Monotonic infrasound and Helmholtz resonance at Volcan Villarrica (Chile)

Transcription

1 GEOPHYSICAL RESEARCH LETTERS, VOL. 38,, doi: /2011gl046858, 2011 Monotonic infrasound and Helmholtz resonance at Volcan Villarrica (Chile) A. Goto 1 and J. B. Johnson 2 Received 3 February 2011; accepted 7 February 2011; published 18 March [1] Monotonic infrasound with stable peaked frequency of 0.77 Hz was recorded at Volcan Villarrica in January Similar monotonic infrasound had been previously reported at Villarrica (e.g., Ripepe et al. [2010]). Using joint infrasound and visual observations from a suspended camera we demonstrate that the likely source of infrasound is Helmholtz resonance produced from a cavity with volume 10 5 m 3 that separates the active convecting lava lake from an overhanging spatter roof. Spatter roof dimension (65 m diameter) and vent diameter (10 m) in the roof are constrained from video observations. Assuming a cylindrical cavity we infer a cavity height of 31 m that is corroborated by video records of spatter drips. The drips take as long as 2.2 s to fall from the roof into the lake, corresponding to a height of more than 24 m, which is in good agreement with the observed resonance frequency. Citation: Goto, A., and J. B. Johnson (2011), Monotonic infrasound and Helmholtz resonance at Volcan Villarrica (Chile), Geophys. Res. Lett., 38,, doi: /2011gl Introduction [2] Many erupting and open vent volcanoes produce prodigious sound in the infrasonic band [e.g., Johnson et al., 2004]. Observational studies have focused on the relation between frequency, intensity and duration of radiated infrasound with optical observations of physical source motions. For example, Oshima and Maekawa [2001] observed the occurrence of Merapi type pyroclastic flow on Unzen volcano. Ripepe and Marchetti [2002] and Rowe et al. [2000] characterized infrasound associated with strombolian style explosions at Stromboli and Erebus. Matoza et al. [2009] compared jet noise infrasound from large vulcanian and plinian eruptions on Mount St. Helens (USA) and Tungurahua (Ecuador). Many others have studied open vent volcanoes, which are often notable for intense infrasound production despite the concurrence of vigorous eruptive activity. For example, Kilauea s Halema`uma`u Vent [Fee et al., 2010], some of the vents at Stromboli [e.g., Johnson, 2004], Turrialba since 2008 (unpublished data), as well as Volcan Villarrica [Ripepe et al., 2010; Johnson et al., 2004] have produced continuous and intense infrasonic tremor during periods of passive degassing. [3] At Villarrica Johnson et al. [2004] first reported on the persistent and intense infrasound corresponding to a period in 1 Center for Northeast Asian Studies, Tohoku University, Sendai, Japan. 2 Department of Earth and Environmental Science, New Mexico Institute of Mining and Technology, Socorro, New Mexico, USA. Copyright 2011 by the American Geophysical Union /11/2011GL December, They observed episodic infrasonic N shaped onsets followed by a coda lasting for 5 10 sec repeating at intervals of 1.5 minutes. These waveforms hint at repeating bubble trains beginning with an initial pressurized bubble and followed by a sustained sequence of bursting bubble slugs. The spectra of this episodic infrasound were sharply peaked at 0.75 Hz, similar to the peaked spectra observed by our group in January 2010 and presented in this paper. Ripepe et al. [2010] also conducted additional infrasound surveys of Villarrica in November 2004 and March 2009 and recorded continuous (instead of episodic) infrasound that was sharply peaked at frequencies of 1.0 and 0.75 Hz respectively. (Although Ripepe et al. [2010] term this tremor as monochomatic we refer to it from hereon as monotonic considering that is qualifies sound rather than color). Ripepe et al. [2010] proposed a resonant phenomenon associated with a stable conduit and vent system, but rejected air column (i.e., organ pipe modes) resonance as the source of infrasound based on what they considered were unrealistic geometries, i.e., 220 and 370 m deep magma free surface for a 100 m wide conduit to generate 0.7 and 1 Hz spectral peaks. [4] Villarrica s intense infrasound has been observationally associated with vigorous roiling of a persistently active convecting lava lake surface from which small strombolian explosions are occasionally observed from the crater rim and during overflights. The basaltic to basaltic andesite lava lake is typically overhung by spatter roof grown through accretion and agglutination of the pyroclastic material on the inner crater walls [Palma et al., 2008]. The spatter roof is broken by a central circular vent through which the lava lake can be glimpsed from directly overhead. In January 2010 we suspended a downward directed video camera above the open vent and coincidentally recorded infrasound to better understand the genesis of Villarrica s intense sounds. With the camera we were able to observe degassing activity at the surface of the lava lake and noted a lack of correlation between infrasound pulse intensity and discrete Stromboliantype activity or other lava lake disruption. Based upon this finding and using camera observations to quantify the size and shape of the vent we demonstrate that Helmholtz resonance reasonably explains the continuous infrasound. 