Volcanic Infrasound Research Articles

Investigating 10 Yr of Volcanoacoustic Activity at Tungurahua Volcano, Ecuador, Aided by Machine Learning

Abstract Here, we present the results of applying diverse data processing and machine learning tools to investigate a very large dataset obtained from single station infrasonic recordings from the last 10 yr of the most recent period of explosive activity at Tungurahua volcano, Ecuador. To increase the quality and quantity of information extracted from the large data set and enhance pattern recognition, we combined traditional techniques with more recent ones. We divided the investigation into sequential steps: detection, discrimination, cleaning, and clustering. For the detection step, we tested the classical short-term average/long-term average algorithm and an algorithm specific for explosions detection called “Volcanic INfrasound Explosions Detector Algorithm (VINEDA)” and detected 118,516 events. To clean the detected signals from potential false positives, we used supervised classification that reduced the events to 75,483, and a catalog cleaning procedure using shallow learners including support vector machines, random forests, and a single layer neural network, trained using data from a manual catalog, to a final number of 36,359 events. This led to a sixfold increase in detected explosions compared to the manual catalog. Then, we applied hierarchical clustering to a well-studied time window of activity using two independent difference metrics: dynamic time warping and waveform cross correlation and showed the insights and drawbacks from this approach. We showed that the different techniques were able to reveal repeating and striving events between selected different eruptive phases and associated them to possible changes in eruptive dynamics. Finally, to analyze the whole dataset at once we used a convolutional autoencoder network and obtained similar results to the classical clustering in a fraction of the time. We identified different families of explosions that appeared, sometimes intermittently, and revealed various potentially competing eruptive processes during the whole time period.

Rapid Location of Remote Volcanic Infrasound Using 3D Ray Tracing and Empirical Climatologies: Application to the 2011 Cordón Caulle and 2015 Calbuco Eruptions, Chile

AbstractInfrasound (<20 Hz) can propagate thousands of kilometers through the atmosphere, enabling global source location using networks of arrays such as the International Monitoring System infrasound network. However, atmospheric spatiotemporal variability poses a major challenge to locating infrasound sources. Strong horizontal cross‐winds deviate the observed infrasound arrival azimuths, producing source mislocations that can be on the order of hundreds of kilometers. We introduce a method that combines empirical climatologies (HWM14/NRLMSIS2.0) and 3D ray tracing (infraGA) to obtain first‐order, robust, and rapid estimates of the backazimuth deviations for source location procedures. For each trial source node and receiver path, day of the year and time, we perform an automatic iterative search for infrasound ground intercepts around the target station and obtain the corresponding backazimuth deviation. We test the method using IMS infrasound data from stations up to ∼5,000 km range for two similar explosive eruption case studies: 2011 Puyehue‐Cordón Caulle and 2015 Calbuco. We obtain a source mislocation reduction up to ∼84% (242–38.7 km) and ∼75% (366–93.1 km) for Puyehue‐Cordón Caulle and Calbuco eruptions, respectively. To evaluate the approach, we repeat the procedure using more realistic hybrid atmospheric descriptions; we obtain comparable results (up to ∼75% mislocation reduction for both eruptions). Potential applications of the approach include long‐range volcano monitoring in near‐real time by using pre‐computed look‐up tables or large‐scale, multi‐year reanalyses of infrasound waveform archives.

