Tornillos and Explosions

We have been analysing tornillos, a particular type of seismic signal from Galeras, to learn about the movement of fluids in the volcanic system. In 1992 and 1993, tornillos occurred at Galeras prior to explosions (Narváez et al, 1997), so they have been considered a precursory signal. They are distinct seismic events with identifiable onsets and relatively long, gradually decaying event tails (codas) and their name comes from the resemblence of their shape on the seismic record to a screw (Figure 13.33). Ninety tornillos recorded at Galeras Volcano, Colombia, from December 1999, to December 2002, were not immediately associated with explosions. Then, no tornillos were observed from December 2002, until well after the beginning of eruptive activity, in September 2004 (Figure 13.34. As a class, the tornillos are complex, each having from one to 15 spectral peaks between 1 and 40 Hz. The peaks for frequencies which extend into the coda are extremely narrow, while those present only during the initial excitation are relatively broad. The frequency of the lowest spectral peak present in any of the tornillos recorded between December 1999, and December 2002, was higher than 1.6 Hz. In contrast, tornillos occuring after the eruption began included frequencies between 1.0 and 1.3 Hz.

Figure 13.34: Occurrence of tornillos and their frequencies (dots). Triangles mark reported explosions.
\begin{figure}\begin{center}
\epsfig{file=hell05_1_2.ps, width=8cm}\end{center}\end{figure}

Most analysis of tornillos has concentrated on the coda which gives them their name. However, each tornillo waveform is actually made up of three parts (Figure 13.35). In the high quality broadband recordings, the tornillos start with a very small, but clearly recognizable P-onset. This is followed about 0.25 s later by the arrival of wave energy on the horizontal traces, probably S-waves. This arrival becomes a complex wavepacket lasting between two and three seconds which then transitions into the coda. Preliminary comparisons of these wavepackets for many different tornillos suggest two things: that they can be classified into families, and that these families are only indirectly related to the exact frequency or family of frequencies present in the long-lasting coda into which they segue.

Figure 13.35: Displacment seismograms of the beginning of a tornillo. The first onset has polarization consistent with a P-wave, while the second onset is likely to be an S-wave. It is followed by a 2-3 s transition before the characteristic coda develops.
\begin{figure}\begin{center}
\epsfig{file=hell05_1_3.ps, width=8cm}\end{center}\end{figure}

We are currently investigating and quantifying the characteristics of both the onsets and the intermediate wavepackets more thoroughly, as they bear important information about the mechanism which triggers a tornillo.

For the spectral peaks below 5 Hz, the polarization both remains constant during an individual tornillo, and varies little from one tornillo to the next. This suggests that tornillos are all generated within a limited volume of the volcanic edifice. The variation in the polarization at higher frequencies should allow us to constrain the size of this volume (Hellweg, 2003)..

Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail: www@seismo.berkeley.edu
© 2005, The Regents of the University of California