• Exploring infrasound wavefields to characterize volcanic eruptions

      Iezzi, Alexandra M.; Fee, David; Tape, Carl; West, Michael; Izbekov, Pavel; Haney, Matthew (2020-08)
      Infrasound has become an increasingly popular way to monitor and characterize volcanic eruptions, especially when combined with multidisciplinary observations. Regardless of how close the infrasound instruments are to the eruption, the effects from propagation must be considered prior to characterizing and quantifying the source. In this dissertation, we focus on modeling the effects of the atmosphere and topography on the recorded infrasound waveforms in order to better interpret the acoustic source and its implications on the volcanic eruption as a whole. Alaska has 54 historically active volcanoes, one third of which have no local monitoring equipment. Therefore, remote sensing (including that of infrasound arrays) is relied upon for the detection, location, and characterization of volcanic eruptions. At long ranges, the wind and temperature structure of the atmosphere affects infrasound propagation, however, changes in these conditions are variable both in time and space. We apply an atmospheric reconstruction model to characterize the atmosphere and use infrasound propagation modeling techniques for a few recent eruptions in Alaska. We couple these atmospheric propagation results with array processing techniques to provide insight into detection capability and eruption dynamics for both transient and long-duration eruptions in Alaska. Furthermore, we explore the future implementation of this long-range infrasound propagation modeling as an additional monitoring tool for volcano observatories in real time. The quantication of volcanic emissions, including volume flow rate and erupted mass, is possible through acoustic waveform inversion techniques that account for the effects of propagation over topography. Previous volcanic studies have generally assumed a simple acoustic source (monopole), however, more complex source reconstructions can be estimated using a combination of monopole and dipole sources (multipole). We deployed an acoustic network around Yasur volcano, Vanuatu, which has eruptions every 1-4 minutes, including acoustic sensors along a tethered aerostat, allowing us to better constrain the acoustic source in three dimensions. We find that the monopole source is a good approximation when topography is accounted for, but that directionality cannot be fully discounted. Inversions for the dipole components produce estimates consistent with observed ballistic directionality, though these inversions are somewhat unstable given the station conguration. Future work to explore acoustic waveform inversion stability, uncertainty, and robustness should be performed in order to better estimate and quantify the explosion source. Volcanic explosions can produce large, ash-rich plumes that pose great hazard to aviation. We use a single co-located seismic and infrasound sensor pair to characterize 21 explosions at Mount Cleveland, Alaska over a four-year study period. While the seismic explosion signals were similar, the acoustic signals varied between explosions, with some explosions exhibiting single main compressional phase while other explosions had multiple compressions in a row. A notable observation is that the seismo-acoustic time lag varied between explosions, implying a change in the path between the source and receiver. We explore the influence of atmospheric effects, nonlinear propagation, and source depth within the conduit on this variable seismo-acoustic time lag. While changes in the atmospheric conditions can explain some of the observed variation, substantial residual time lags remain for many explosions. Additionally, nonlinear propagation does not result in a measurable difference for the acoustic onset. Therefore, using methods such as seismic particle motion analysis and cross correlation of waveforms between events, we conclude that varying source depth within the conduit likely plays a key role in the observed variation in the seismo-acoustic time lags at Mount Cleveland.