Long range infrasound monitoring of Yasur volcano

The atmospheric injection of gas and material produced by an explosive volcanic eruption determines a rapid compression of the atmosphere, which subsequently propagates as longitudinal elastic waves (sound). The size of the source, generally greater than tens of meters, and its duration, longer than a few seconds, result into an emitted signal that is particularly rich in low frequencies (f < 20 Hz), thus determining an efficient infrasound radiation. Thanks to the low spectral content and the reduced attenuation in the atmosphere, infrasound is capable of propagating for very large distances.In this study we show how the infrasonic monitoring at regional distances (> 100 km) is efficient in recording and characterizing volcanic events. For the purpose of our study, infrasound signal radiated from Yasur volcano (Tanna Island, Vanuatu) was studied for a period of twelve years (January 2008 – December 2019). Signals from Yasur were registered at a source-to-receiver distance of 400 km by the IS22 infrasound array, located in New Caledonia, part of the Comprehensive Nuclear-Test-Ban Treaty (CTBT) International Monitoring System (IMS). The predominantly explosive Strombolian activity of this volcano makes it a perfect subject to be studied by infrasound technology. Detections of volcanic infrasound are modulated according to the seasonal variation of stratospheric winds and corrected for attenuation accounting for real atmospheric specification between the source and the receiver to retrieve the pressure at the source. Next, they are used to evaluate long-term (yearly) and short term (hourly) variations of activity over the period of analysis. Results are compared with thermal anomalies recorded by the MODerate resolution Imaging Spectroradiometer (MODIS) installed on NASA's Terra and Aqua satellites.We show that even at regional distances (400 km) it is possible to follow the long term (yearly) fluctuations of ordinary explosive activity during periods of optimal propagation of infrasonic waves in the atmosphere. In addition, we show that the time resolution retrieved from the signal analysis allows to follow variations of volcanic activity at hourly time scale, thus representing a valuable source of information, in particular in areas where local geophysical observations are missing.

Global shockwaves of the Hunga Tonga-Hunga Ha’apai volcano eruption measured at ground stations

The eruption of the Tonga volcano created globally propagating spherical shockwaves in the atmosphere. Analyses are done to data from two southern U.S. stations of the author sampling at 3-21s intervals and 189 weather stations at 1-5min intervals. The shockwaves arrived from two routes in the atmosphere: the shortest spherical arc and the longer spherical arc through the antipole. In most stations, signals up to the 6th path of shockwaves were recorded as the waves traveled around the globe multiple times. The speed of shockwaves is estimated to be 309.5± 2.9m/s, consistent with the speed of sound at the top of the troposphere where a waveguide exists. Discussion is made on the post-shockwave ringing of 4-8min as higher amplitude oscillations above the level of pre-shockwaves background noise. A theoretical wave dispersion is derived which verifies that the spherical shockwave's phase speed is the same as the speed of sound.

Diverse Eruptive Activity Revealed by Acoustic and Electromagnetic Observations of the 14 July 2013 Intense Vulcanian Eruption of Tungurahua Volcano, Ecuador

AbstractDuring the powerful July 2013 eruption of Tungurahua volcano, Ecuador, we recorded exceptionally high amplitude, long‐period infrasound (1,600‐Pa peak‐to‐peak amplitude, 5.5‐s period) on sensors within 2 km of the vent alongside electromagnetic signals from volcanic lightning serendipitously captured as interference. This explosion was one of Tungurahua's most powerful vulcanian eruptions since recent activity began in 1999, and its acoustic wave is among the most powerful volcanic infrasound ever recorded anywhere. We use these data to quantify erupted volume from the main explosion and to classify postexplosive degassing into distinct emission styles. Additionally, we demonstrate a highly effective method of recording lightning‐related electromagnetic signals alongside infrasound. Detailed chronologies of powerful vulcanian eruptions are rare; this study demonstrates that diverse eruptive processes can occur in such eruptions and that near‐vent infrasound and electromagnetic data can elucidate them.

Volcanic eruption volume flux estimations from very long period infrasound signals

AbstractWe examine very long period infrasonic signals accompanying volcanic eruptions near active vents at Lokon‐Empung volcano in Indonesia, Aso, Kuchinoerabujima, and Kirishima volcanoes in Japan. The excitation of the very long period pulse is associated with an explosion, the emerging of an eruption column, and a pyroclastic density current. We model the excitation of the infrasound pulse, assuming a monopole source, to quantify the volume flux and cumulative volume of erupting material. The infrasound‐derived volume flux and cumulative volume can be less than half of the video‐derived results. A largely positive correlation can be seen between the infrasound‐derived volume flux and the maximum eruption column height. Therefore, our result suggests that the analysis of very long period volcanic infrasound pulses can be helpful in estimating the maximum eruption column height.

