Abstract:
Volcanic earthquakes are a key indicator of unrest and eruptive risk. Analysis of seismic source signatures allows the inference of underlying processes and properties in the edifice. However, volcano seismic signals are also affected by scattering and intrinsic attenuation along the wave propagation path. These path effects contain useful information that may be inverted to help constrain the volcano structure, but without proper treatment can obscure information present in the source. Volcanic fluids include gases, water/hydrothermal brines and viscous magmas and have a wide range of viscosities. Fluid viscosity is known to represent a key control on the style of volcanic eruptions, but the link between viscosity and volcano seismicity is not well understood. Furthermore, fluids also influence the attenuation of seismic waves, but more data is needed to understand how different fluids control wave velocities and attenuation, and the effect this has on propagating waveforms. This is important for reliably interpreting volcano processes and inverting for edifice and fluid properties by analysis of volcano seismic signals and through volcano tomography studies. This thesis reports on the use of controlled laboratory experiments to calibrate physical changes in volcanic rocks and the fluids within to field seismic signatures, particularly focusing on the effect of fluid viscosity. The effect of variable fluids on wave generation and propagation is studied by three approaches: 1) An experimental comparison of the attributes of elastic waves (acoustic emissions) generated by geomechanical rock fracturing and the depressurisation of a range of fluids. 2) Experimental quantification of wave speeds and attenuation with changing pore fluid stiffness and viscosity, in rocks from White Island volcano (Whakaari), New Zealand. 3) Spectral Element Method, numerical simulations exploring how path effects influence the waveform and frequency characteristics of Volcano-Tectonic (VT) earthquakes at laboratory and field scales. By deforming rock samples until brittle failure and subsequently depressurising fluids of varying viscosity through the damage zone, we simulate processes associated with Volcano-Tectonic and Long-Period (LP) (and tremor) seismicity. Acoustic emissions (AEs) are recorded on a dense receiver array to accurately characterise the source, and are found to resemble well known families of volcano seismicity observed in the field, providing insight into some of the mechanisms that generate different types of seismicity in volcanic settings. Key seismological indicators are found to be dependent on the viscosity of the venting fluid. The number, initial rate and dominant spectral frequency of depressurisation AEs are found to inversely correlate with the fluid viscosity. A dependence of AE spectral content on fluid viscosity motivates the study of active-source wave propagation experiments. Ultrasonic P-wave attenuation and wave speeds are measured for intact and fractured lavas and ash tuffs from White Island, for variable fluid saturation, at in-situ pressure conditions. P-wave velocity and attenuation quality factor are found to correlate with pore fluid viscosity and bulk modulus and inversely correlate with sample porosity and fracturing. The sensitivity of wave speed and attenuation to changes in pore fluid is controlled by the rock porosity. We highlight the importance of seismic velocities in delineating volcanic structure and attenuation in monitoring time-lapse or spatial variations in fluid properties, particularly as a result of magmatic degassing. Attenuation quality factors and wave velocities determined during the previous ultrasonic experiments on volcanic rocks are used as input parameters in numerical modelling of wave propagation at laboratory and field scale. We find that wave propagation through low velocity, high attenuation media, representing gas- or partially gas-charged ash tuffs, can cause distortion of VT-type waveforms, with the depletion of high frequencies causing them to resemble LP signals at relatively short source-receiver distance. This highlights the importance of accounting for path effects before using volcano seismic signatures for the inversion of source processes and properties.