Tales of intrusion and eruption: bringing magmatic timescales to eruption monitoring

Understanding sub-surface magmatic processes is key for assessing volcanic eruption timescales, eruption styles and their associated hazards. Geophysical and geochemical techniques are commonly used to monitor volcanoes to better understand the magmatic processes occurring at depth. However, although geophysical techniques can be implemented in near-real time, there is currently no geochemical or petrological technique that can be correlated with the geophysical datasets on this timescale. Diffusion modelling is a petrological technique used to model timescales of magmatic processes using the composition of zoned minerals. However, in its current form, it is not suitable as an eruption monitoring tool due to the lengthy processing time required. In this thesis, I present new diffusion modelling methodology for Fe-Mg diffusion in olivine that can be implemented for use as an eruption monitoring tool in near-real time. To refine and streamline the processing methodologies I analysed olivine-rich samples from Piton de la Fournaise, La Réunion, Mauna Loa, Hawaii and Vatnaöldur, Iceland. Using traditional processing methods, I applied suitable model parameters (e.g. temperature) and geometrical corrections (e.g. those for anisotropy) to a lava flow sample from Piton de la Fournaise and identified a simple, single crystal population. I used field, textural, compositional and timescale data (using the same traditional methods) to identify at least two crystal populations within two different samples from Mauna Loa (Hapaimamo and Moinui). Each of these populations, that vary in complexity, were considered to assess how each part of the processing workflow could be streamlined and I considered how to apply the necessary geometrical corrections at crystal population level rather than to individual crystal traverses. In doing so, I created a rapid processing workflow for Fe-Mg diffusion in olivine. I stress tested this new streamlined methodology using tephra samples from Vatnaöldur. I processed the samples under simulated eruption conditions to quantify how rapid the new workflow could be. It took ~26 hours from initial sample preparation to timescale interpretation; this is significantly faster than traditional methods. I have discussed the potentials and pitfalls of diffusion modelling as a monitoring tool so the new rapid methods developed in this thesis can be implemented during volcanic eruptions.