2. Observation [5] We deployed four stations consisting of infrasound microphones, audio microphone, three components broadband seismometers (Guralp CMG 40 T), and weather stations (Figure 1). The infrasound data presented here come from microphones incorporated AllSensors MEMS transducers, which have a flat frequency response between 0.02 Hz and Nyquist and a linear dynamic range from 1of5

2 Figure 1. Location map of Volcan Villarrica and distribution of observation sites. 5.5 mpa (RMS noise floor from 0.5 to 2 Hz) to +/ 250 Pa. Two stations were on the eastern and western rim of the crater (VSUM and VBEN), while the others were on the flanks of volcano 2.7 and 3.8 km NNW of the crater (VWIN and VSKI), respectively (Figure 1). Data were recorded at 1000 Hz at all stations continuously between Jan 21st and Jan 23rd 2010 and were synchronized with GPS timing. [6] Coincident video observations were made intermittently ( 2 hours per day during three days) with a high resolution ( pixels) 30 fps CASIO EX F1 camera. The camera was housed in a plastic enclosure with a clear glass window and the entire camera package was suspended from the midpoint of a 150 m long cable spanning the summit crater. Another cable was oriented perpendicular to this wire to provide horizontal stability of the suspended payload. Video was time synchronized to the nearest frame (1/30 s) by manually filming a handheld GPS clock display at the beginning of each 60 minute recording interval. [7] The position of the camera and size and distance to the top of the spatter roof were determined through placement of a GPS in the camera enclosure during filming. We filmed with the camera s widest angle field of view (7.3 mm focal length, corresponding to 36 mm on 35 mm film equivalent) and found that a 29 m camera translation corresponded to 570 pixels at the distance of the spatter roof. From this we deduced the spatial image resolution ( 5 cm/pixel) as well as horizontal extent of the cavity roof (65 m), and major and minor axis lengths of elliptic vent (12.5 m and 9.5 m). The camera s zoom setting also allowed an estimate of 65 m for the vertical distance between camera position (approximately equal to the lowest point on the Villarrica crater rim) and top of the spatter roof. 3. Infrasound Monotonic Tremor and Helmholtz Resonance [8] Infrasound recorded on the summit and flank stations shows a high degree of correlation (Figure 2). Monotonic and stable waveforms with a dominant frequency of 0.77 Hz (with Hz standard deviation) are prominent on each station during our observation period (Figure 3). During three days of recording infrasonic tremor varied in root mean squared amplitude from peak values of 20 Pa to 100 Pa (on the crater rim), but relative signal amplitude (recorded across all network stations) was consistent suggesting a minimal variation in atmospheric propagation effects. Network infrasound excess pressures demonstrated an expected 1/r fall off in pressure with distance. [9] Associated video imagery of the active lava lake showed persistent active convection and bubble bursting on Figure 2. Ten second sequence of infrasound recorded on summit (100 m from vent) and flank station (3.8 km from the vent). After time shift for expected infrasound flight paths signals correlate well. Amplitude fall off is predicted by 1/r geometric spreading. Corresponding activity shown in time synced video stills indicates a roiling lava lake surface, but no vigorous explosions. 2of5