Using infrasound to constrain ash plume rise

Airborne volcanic ash advisories are currently based on analyses of satellite imagery with relatively low temporal resolution, and numerical simulations of atmospheric plume dispersion. These simulations rely on key input parameters such as the maximum height of eruption plumes and the mass eruption rate at the vent, which remain loosely constrained. In this study, we present a proof-of-concept workflow that incorporates the analysis of volcanic infrasound with numerical modelling of volcanic plume rise in a realistic atmosphere. We analyse acoustic infrasound records from two explosions during the 2009 eruption of Mt. Redoubt, USA, that produced plumes reaching heights of 12–14 km. We model the infrasonic radiation at the source under the assumptions of linear acoustic theory and calculate variations in mass ejection velocity at the vent. The estimated eruption velocities serve as the input for numerical models of plume rise. The encouraging results highlight the potential for infrasound measurements to be incorporated into numerical modelling of ash dispersion, and confirm their value for volcano monitoring operations.

The validation of SSWs forecasts in atmospheric global circulation models

While the influence of the troposphere on the stratosphere is well known, recent observational and modeling studies have demonstrated that the stratosphere has an impact on the troposphere as well. The dynamical coupling between stratosphere and troposphere is particularly strong during sudden stratospheric warming (SSW) events. The correct forecasting of the onset and duration of SSW events is therefore important and is a current challenge for weather forecasting centers. As there is a lack of observations in the upper stratosphere with good temporal and spatial coverage, additional techniques may be helpful to constrain SSWs. This is illustrated using volcanic infrasound measurements. The observations are compared with nowcast and forecast models up to 10 days. While a general agreement is found during the summer period, larger discrepancies are found during the equinox and major SSW of January 2013.

Detection of volcanic infrasound with a ring laser interferometer

Over the last 15 years, large horizontally mounted ring lasers have been used to study numerous geophysical phenomena. This paper provides examples of the sensitivity of large active ring laser interferometers to far field infrasound emissions from explosive volcanic eruptions. Volcanic infrasound is reported from representative eruptions of volcanoes Kelut (Kelud), Klyuchevskaya (Kliuchevskoi), Puyehua, Santa Maria, Sakurajima, and Tungurahua. The detected infrasound frequencies are in basic agreement with the far field air wave frequencies from the 1980 eruption of Mount St. Helens and the 1991 eruption of Mount Pinatubo. Cavities of large horizontally mounted active ring lasers contain two counter-propagating waves that in the presence Earth's rotation become traveling waves of slightly different frequencies. The Sagnac or beat frequency due to the difference in the traveling wave frequencies is modulated by geophysical phenomena, in this case volcanic infrasound. Signatures of the infrasound are found in the frequency modulated side bands.

Comparing near-regional and local measurements of infrasound from Mount Erebus, Antarctica: Implications for monitoring

