11:30 AM - 11:45 AM
[SVC41-22] An evaluation of small infrasound array observation at Stromboli
Keywords:Volcano, Infrasound, Observation
Volcanic infrasound brings information of surface activity of a volcano and its monitoring has been considered important to evaluate volcanic activity. Source location of infrasound is an essential information. There are two methods to estimate the source location. The one is an array observation and the other is a network observation. The former estimates the direction of wave propagation by closely distributed sensors. The latter estimates the source locations using the arrival time lags at all observation points. These observations, however, require a good amount of costs for equipments, installation, and maintenance. Here we propose a smaller scale array observation and the use of MUSIC algorithm for a high-resolution analysis. This study evaluates the performance of the method through a field experiment at Stromboli.
A smaller scale array, which consisted of 3 infrasonic sensors within the aperture of 20 m, was deployed near the summit of Stromboli volcano (Italy) in 25-27 June, 2017. For a few hours during the observation, a video camera was deployed in the neighborhood, of which data were available for reference. There were 4 active craters separate by about 10 degrees or more in the azimuth from the array.
At first, we analyzed test infrasound signals of which source craters were identified using the video data. Then, the parameters of the analyses and relative directions from the array to the craters were adjusted. It was found that sufficient resolution of azimuth was obtained by the array, though sound speed and elevation-angle were not resolved. Then, we analyzed unknown infrasound signals.
A theory of error estimation of MUSIC algorithm (Swindlehurst and Kailath, 1992) showed that the azimuths were obtained with a precision of less than 2 degrees. Actually, all of the azimuth estimates of strong signals from the individual craters were within ± 2 degrees around the means. On the other hand, results of weak signals were not so clearly pointing the craters.
We also conducted laboratory experiment with a hundredth scale of the field experiment. Comparing the results, we found frequency dependence in the azimuth estimations only with the field data. We tentatively infer the reason that the different frequency bands are from different sources and/or the lower-frequency bands are more contaminated by topographic reflection, scattering, or wind noise, which are not included in the laboratory data.
In conclusion, a three-element infrasound array of a 20-meter aperture has a good resolution at least to distinguish sound source azimuths separated by about 10 degrees, and is useful in observation of active volcanoes. The accuracy could be further improved by adequately improving methods of array installation and analysis.
A smaller scale array, which consisted of 3 infrasonic sensors within the aperture of 20 m, was deployed near the summit of Stromboli volcano (Italy) in 25-27 June, 2017. For a few hours during the observation, a video camera was deployed in the neighborhood, of which data were available for reference. There were 4 active craters separate by about 10 degrees or more in the azimuth from the array.
At first, we analyzed test infrasound signals of which source craters were identified using the video data. Then, the parameters of the analyses and relative directions from the array to the craters were adjusted. It was found that sufficient resolution of azimuth was obtained by the array, though sound speed and elevation-angle were not resolved. Then, we analyzed unknown infrasound signals.
A theory of error estimation of MUSIC algorithm (Swindlehurst and Kailath, 1992) showed that the azimuths were obtained with a precision of less than 2 degrees. Actually, all of the azimuth estimates of strong signals from the individual craters were within ± 2 degrees around the means. On the other hand, results of weak signals were not so clearly pointing the craters.
We also conducted laboratory experiment with a hundredth scale of the field experiment. Comparing the results, we found frequency dependence in the azimuth estimations only with the field data. We tentatively infer the reason that the different frequency bands are from different sources and/or the lower-frequency bands are more contaminated by topographic reflection, scattering, or wind noise, which are not included in the laboratory data.
In conclusion, a three-element infrasound array of a 20-meter aperture has a good resolution at least to distinguish sound source azimuths separated by about 10 degrees, and is useful in observation of active volcanoes. The accuracy could be further improved by adequately improving methods of array installation and analysis.