3 When vent length is negligibly short, Helmholtz resonance frequency is given by [Fletcher and Rossing, 1998]: f ¼ c rffiffiffiffiffiffiffiffiffiffiffi r 2 ¼ c rffiffiffiffiffiffiffiffiffiffi r ð3þ 2 1:7Vr 2 1:7V In equations (1) (3) sound velocity is a function of temperature and gas composition in the cavity, given by p c ¼ ffiffiffiffiffiffiffiffiffi RT ð4þ Figure 3. Normalized amplitude spectra for Villarrica infrasound tremor (filtered between 0.25 Hz and 50 Hz) is sharply peaked at 0.77 Hz +/ 0.2 Hz during three days of observation in Each spectrum corresponds to one hour of tremor and possesses a peak frequency displayed in frequency histograms (inset). Secondary spectral peak at about 0.22 Hz is microbarom noise not associated with the volcano. the lava lake surface, which was uncorrelated with discrete events in the tremor record (Figure 2). Attempts to relate individual gas bubble bursts to discrete explosions were inconclusive. For this reason, and based upon the observed vent geometry, we propose that a cavity volume resonance could alternatively explain the monotonic infrasound. Previous studies by Fee et al. [2010] and Vergniolle and Caplan Auerbach [2004] had attributed volcano infrasound at Kilauea and Shishaldin to Helmholtz resonance, for which the classic resonance frequency f is given by: f ¼ c rffiffiffiffiffiffi S ð1þ 2 VL where c is sound velocity in the cavity, S is neck cross sectional area, L is neck length and V is cavity volume. In the case of Villarrica s vent L would corresponds to vent length penetrating the spatter roof. In practice, an extra air volume proportional to the neck radius moves together with the air above and below a short neck and effectively decreases the resonance frequency. This end effect may be added to the geometrical length of the neck and is calculated as 0.85 times the radius for a flanged end and 0.61 times radius at non flanged (pipe) end [e.g., Fletcher and Rossing, 1998]. By considering the vent as a circular flanged hole with radius r, the Helmholtz resonance frequency is then given by: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi f ¼ c r 2 ð2þ 2 VðLþ 1:7rÞ where g is specific heat ratio, R is gas constant (J/kg/K) and T is absolute temperature. Assuming Villarrica volcanic gas concentrations is 95 mol% H 2 O, 2.0 mol% CO 2, 2.1 mol% SO 2, and less than 1 mol% of other species [Shinohara and Witter, 2005] and using mixing theory for each gas species [Morrissey and Chouet, 2001], g and R are 1.31 and J/kg/K. We propose 200 C as a reasonable cavity gas temperature following Fee et al. [2010], who estimated the average temperature within the cavity in Halema`uma`u crater, Kilauea from FLIR imagery. Although gas concentrations and vent geometries differ at the two systems, we consider the open vent degassing and mafic chemistry of the two volcanoes to be similar. [10] Assuming a 200 C cavity temperature the sound velocity is calculated to be 514 m/s from equation (4). Using this value with f = 0.77 Hz and r = 5 m we obtain a cavity volume of m 3 from equation (3). For the same infrasound frequency cavity temperatures of 100 C and 300 C (456 m/s and 566 m/s) would yield respective volumes of m 3 and m 3. If the cavity is cylindrical and has the same diameter as that of the solidified roof (65 m), the uniform cavity height would be 31 m at 200 C for volcanic gases. [11] Cavity gas might also mix with ambient atmosphere whose g and R are 1.40 and J/kg/K resulting in 0.85 times lower velocity than Villarrica volcanic gas for the same temperatures. Although the atmosphere volcanic gas mixing ratio in cavity is unknown, equation (3) would predict a volume and height estimations as low as m 3 and 23 m, respectively, for a cavity filled with atmospheric air at 200 C. The actual cavity height should then most probably be somewhere between 23 m and 31 m. 4. Cavity Geometry Inferred From Video Image Analysis [12] We use fall times for molten lava blebs to estimate the cavity height and confirm the predicted Helmholtz cavity volume. In the video we are able to identify occasional lava dripping from the edge of vent and falling into the lava lake (Figure 4). Their apparent motion in the video image is evident due to the slightly oblique line of sight at the margins of the vent. Falling blebs occur primarily after lava spattering and they are identified from their straight motion, which ends at the surface of the lava lake. [13] Using lava fragment fall times we estimate cavity height. Forty four independent measurements of fall time range from 1.3 to 2.2 sec with a 1.76 s average. Ignoring drag forces of the atmosphere these correspond to free fall distances between 8 and 24 m with an average of 15.5 m. We speculate that the wide range in estimates could be due in part to dynamic levels of the lava surface (due to bubble 3of5

