Inferring magma ascent timescales and reconstructing conduit processes in explosive rhyolitic eruptions using diffusive losses of hydrogen from melt inclusions

Abstract

We integrate field observations with geochemical data from quartz-hosted melt inclusions to develop a hydrogen diffusion-based method to shed light on timescales and conduit processes during the opening stages of two supereruptions. We focus on the initial explosive phases (first ~5% of the total magma volume erupted) from: (1) the episodic Oruanui eruption, Taupo NZ (530 km3, 25.4 ka) and (2) the continuous Bishop Tuff eruption, Long Valley (650 km3, 0.765 Ma). Quartz-hosted melt inclusions for both eruptions exhibit coherent trends for major and trace elements (e.g. Al2O3, SiO2, Nb, U, Rb). However, in the first three phases (of ten) in the Oruanui eruption there is significant variability in measured H2O values (2.6–5.8 wt%), particularly in phases 1 and 3, each of which reflects the opening stages of two respective inferred vent sources. Phase 3 inclusions also record greater scatter in Li concentrations. In contrast, melt inclusions from the first two Bishop Tuff fall layers yield a narrower range of H2O concentrations (4.0–5.8%, with 90% between 4.5 and 5.8 wt%). We interpret the scatter in measured H2O concentrations to represent changes during slow magma ascent or partial stalling within the conduit at lower pressures. Using a diffusion model, we estimate that 61% of the Oruanui melt inclusions experienced modification of original H2O concentrations due to slow ascent in the conduit system prior to eruptive quenching, with the longest timescales (>5 days) mostly in phase 3 inclusions. In contrast, only 33% of melt inclusions from the initial Bishop fall layer (F1) experienced diffusive losses from their starting concentrations, implying that most magma spent less than a day ascending from the storage region at the eruption onset. Similarities between the Oruanui and the previously studied Huckleberry Ridge Tuff include comparable extents and distribution of diffusive losses (seen as H and Li scatter), and field evidence for episodic activity. Our observations imply that in these two systems, the first-erupted magmas rose from reservoirs with low degrees of overpressure and that magma mobilization was controlled by external factors (e.g. tectonics) rather than by strong volatile over-pressures. In contrast the Bishop eruption may well have been initiated by overpressure, leading to continuous, escalating activity.