Infrasound Explosion and Coda Signal Investigated with Joint Analysis of Video at Mount Erebus, Antarctica

Document Type


Publication Date



Mount Erebus is a natural laboratory where lava lake strombolian activity can be directly observed from the crater rim, allowing for a unique opportunity to study infrasound generated during explosive bubble bursting. Erebus lava lake eruptions occur when pressurized bubble slugs distend the lava lake surface and then burst within a few tenths of seconds. Data from two infrasound stations and one video camera were used to study 13 explosions occurring in January of 2006. Infrasound recordings from the two stations at ~330 m from the source exhibit variable amplitudes (40–160 Pa), but similar shapes composed of high-amplitude ~1 s bipolar pulses followed by codas exhibiting 3 or 4 decaying oscillations and lasting ~3 s. Infrasound pulses and infrasound-inferred source volumes are compared to video imagery, which is used to quantify the expanding volumes of distending bubbles and subsequent erupted pyroclasts. Video-inferred volumes scale well with infrasound-inferred volumes during the first few tenths of a second of an eruption, assuming a monopole source. Afterwards, an inflection in infrasound waveform, corresponding to a more rapid increase in the pressure record, indicates the fragmentation of the bubble membrane. At this point, infrasound is still produced by expanding gases and pyroclasts, but infrasound-inferred volumes are smaller than the image processing suggests. This is due to the fact that the image processing tracks ballistics, which expand inertially and without exerting a significant volume change. After 0.5 to 1.5 s the volumetric expansion is no longer evident in the infrasound record, which is now comprised of a coda characterized by decaying ~1 Hz oscillations. Because the coda does not correspond to any motions evident in the video, we conclude that it represents an acoustic resonance of the void left in the lava lake. This resonance is modulated by echoes off crater topography giving rise to two peak frequencies that dominate the coda spectra and are different for the two infrasound recording sites. The two coda spectral peaks are well explained by site-dependent propagation effects. Deconvolving crater topography effects, which are different for the two stations, reveals a single frequency peak at 1.5 Hz and is attributed to resonance of the void left in the lava lake, which we model as an open cylindrical pipe or Helmholtz resonance.