Acoustic signals at Shishaldin volcano
Part of my research at the Alaska Volcano Observatory focuses on the 1999 eruption of Shishaldin volcano. Shishaldin is a spectacular stratovolcano on Unimak island, the easternmost island in the Aleutian chain. It is among the most active volcanoes in Alaska, with literally dozens of historical eruptions.

Figure 1: Location of Shishaldin volcano and Unimak island. Black circles represent the locations of seismic stations operated by the Alaska Volcano Observatory. The fishing communities of False Pass and Cold Bay are marked with triangles. The pressure sensor is co-located with seismic station SSLN. Note that Shishaldin cannot be seen from False Pass because the view is blocked by Round Top and Isanotski volcanoes.

The 1999 eruption of Shishaldin is particularly interesting because it exhibited a range of eruptive styles. On April 18 1999, lava spattering was observed at the volcano's summit. Satellite imagery shows that there was a thermal anomaly (hot spot) at Shishaldin's summit for quite some time before this, so it is entirely possible, even likely, that this type of activity had been occurring for some time. On April 19, however, the volcano produced an explosive eruption that sent an ash plume to heights exceeding 14 km (a Subplinian eruption). Immediately following the explosive phase, however, the volcano returned to its relatively more passive ways and returned to Strombolian behavior, with magma bubbles bursting within the conduit.

My interest in the 1999 eruption has to do with data collected on a pressure sensor. This instrument is a type of microphone that listens to sounds produced by the volcano. (My affection for volcano acoustics stems from research that I did with Fred Duennebier at the University of Hawai`i, studying hydroacoustic signals from Lo`ihi and Kilauea volcanoes).

During the 1999 Shishaldin eruption, the pressure sensor recorded a number of interesting signals. The first is a strongly bandlimited "humming" signal, shown at the left. This signal was recorded for ~10 hours prior to the Subplinian eruption. Although the frequency never changed, the amplitude increased with time, as shown in the figure to the right. Work that I have done in collaboration with Sylvie Vergniolle of the Institut de Physique du Globe de Paris suggsts that this signal resulted from the collapse of athe shallow portion of a foam that had been building in the Shishaldin conduit. This work can be accessed here.

A second intriguing acoustic signal was a sequence of discrete, impulsive signals, as shown to the right. These signals are virtually identical to acoustic signals recorded at a number of other volcanoes, including Karymsky, Arenal, Erebus, Stromboli and Etna. At Erebus and Stromboli these events occur coincident with magma bubble bursts. An analysis of these events from the perspective of quantifying gas flux, estimating bubble size, etc.may be found here .

We recorded two episodes of these explosion signals. The first occurred immediately following the Subplinian eruption, and was composed of relatively small signals with very low (<1 Hz) frequencies (seen in the spectrogram to the left). The later episode, on April 22-23, saw extremely large explosions, with peak-to-peak amplitudes exceeding 100 Pa over 6 km from the vent (righthand spectrogram).

Perhaps the most interesting perspective is to look at the pressure sensor and seismic records in tandem. The figure below shows spectrograms of the pressure sensor (top) and one of the seismometers (bottom) for the several hour time period surrounding the Subplinian eruption.

An interesting relationship between the acoustic and seismic signals is evident in this figure. When the humming signal stops (time "A"), the tremor amplitude increases dramatically on seismic station ISNN. This is followed by a short time period (just before time B) where the acoustic sensor recorded near silence. And at time B, both instruments recorded a strong broadband signal that we believe represents the Subplinian eruption. Times D and E are the beginning of two episodes of explosions, as shown in an earlier figure.

One of the questions that Sylvie and I have been working on using these data has to do with the transition from Strombolian to Subplinian activity and back to Strombolian bubble bursts. We believe that the relationship between the humming signal and the Subplinian episode may elucidate the processes occurring within the conduit at this time. It is also notable how much more detail is evident in the acoustic record...in general, signals passing through the atmosphere are less affected by attenuation and filtering than those passing through the earth. So we get a clearer view of vent processes. We have several more papers in review on this topic which will be posted here if and when they are published.



Return to Jackie's home page