Earthquake... Volcano? Tales of seismically-triggered eruptions

How do tectonic earthquakes affect volcanoes? This research question forms the common thread in this thesis. Earthquakes have been reported to trigger a range of phenomena at volcanic centres: seismicity, ground deformation, gas emissions and, sometimes, magmatic eruptions. Such volcanic eruptions are then referred to as seismically-triggered eruptions. Today, anticipating seismically-triggered eruptions remains a challenge because we do not fully grasp the underlying physical mechanisms. This thesis is divided in three main chapters, each using a different approach to constrain the mechanisms involved in seismically-triggered eruptions. The thesis ends with implications for the fundamental goal of volcanology—improving our ability to forecast volcanic eruptions. In Chapter 2, I establish a theoretical framework to conceptually organize and under- stand the topic of seismically-triggered eruptions. I summarize the available evidence for seismically-triggered eruptions and review viable physical mechanisms to explain them. By considering the favourable conditions for each of the mechanisms, I derive a novel classification of volcanoes based on their susceptibility to being seismically-triggered. I find that every volcano has the potential to be seismically-triggered, though via different mechanisms. I outline three observable parameters to help assess the relevant mechanisms at any given volcano: (1) magma viscosity, (2) open- or closed-system degassing and (3) the presence or absence of an active hydrothermal system. In general, non- eruptive unrest is the most common seismically-triggered phenomenon, often associated with hydrothermal systems. Magmatic seismically-triggered eruptions only occur at vol- canoes that are already primed for eruption. I conclude this chapter by discussing how the characteristics of an earthquake also exert some control on the triggering mechanisms. This analysis shows that high peak ground velocities, low seismic waves frequencies and large static stress change amplitudes, all favour seismically-triggered eruptions. For Chapter 3, I designed a new experimental setup in order to explore the effect of pressure oscillations on vesiculation in rhyolite. Hydrous (0.11 ± 0.01 wt.% H2O) rhyolite samples were placed inside cylindrical ceramic crucibles and mechanically pressurized by applying force with a ceramic plunger. Two sets of experiments were performed. In the first set, pressure was kept constant at 177 kPa whereas in the second set, pressure followed a sinusoid centred about 177 kPa, with an amplitude of 71 kPa and a frequency of 0.1 Hz. In both cases, samples were supersaturated in water and expanded due to vesiculation. Bubbles preferentially formed on the outer margin of the samples, either via heterogneous nucleation or by diffusion into trapped interstitial air at the sample- crucible interface. For oscillating experiments, sample expansion was consistently lower, and the amount of bubbles at the sample margin was also reduced. I consider the effect of pressure oscillations on bubble nucleation, growth and coalescence, as well as gas loss by filtration through the ceramic crucibles. Pressure oscillations drive the samples in and out of water saturation, thereby reducing bubble nucleation and growth. While this analysis is consistent with my observations, more tests are needed to confirm this unexpected result. I finally discuss how the results obtained with this new experimental setup can be scaled to natural scenarios of seismically-triggered eruptions, where magmas usually contain more dissolved water (2-6 wt.%) and are typically stored at higher confining pressures (100-250 MPa). In Chapter 4, I study the case of the 1960 eruption of Cordón Caulle, in southern Chile. This eruption occurred less than two days after the Mw9.5 Great Chilean earth- quake and is generally considered to have been seismically-triggered. Cordón Caulle also produced two similar historical eruptions that were not seismically-triggered in 1921-1922 and 2011-2012. I collected and compared pumice lapilli samples from the 1921-1922, 1960 and 2011-2012 eruptions, in order to decipher any effect from the Mw9.5 earthquake. The 1960 samples are indistinguishable from the 2011-2012 samples, based on major element matrix glass composition, but the glass from 1921-1922 is slightly less evolved. I then apply rhyolite-MELTS geobarometry to calculate storage and extraction pressures. Storage pressures for all eruptions correspond to the shallow crust, at 90-112 MPa (4.0-5.0 km) for the 1921-1922 eruption, 123-143 MPa (5.4-6.3 km) for the 1960 eruption and 79-146 MPa (3.5-6.6 km) for the 2011-2012 eruption. Extraction pressures, i.e. the pressures at which melt segregated from its parental mush, are within the range 70-200 MPa (3.1-9.0 km) for all three eruptions, suggesting that melt was produced in the same, laterally extensive mush zone. These new geobarometry results are in agreement with current models of the magmatic system at Cordón Caulle from different data sources. Finally, I discuss whether the difference in storage pressure in 1960 could be related to the Great Chilean earthquake. The 1960 magma chamber was ∼1.3 km deeper than the 1921-1922 magma chamber. Thus, assuming similar initial volatile contents, the 1960 magma contained less exsolved volatiles during storage, which may have contributed to lower overpressures. The Mw9.5 earthquake unclamped the system, perhaps accelerating dyke opening, ascent and eruption of a magma likely already bound to erupt. In summary, assessing whether a tectonic earthquake increases the probability of eruption at a given volcano remains a major challenge for volcanologists. This thesis shows the breadth and complexity of seismically-triggered eruptions, and highlights the need for increased multidisciplinary approaches to the problem. An earthquake can modulate the time to eruption, for volcanic centres that are already in a critical state, i.e. ready to erupt on timescales from days to a few years. Volcanoes that are in a critical state should then be the monitoring priority, in the wake of a tectonic earthquake