4 Figure 4. Lava fragments falling (arrowed) from vent edge, captured from video. Image frames are 0.4 s apart. A, B, and C refer to three different falling spatter blebs that are tracked in the image frames. slug arrival and surface disruption), but more likely, poor visibility due to volcanic fume may result in a bias that serves to decrease the estimated fall times. For these reasons we propose that the actual cavity depth could correspond to a fall of at least 2.2 s (or 24 m). The 24 m dimension is similar to the m estimate determined from the Helmholtz resonance model (equation (3)). 5. Discussion [14] The general agreement between Helmholtz inferred cavity height and observed cavity height, determined through spatter fall times, shows the plausibility of Helmholtz resonance for observed monotonic infrasound at Villarrica. Based upon repeated observation of a vent at Villarrica over the last ten years including our group, Palma et al. [2008], and Shinohara and Witter [2005], we suggest that this vent configuration is a common and stable feature of the Villarrica vent. We further speculate that the infrasound observed on December 2002, November 2004 and March 2009, which was all monotonic [Ripepe et al., 2010], might have corresponded to periods of time with similar vent geometry, which could also have resulted in Helmholtz resonance. It is notable that peak infrasound at Villarrica has been remarkably stable (between 0.7 to 0.8 Hz) during various campaigns conducted over many years. An exception was the observation interval in 2004 when the peak frequency was 1 Hz and the level of the lava free surface was considerably higher [Ripepe et al., 2010]. Intriguingly, Shinohara and Witter [2005] confirmed in November 2004 the existence of a spatter roof with 20 m diameter vent, about four times the area as in the present study. This was the same season as when the 1 Hz infrasound was recorded by Ripepe et al. [2010]. The higher frequency recorded in this season is in agreement with the prediction from equation (3). [15] Ripepe et al. [2010] had proposed an alternative model for the generally monotonic infrasound tremor at Villarrica attributing the sound to oscillatory convection in the conduit. According to this model, upwelling gas bubbles in the central part of the magma column are flanked by liquid moving downwards in convective cells close to the conduit wall, leading to gravity induced inhomogeneity. This model is based on two phase flow analogue experiments showing unimodal frequency pressure oscillations of the bubble column with periods becoming shorter for higher surficial gas velocities. Based upon analogue experiments they speculated that higher frequency infrasound at Villarrica is associated with enhanced degassing. [16] In theory vertical oscillations of the lava column, as well as Helmholtz resonance, can accelerate the atmosphere to explain the observed infrasound amplitudes. Cumulative mass of atmospheric air M(t) moved by magma surface oscillation or gas oscillations in a pipe can be estimated assuming a compact volumetric acoustic source [e.g., Johnson, 2003]: Mt ðþ¼ Z 0 2D Z 0 DP t þ D dt d v where D is the slant distance between infrasound source and observation site, DP is recorded excess pressure and v is average sound velocity along the propagation path. Considering infrasound observations at VSKI (1400 m high, 3.8 km from the vent) resulted in a sinusoidal wave with amplitudes as high as 0.28 Pa we infer that 234 m 3 of air must be displaced during a quarter cycle, or 0.32 s. This would correspond to a rapid lava surface uplift of 0.07 m averaged over the 65 m diameter lake. Alternatively, the same infrasonic pressure records could be produced by a 3 m throw of air mass within the 10 m diameter vent. [17] Oscillating air mass movement in the vent appear quite reasonable in light of recent video observations at Halema`uma`u, where puffs of gas were observed rising and falling at the vent opening with the same frequency as the recorded infrasound [Fee et al., 2010]. They interpreted the rhythmic exhalations to be a visual manifestation of Helmholtz resonance in the pit crater. Indeed, although the fundamental monotone of Kilauea monotonic tremor is a lower frequency than at Villarrica, the two volcanoes may be analogues considering the similarly high amounts of degassing that occurs from their actively convecting lava lakes. [18] Fee et al. [2010] attributed changing monotone frequency from 0.61 Hz to 0.33 Hz to gradual widening of the crater of the Halema`uma`u Vent due to incremental collapse occurring between April and August Ripepe et al. [2010] and our observations show no systematic change in frequency content of Villarrica infrasonic tremor, suggesting a largely non destructive vent geometry over the 2002 to 2010 interval. At Kilauea Fee et al. [2010] also reported on secondary tremor peaks (at 1 Hz) with frequency attributed to acoustic resonance. Although we also observe a subtle secondary peak in