Geophysical signature of unrest episodes at active volcanic systems : insights into the hydrothermal system fingerprint.
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
The characterisation of unrest signals and eruption precursors is one of the main challenges in volcano research because of the complexity of volcanic systems (e.g., the interplay between magmatic and hydrothermal fluids). This thesis addresses the issue for White Island volcano (New Zealand), and presents a comprehensive analysis of geophysical changes associated with the recent unrest/eruptive episode (2011-2013). A modelling strategy was used to 1. Characterise the source of the magnetic and gravity changes during this unrest/eruptive episode, 2. describe the effect of an inclined fumarole on hydrothermal circulation and gravity changes, 3. assess whether volcanic tremor can be used for eruption forecasting at White Island. The observed magnetic changes were inverted for a dipole, and can be explained by temperature changes at shallow depth below the active crater. The lack of signi_cant gravity changes was then used to constrain the heat source responsible for the magnetic changes. The geophysical changes are consistent with a model involving an episode of increased degassing from a possible shallow magmatic intrusion. The effect of a period of increased degassing on hydrothermal circulation and gravity changes in the fumarole area was then investigated. Previous studies inferred an inclined conduit for the main fumarole at White Island (fumarole zero). I therefore investigated the effect of such an inclined conduit on hydrothermal circulation and gravity changes at steady state and associated with an unrest episode, using a numerical modelling approach (TOUGH2). The model was constrained using parameters consistent for fumarole zero (small conduit one order of magnitude more permeable than the surrounding medium). The effect of the fumarole inclination is to shift the hydrothermal plume and the gravity anomaly towards the injection area instead of the fumarole outlet. Such a model implies that regular microgravity measurements can inform on the location of the feeding source of the fumarole. Finally, I calibrated an algorithm implementing the material Failure Forecast Method to issue eruption forecasts from volcanic tremor at White Island. Volcanic tremor increases preceding four out of the five eruptions of August 2011-January 2014 period are well explained by a model where an eruption is a case of material failure due to magma pressurisation. These tremor increases were therefore likely precursory to the eruptions. The good fit between the model and data allowed the issue of reliable eruption forecasts so that four eruptions (out of the five eruptions of the episode) occurred during forecast eruption windows. The probability of having an eruption during a forecast eruption window is 0.21 for the whole period, 37 times higher than the probability of having an eruption on any day, demonstrating that eruption forecasting capabilities can be enhanced using our procedure. We conclude that, at White Island, magnetic and gravity measurements are valuable to characterise the unrest source, and that the evolution of volcanic tremor can be used for eruption forecasting. Magnetic measurements can help characterise unrest because of their sensitivity to temperature changes. Additional gravity measurements allow constraining the source of the magnetic changes, and they could also inform on location of the source of the fluid injection in the fumarole area. The evolution of volcanic tremor can be precursory to eruptions and allow an estimate of the timing of the eruptions onset. This study therefore brings insights into unrest sources and eruption precursors at White Island, while providing methods that could be applied at other volcanoes. It also highlights the importance of continuous measurements to constrain volcanic processes.