Optimal identification of volcanic noise source using integral methods

Abstract

In this PhD thesis a critical approach in data analysis and a new method for acoustic inversion of volcanic infrasound are presented. An accurate review of equations relating the infrasonic pressure to gas velocity, based both on theoretical considerations and data analysis, demonstrates that the actual cases of applicability of such equations are rather sparse. All compact source models are defined in an infinite free-space, but the acoustic source during a strombolian eruption can be located at an unknown depth inside volcanic conduit: that makes the use of equations for free-space propagation no more valid. Furthermore, the presence of many poor-constrained parameters and the scattering with the terrain give rise to errors that prevent the extraction of dynamical quantities from data by using simple expressions, making these methods almost ineffective. In order to extract some informa- tion about the acoustic source directly from experimental data, an inversion problem was set up using numerical simulations, connecting the pressure recorded at microphones to the normal derivative of pressure at the source. By means of a Boundary Element Method, the transfer matrix containing the effect of acoustic propagation and terrain scattering from source panels to the microphones was extracted. Thanks to a simplified geometry of Stromboli volcano (Italy), source conditions were obtained from the inverse problem using two different methods: an analytical system, and an Adaptive Weighted-Sum optimization. The comparison of results obtained by the two methods makes use confident about the validity of the optimization. A validation of the methodology was performed by using a Finite Element Method (FEM), radiating in time domain the source conditions obtained by the inversion. In the case with no propagation inside the volcanic conduit, the signal reconstructed at microphone by the FEM accurately reproduces the actual pressure signal recorded at Stromboli, proving the accuracy of this methodology. Any refinements of the methodology could provide an accurate determination of the acoustic source conditions, giving more insight to the source process of explosive eruptions.