Local (100 s of metres from vent) monitoring of volcanic infrasound is a common tool at volcanoes characterized by frequent low-magnitude eruptions, but it is generally not safe or practical to have sensors so close to the vent during more intense eruptions. To investigate the potential and limitations of monitoring at near-regional ranges (10 s of km) we studied infrasound detection and propagation at Mount Erebus, Antarctica. This site has both a good local monitoring network and an additional International Monitoring System infrasound array, IS55, located 25 km away. We compared data recorded at IS55 with a set of 117 known Strombolian events that were recorded with the local network in January 2006. 75% of these events were identified at IS55 by an analyst looking for a pressure transient coincident with an F-statistic detection, which identifies coherent infrasound signals. With the data from January 2006, we developed and calibrated an automated signal-detection algorithm based on threshold values of both the F-statistic and the correlation coefficient. Application of the algorithm across IS55 data for all of 2006 identified infrasonic signals expected to be Strombolian explosions, and proved reliable for indicating trends in eruption frequency. However, detectability at IS55 of known Strombolian events depended strongly on the local signal amplitude: 90% of events with local amplitudes > 25 Pa were identified at IS55, compared to only 26% of events with local amplitudes < 25 Pa. Event detection was also affected by considerable variation in amplitude decay rates between the local and near-regional sensors. Amplitudes recorded at IS55 varied between 3% and 180% of the amplitude expected assuming hemispherical spreading, indicating that amplitudes recorded at near-regional ranges to Erebus are unreliable indicators of event magnitude. Comparing amplitude decay rates with locally collected radiosonde data indicates a close relationship between recorded amplitude and lower atmosphere effective sound speed structure. At times of increased sound speed gradient, higher amplitude decay rates are observed, consistent with increased upward refraction of acoustic energy along the propagation path. This study indicates that whilst monitoring activity levels at near-regional ranges can be successful, variable amplitude decay rate means quantitative analysis of infrasound data for eruption intensity and magnitude is not advisable without the consideration of local atmospheric sound speed structure. • A study of volcanic infrasound recorded 25 km from Mount Erebus, Antarctica. • 75% of Strombolian events detected at the vent are observed at 25 km. • Large amplitude, temporally smooth variation in infrasound amplitude decay with range • Amplitude decay rate depends on lower tropospheric sound speed structure. • Automatic event detection was able to monitor activity levels at Erebus from 25 km.

Evaluation of wind and temperature profiles from ECMWF analysis on two hemispheres using volcanic infrasound

AbstractIn this paper, we evaluate vertical wind and temperature profiles that are produced by the European Centre for Medium‐Range Weather Forecasts (ECMWF) atmospheric analysis. The evaluation is carried out on both hemispheres: we make use of stratospheric infrasound arrivals from Mount Etna (37°N) and Mount Yasur (22°S). The near‐continuous, high activity of both volcanoes permits the study of stratospheric propagation along well‐defined paths with a time resolution ranging from hours to multiple years. Infrasound observables are compared to theoretical estimates obtained from acoustic propagation modeling using the ECMWF analysis. While a first‐order agreement is found for both hemispheres, we report on significant discrepancies around some of the equinox periods and other intervals during which the atmosphere is in a state of transition and dynamical oscillations of the atmosphere dominate over the general circulation. We present an inversion study in which we make use of measured trace velocity estimates to estimate first‐order effective sound speed model updates in a Bayesian framework. Deviations from the a priori models around the stratopause up to 10% (≈ 30 m s−1) are estimated. Such updates are in line with the results from comparisons between ECMWF analysis and observations from lidar and microwave Doppler spectroradiometer facilities that were colocated during the course of the 2012–2013 Atmospheric dynamics Research and InfraStructure in Europe (ARISE) measurement campaign.

Aeroacoustics of volcanic jets: Acoustic power estimation and jet velocity dependence

A fundamental goal of volcano acoustics is to relate observed infrasonic signals to the eruptive processes generating them. A link between acoustic power and volcanic gas exit velocity V was proposed by Woulff and McGetchin (1976) based upon the prevailing jet noise theory at the time (acoustic analogy theory). We reexamine this approach in the context of the current understanding of jet noise, using data from a laboratory jet, a full‐scale military jet aircraft, and a full‐scale rocket motor. Accurate estimates of require good spatial sampling of jet noise directionality; this is not usually possible in volcano acoustic field experiments. Typical volcano acoustic data better represent point measurements of acoustic intensity at a particular angle θ from the jet axis rather than . For pure air jet flows, velocity‐scaling laws currently proposed for acoustic intensity differ from those for acoustic power and are of the form , where c is the ambient sound speed and nθ varies nonlinearly from ∼5 to 10 as a function of temperature ratio and angle θ. Volcanic jet flows are more complex than the pure air laboratory case, which suggests that we do not currently know how the exponent nθ varies for a volcanic jet flow. This indicates that the formulation of Woulff and McGetchin (1976) can lead to large errors when inferring eruption parameters from acoustic data and thus requires modification. Quantitative integration of field, numerical, and laboratory studies within a modern aeroacoustics framework will lead to a more accurate relationship between volcanic infrasound and eruption parameters.

