreement with results from 2004 data by Langer et al. (2005) and di Grazia et al. (2006). However, although frequency bands above 3 Hz contain less energy, we find distinguishable signals at higher frequencies, whose centroid locations suggest a similar source region.