Villarrica tremor spectra at 1 Hz (see Figure 3) we see no obvious overtones. At this point we are reluctant to comment on the origin of this 1 Hz tone and differentiate between potential organ pipe resonance and oscillations of the magma free surface following Ripepe et al. [2010]. [19] In the present analysis we focused primarily on the frequency content of the Villarrica infrasound tremor, which is insensitive to the gas flux from the magma into the cavity, which excites the Helmholtz resonance through the vent constriction. Considering that gas flux influences cavity pressurization, which in turn controls the amplitude of the Helmholtz resonator, a future fluid dynamical study should attempt to relate gas flux rates to infrasound amplitudes using constraints on the vent geometry. Such a tool would be exceptionally useful considering that Villarrica degassing has been observed to vary significantly over time. Palma et al. [2008] showed the SO 2 flux ranging ton/day between Feb Mar 2000 and Jan Feb 2006 while Shinohara ð5þ 4of5

5 and Witter [2005] calculated ton/day SO 2 degassing. Based upon the descriptions by Palma et al. [2008] we consider the activity level during our January 2010 observation period to be relatively low. A calibrated tool to relate infrasound radiation to total gas flux would be invaluable. [20] Acknowledgments. We appreciate logistical assistance from Wener Keller (POVI) and OVDAS. This work was made possible through field assistance from Jake Anderson, Richard Sanderson, and Nick Varley. Discussions with Joe Wolfe and Yuji Hattori were helpful for our modeling. Funding was provided by NSF EAR grant , a grant from the National Geographic Society Expeditions Council, and Grants in Aid for Scientific Research (C) from the Japan Society for the Promotion of Science. References Fee, D., M. Garces, M. Patrick, B. Chouet, P. Dawson, and D. Swanson (2010), Infrasonic harmonic tremor and degassing bursts from Halema`uma`u Crater, Kilauea volcano, Hawaii, J. Geophys. Res., 115, B11316, doi: /2010jb Fletcher, N. H., and T. D. Rossing (1998), The Physics of Musical Instruments, 2nd ed., Springer, New York. Johnson, J. B. (2003), Generation and propagation of infrasonic airwaves from volcanic explosions, J. Volcanol. Geotherm. Res., 121, 1 14, doi: /s (02) Johnson, J. B. (2004), Source location variability and volcanic vent mapping with a small aperture infrasound array at Stromboli volcano, Italy, Bull. Volcanol., 67, 1 14, doi: /s Johnson, J. B., R. C. Aster, and P. R. Kyle (2004), Volcanic eruptions observed with infrasound, Geophys. Res. Lett., 31, L14604, doi: /2004gl Matoza, R. S., D. Fee, M. A. Garcés, J. M. Seiner, P. A. Ramón, and M. A. H. Hedlin (2009), Infrasonic jet noise from volcanic eruptions, Geophys. Res. Lett., 36, L08303, doi: /2008gl Morrissey, M. M., and B. A. Chouet (2001), Trends in long period seismicity related to magmatic fluid compositions, J. Volcanol. Geotherm. Res., 108, , doi: /s (00) Oshima, H., and T. Maekawa (2001), Excitation process of infrasonic waves associated with Merapi type pyroclastic flow as revealed by a new recording system, Geophys. Res. Lett., 28, , doi: /1999gl Palma, J. L., E. S. Calder, D. Basualto, S. Blake, and D. A. Rothery (2008), Correlations between SO 2 flux, seismicity, and outgassing activity at the open vent of Villarrica volcano, Chile, J. Geophys. Res., 113, B10201, doi: /2008jb Ripepe, M., and E. Marchetti (2002), Array tracking of infrasonic sources at Stromboli volcano, Geophys. Res. Lett., 29(22), 2076, doi: / 2002GL Ripepe, M., E. Marchetti, C. Bonadonna, A. J. L. Harris, L. Pioli, and G. Ulivieri (2010), Monochromatic infrasonic tremor driven by persistent degassing and convection at Villarrica volcano, Chile, Geophys. Res. Lett., 37, L15303, doi: /2010gl Rowe,C.A.,R.C.Aster,P.R.Kyle,R.R.Dibble,andJ.W.Schlue (2000), Seismic and acoustic observations at Mount Erebus volcano, Ross Island, Antarctica, , J. Volcanol. Geotherm. Res., 101, , doi: /s (00) Shinohara, H., and J. B. Witter (2005), Volcanic gases emitted during mild Strombolian activity of Villarrica volcano, Chile, Geophys. Res. Lett., 32, L20308, doi: /2005gl Vergniolle, S., and J. Caplan Auerbach (2004), Acoustic measurements of the 1999 basaltic eruption of Shishaldin volcano, Alaska 2. Precursor to the Subplinian phase, J. Volcanol. Geotherm. Res., 137, , doi: /j.jvolgeores A. Goto, Center for Northeast Asian Studies, Tohoku University, 41 Kawauchi, Aoba ku, Sendai, Miyagi , Japan.(ak goto@cneas. tohoku.ac.jp) J. B. Johnson, Department of Earth and Environmental Science, New Mexico Institute of Mining and Technology, 801 Leroy Place, Socorro, NM 87801, USA. 5of5