Passive acoustic remote sensing and anomalous infrasound propagation studies

In this talk, we will present recent work on various infrasound remote sensing studies. We will focus on bi-directional stratospheric ducting during a Sudden Stratospheric Warming (SSW) event and the associated infrasonic signature. We present infrasound data in which the described effect is captured with microbarom signals in the Mediterranean region. Microbarom source locations are modeled using operational ocean wave models. The modeling reveals a previously unidentified microbarom source region in the Eastern Mediterranean besides the more typical microbarom source region in the Atlantic Ocean. This work illustrates that the classic paradigm of a unidirectional stratospheric duct for infrasound propagation can be broken during a SSW event. Furthermore, we will present a case study in which the influence of atmospheric dynamics on infrasound propagation is studied. We make use of over 6 years of nearly continuous volcanic infrasound recordings from Mount Etna, Italy (37 N) that are available through the Atmospheric dynamics Research InfraStructure in Europe (ARISE) network. The infrasound observables are compared to theoretical estimates obtained from propagation modeling using existing European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric databases. While a good agreement is often found, we also report on significant discrepancies around the equinox period and during intervals during which anomalous detections occur during the winter.

Acoustic multipole source model for volcanic explosions and inversion for source parameters

SUMMARY Volcanic explosions are accompanied by strong acoustic pressure disturbances in the atmosphere. With a proper source model, these acoustic signals provide invaluable information about volcanic explosion dynamics. Far-field solutions to volcanic infrasound radiation have been derived above a rigid half-space boundary, and a simple inversion method was developed based on the half-space model. Acoustic monopole and dipole sources were estimated simultaneously from infrasound waveforms. Stability of the inversion procedure was assessed in terms of variances of source parameters, and the procedure was reliable with at least three stations around the infrasound source. Application of this method to infrasound observations recorded at Tungurahua volcano in Ecuador successfully produced a reasonable range of source parameters with acceptable variances. Observed strong directivity of infrasound radiation from explosions at Tungurahua are successfully explained by the directivity of a dipole source model. The resultant dipole axis, in turn, shows good agreement with the opening direction of the vent at Tungurahua, which is considered to be the origin of the dipole source. The method is general and can be utilized to study any monopole, dipole or combined sources generated by explosions.

Studying possibilities for the classification of infrasonic signals from different sources

Atmospheric infrasonic signals were classified on the basis of data obtained in the United States (University of Alaska, Fairbanks) and in the Antarctic region (Windless Bight) from 1980 to 1983. The data archive included five classes of signals from different sources: explosions, mountain associated waves, microbaroms, volcanic infrasound, and auroral infrasonic waves. This classification was based on the theory of testing statistical hypotheses. The possibilities of separating these classes were studied. It was shown that the signals (from the archive used) that are characteristic of explosions and volcanic activity can be rather easily separated from those characteristic of mountain associated waves, microbaroms, and auroral infrasonic waves.

A global study of volcanic infrasound characteristics and the potential for long-range monitoring

Volcanic infrasound from explosive eruptions is detectable at distances up to 1000 s of km and is often recorded on the International Monitoring System (IMS) infrasound network. Though infrasound produced by individual volcanoes has previously been studied, more general relationships between infrasound characteristics and eruption parameters remain unclear. Therefore, using a dataset comprising 110 events at 39 globally distributed volcanoes, we assess the ability of the IMS to monitor remote volcanoes and investigate whether volcanic infrasound exhibits consistent trends across a range of eruption styles and intensities. Acoustic pressure amplitude and frequency content of the volcanic infrasound were measured, and acoustic energy was also calculated. These were compared, along with the maximum distance of detection for each event, to the maximum plume height reported for the event, which was used as a proxy for event intensity. Of the events investigated 62% were detected by at least one infrasound station. The maximum distance at which an event was detected generally increased with the plume height, and infrasound produced during the eruption of Manam, PNG in January 2005, which produced a plume which rose over 20 km above the summit, was detected at a distance of 10,671 km. Total acoustic energy and acoustic pressure amplitude (corrected for geometrical spreading) increased with plume height, whilst the lowest frequency present in the volcanic infrasound decreased. Additionally, relationships seen between infrasound characteristics and eruption plume height indicate that analyses of this data alone could constrain vital estimates of plume heights for otherwise un-observed eruptions. Taken together these findings indicate that if made available for this purpose in the future, the IMS infrasound network could complement existing monitoring techniques by detecting and characterising eruptions at remote volcanoes which often have no local monitoring networks.

