Explosive volcanic eruptions are propelled to the surface by the exsolution of vapour bubbles from magma due to decompression. A long-held view is that the amount of H₂O dissolved in the magma at depth controls the intensity of an explosive eruption. Growing evidence from studies reporting H₂O concentrations of melt inclusions (MIs) do not support this view. Instead, the rate at which magma ascends to the surface may play an important role in modulating the eruption style. Slow magma ascent allows the vapour bubbles to rise ahead of the magma, thereby diffusing the driving force for an explosive eruption, whereas for fast magma ascent, the bubbles remain essentially trapped within the magma, causing acceleration and the potential for an explosive eruption.
Chapter 1 presents a new modelling approach to constrain magma decompression rate based on the incomplete diffusive re-equilibration of H₂O in olivine-hosted melt inclusions. We apply this chronometer to two contrasting eruptions at Cerro Negro volcano in Nicaragua: the 1992 VEI 3 and 1995 VEI 2 eruptions. Both eruptions have the same basaltic composition (SiO₂ ∼ 50 wt%) and maximum volatile concentrations (H₂O ∼ 4.7 wt%). However, MIs from the less explosive 1995 eruption appear to have experienced more water loss compared to those from the 1992 eruption, which is consistent with slower magma ascent.
We present a parameterization of the numerical diffusion model in chapter 2, which significantly reduces the calculation time, facilitating the use of Monte Carlo simulations to evaluate uncertainties. We use this parameterization to create a regime diagram that can be used to guide when melt inclusions may be used as magma hygrometers and when they are better suited to act as magma speedometers. We develop diagnostic tools to recognize where and when water loss has occurred in a magma’s ascent history, and we outline quantitative tools that may be used to restore the primary and/or pre-eruptive water content.
We find that one of the largest sources of uncertainty in modelling diffusive re-equilibration of H₂O in MIs and olivines is the diffusivity of H+ in olivine. We present new experimental constraints on H+ diffusivity in olivines from Cerro Negro (1992 eruption) and Etna (3930 BP ‘Fall Stratified’ eruption) (chapters 1 and 3, respectively). Our results show that H+ diffusion is highly anisotropic with the diffusivity along the [100] direction more than an order of magnitude faster than along [010] or [001], implying a large role for the ‘proton-polaron’ diffusion mechanism, which shares this anisotropy. We also find that the lower forsterite (Fo ~ 80) olivines from Cerro Negro have significantly faster H+ diffusivity than higher forsterite (Fo ~ 90) olivines from Etna. The results for Etna agree well with other estimates on high forsterite olivines from San Carlos and Kilauea, suggesting that the Fe content of the olivine strongly affects the H+ diffusivity.
In chapter 4, we apply the methods from the first three chapters to an unusually explosive eruption of picritic magma at Etna, Sicily in 3930 BP (termed the ‘Fall Stratified’ eruption). MIs from this eruption show limited evidence for water loss and so cannot be modelled to determine decompression rate. Instead, we model H+ diffusion profiles within the olivine crystals themselves and determine rapid ascent rates of ~15 m/s. We perform rehomogenization experiments on the MIs to accurately assess their pre-eruptive CO₂ concentrations, and find nearly 1 wt.% CO₂. Solubility modelling indicates that these MIs must have been trapped at near Moho depths (24–30 km). The magma’s high CO₂ concentration and deep initial pressures may have been responsible for the magma’s rapid ascent, which ultimately led to its great eruption intensity.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/d8-zkzt-a130 |
| Date | January 2021 |
| Creators | Barth, Anna Claire |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0021 seconds