Revisiting the source process for VLP signals observed at Satsuma-Iwo jima, Japan

Abstract:

In April 1997, characteristic broadband seismic signals accompanied with very-long-period (VLP) repetitious seismic pulses were observed at Satsuma-Iwojima, Japan. Ohminato (2006) conducted waveform analyses and showed that the waveforms are well explained by the sudden expansion of an inclined shallow crack. In order to explain the repetitious property of the signals, he proposed a model of sudden vaporization of superheated water filling the crack as a physical model of the sudden crack expansion. This model explains the source geometry and the repetitious property of the VLP signals. However, this model has a serious drawback. The crack surface is not smooth and the rough surface would have small cavities containing small bubbles which can easily grow with increasing temperature and break the state of superheated condition of the water. Thus, for water in a container with rough surface such as a crack in the rock, it is difficult to keep the superheated condition.

Recently, Namiki and Kagoshima (2014) conducted lab experiments using syrup as a magma analogue. The experiments show that the expansion of one void leads to expansion of another void above it by the rupture of film separating these two voids. The void expansions propagate upward and finally cause intermittent outgassing when they reach the magma surface. Suppose that the upward propagation of void expansions occurs in the conduit of Satsuma-Iwojima and that the top of the conduit is connected to the shallow inclined crack. When the void expansions propagate upward and finally reach the top of the conduit, the gas from the void would be injected into the crack and the crack expansion would excite VLP signals. Every time a large void reaches the top of the conduit, the crack would repeat expansion. The internal pressure increase of the crack occurs simultaneously with the void pressure decrease. Why is it that only the VLP signals corresponding to the expanding source are observed? The time constant at which the void pressure decrease is transferred to the surrounding conduit wall is determined by the viscoelastic property of bubbly magma. If the time constant is large enough, it is possible that only the pressure increase in the crack contributes to the observed seismic signals and the contribution of the pressure decrease in the void is small enough not to be observed.