Long-range acoustic observations of the Eyjafjallajökull eruption, Iceland, April-May 2010

[1] The April–May 2010 summit eruption of Eyjafjallajokull, Iceland, was recorded by 14 atmospheric infrasound sensor arrays at ranges between 1,700 and 3,700 km, indicating that infrasound from modest-size eruptions can propagate for thousands of kilometers in atmospheric waveguides. Although variations in both atmospheric propagation conditions and background noise levels at the sensors generate fluctuations in signal-to-noise ratios and signal detectability, array processing techniques successfully discriminate between volcanic infrasound and ambient coherent and incoherent noise. The current global infrasound network is significantly more dense and sensitive than any previously operated network and signals from large volcanic explosions are routinely recorded. Because volcanic infrasound is generated during the explosive release of fluid into the atmosphere, it is a strong indicator that an eruption has occurred. Therefore, long-range infrasonic monitoring may aid volcanic explosion detection by complementing other monitoring technologies, especially in remote regions with sparse ground-based instrument networks.

Finite-difference time-domain modeling of transient infrasonic wavefields excited by volcanic explosions

[1] Numerical modeling of waveform diffractions along the rim of a volcano vent shows high correlation to observed explosion signals at Karymsky Volcano, Kamchatka, Russia. The finite difference modeling assumed a gaussian source time function and an axisymmetric geometry. A clear demonstration of the significant distortion of infrasonic wavefronts was caused by diffraction at the vent rim edge. Data collected at Karymsky in 1997 and 1998 were compared to synthetic waveforms and variations of vent geometry were determined via grid search. Karymsky exhibited a wide range of variation in infrasonic waveforms, well explained by the diffraction, and modeled as changing vent geometry. Rim diffraction of volcanic infrasound is shown to be significant and must be accounted for when interpreting source physics from acoustic observations.

Tracking near-surface atmospheric conditions using an infrasound network

Continuous volcanic infrasound signal was recorded on a three-microphone network at Kilauea in July 2008 and inverted for near-surface horizontal winds. Inter-station phase delays, determined by signal cross-correlation, vary by up to 4% and are attributable to variable atmospheric conditions. The results suggest two predominant weather regimes during the study period: (1) 6-9 m/s easterly trade winds and (2) lower-intensity 2-5 m/s mountain breezes from Mauna Loa. The results demonstrate the potential of using infrasound for tracking local averaged meteorological conditions, which has implications for modeling plume dispersal and quantifying gas flux.

Airwaves generated by an underwater explosion: Implications for volcanic infrasound

A shallow explosion in a fluid is one of the fundamental processes of airwave generation in volcanic eruptions. To better understand the mechanism of the airwave generation, underwater explosion experiments were conducted. Although the underwater explosions have been intensely studied over the last century, airwaves have received little attention. In this study, pressure waves were measured in air and under water, and the corresponding motion of the water surface was captured by a high‐speed video camera. Airwaves and associated styles of the surface motion show similarities determined by the scaled depth, defined as the depth divided by the cubic root of the explosion energy. The air waveforms are quite different from the pressure waves measured in the water and cannot be completely explained by linear transmission of pressure waves from water to air. The mechanism of airwave generation is a combination of wave transmission and dynamics of the interface boundary. The first mechanism directly reflects the explosion source, and the same source is also observed as the underwater pressure waves. On the other hand, the second is not necessarily visible in the underwater pressure waves but will provide information on the mechanical properties and behavior of the material above the explosion source. Each mechanism is analyzed in the experiments.