Temporal evolution of long-period seismicity at Etna Volcano, Italy, and its relationships with the 2004–2005 eruption

Abstract

Between December 2004 and August 2005, more than 50,000 long-period events (LP) accompanied by very-long period pulses (VLP) were recorded at Mt. Etna, encompassing the effusive eruption which started in September 2004. The observed activity can be explained by the injection of a gas slug formed within the magmatic column into an overlying cavity filled by either magmatic or hydrothermal fluids, thus triggering cavity resonance. Although a large number of LP events exhibit similar waveforms before the eruption, they change significantly during and after the eruption. We study the temporal evolution of the LP-VLP activity in terms of the source movement, change of the waveforms, temporal evolution of the dominant resonance frequencies and the source Q factor and changes in the polarization of the signal. The LP source locations before and after the eruption, respectively, do not move significantly, while a slight movement of the VLP source is found. The intensity of the LP events increases after the eruption as well as their dominant frequency and Q factor, while the polarization of the signals changes from predominantly transversal to pure radial motion. Although in previous studies a link between the observed LP activity and the eruption was not found, these observations suggest that such a link was established at the latter end of the eruptive sequence, most likely as a consequence of a reestablishment of the pressure balance in the plumbing system, after it was undermined due to the discharge of large amounts of resident magma during the eruption. Based on the polarization properties of the signal and geological setting of the area, a fluid-filled crack is proposed as the most likely source geometry. The spectral analysis based on the autoregressive-models (SOMPI) is applied to the signals in order to analyse the resonance frequencies and the source Q-factors. The results suggest water and basalt at low gas volume fraction as the most likely fluids involved in the source process. Using theoretical relations for the “slow waves” radiated from the fluid-filled crack, we also estimate the crack size for both fluids, respectively.