Page 22 - Spring 2018

Volcanic Eruption Infrasound
reported in Universal Time \[UT\]). The seismometer (Figure 3a) records a near-continuous sequence of rhythmic volca- no-seismic disturbances called long-period events (LPs), a volcano-seismology term that refers to signals with domi- nant frequencies from 0.5 to 5 Hz. At the beginning of the plot before the steam explosion, the LPs occur in cyclic regu- larity, with a prominent peak in the interevent time distribu- tion (Matoza and Chouet, 2010). During the steam explosion that begins on March 9, 2005 (day 68), at 01:26:17 (Figure 3a, blue arrow), the seismic LPs merge closer together in time and transition into continuous eruption tremor as fluid (primarily steam and entrained ash) is ejected into the at- mosphere. After the steam explosion, the LP rate decelerates and returns to the background rate within about 90 minutes.
The 2004-2008 eruption of Mount St. Helens was closely monitored with multiple types of instrumentation. Thus, we have accurate knowledge of how the seismic recordings relate to the eruption process. However, without any other data, it might have been difficult to determine exactly when (and even if) an explosive eruption had occurred because similar seismic sequences as those shown in Figure 3a can be recorded even in the absence of an explosive eruption. However, the infrasound data (Figure 3b) reduce this am- biguity and provide clear evidence that an eruption has occurred (Matoza et al., 2007). Before and after the steam explosion, the infrasound data record only ambient noise. A clear signal begins coincident with the ejection of fluid into the atmosphere and continues to indicate the sustained jetting of fluid for a total of about 53 minutes. Thus, the in- frasound data delineate the timing and duration of the ex- plosive eruption process.
Long-Range Atmospheric
Propagation of Volcanic Infrasound Low-frequency sound from volcanoes has long been known to propagate great distances. The powerful 1883 Kraka- toa (located in present-day Indonesia) eruption produced acoustic-gravity waves that circumnavigated the globe mul- tiple times and were recorded on weather barometers (Ga- brielson, 2010). Long-range infrasound propagation occurs for two main reasons. First, absorption is relatively low at these frequencies, about 5 × 10−5 dB/km at 1 Hz compared with about 2 dB/km at 500 Hz (Sutherland and Bass, 2004). Second, strong vertical gradients in the temperature and horizontal winds create infrasonic waveguides in the tropo- sphere (approximately 0 to 15 km altitude), stratosphere (ap- proximately 15 to 60 km), and thermosphere (approximately
85 to 120 km). Substantial spatiotemporal atmospheric vari- ability, particularly changes in wind speed and direction, create variability and complexity in infrasound propagation. For example, numerous studies have found that the seasonal stratospheric winds (typically blowing to the east in the win- ter hemisphere and to the west in the summer hemisphere) largely control global infrasound propagation and detection (e.g., Le Pichon et al., 2009). Moreover, there has recently been a focus on creating better, more accurate, and seamless atmospheric models from the ground through the thermo- sphere to more accurately model infrasound propagation at long ranges (Blanc et al., 2017). Similarly, because long- range infrasound propagation is sensitive to wind and tem- perature changes, certain volcanoes that radiate infrasound near continuously have shown promise as a tool to passively detect changes in atmospheric structure that would other- wise be unobservable (e.g., Assink et al., 2013).
Remote Detection of
Volcanic Infrasound
The data shown in Figure 3 were collected as part of a project to evaluate the potential of remote infrasound arrays to pro- vide useful information about explosive eruptions (Garces et al., 2008). In addition to the local data shown from Mount St. Helens, an identical infrasound array was deployed approxi- mately 250 km to the east, at a location where infrasound ducted in the stratosphere can be expected to return to the ground during typical winter conditions (Matoza et al., 2007). Similar waveforms were recorded at the near and distant ar- rays, indicating that remote infrasound observations can rep- resent an accurate record of the source. The utility of long- range infrasound signals has been confirmed by numerous observations of eruption signals on remote infrasound arrays.
Larger and more energetic eruptions produce more power- ful infrasound signals that can be recorded farther from the source. The June 2009 eruption of Sarychev Peak volcano on a remote, uninhabited island in the Kurile Islands (a vol- canic island chain stretching between Japan and the Kam- chatka Peninsula of Russia) provides an illustrative example (Matoza et al., 2011; Figure 4). Due to the remote location, local ground-based geophysical observations were nonexis- tent and the eruption did not register on any remote seismic stations (e.g., seismic stations at distances of 352 km, 512 km, and 800 km). The first indication that an eruption had occurred was in satellite data acquired on June 11, 2009, that showed a thermal anomaly and weak ash emissions. Subse- quently, at 22:16 on June 12, 2009, spectacular photographs
20 | Acoustics Today | Spring 2018