Magma ascent rate and initial water concentration inferred from diffusive water loss from olivine-hosted melt inclusions

As the water concentration in magma decreases during magma ascent, olivine-hosted melt inclusions will reequilibrate with the host magma through hydrogen diffusion in olivine. Previous models showed that for a single spherical melt inclusion in the center of a spherical olivine, the rate of diffusive reequilibration depends on the partition coefficient and diffusivity of hydrogen in olivine, the radius of the melt inclusion, and the radius of the olivine. This process occurs within a few hours and must be considered when interpreting water concentration in olivine-hosted melt inclusions. A correlation is expected between water concentration and melt inclusion radius, because small melt inclusions are more rapidly reequilibrated than large ones when the other conditions are the same. This study investigates the effect of diffusive water loss in natural samples by exploring such a correlation between water concentration and melt inclusion radius, and shows that the correlation can be used to infer the initial water concentration and magma ascent rate. Raman and Fourier transform infrared spectroscopy measurements show that 31 melt inclusions (3.6–63.9 μm in radius) in six olivines from la Sommata, Vulcano Island, Aeolian Islands, have 0.93–5.28 wt% water, and the host glass has 0.17 wt% water. The water concentration in the melt inclusions shows larger variation than the data in previous studies (1.8–4.52 wt%). It correlates positively with the melt inclusion radius, but does not correlate with the major element concentrations in the melt inclusions, which is consistent with the hypothesis that the water concentration has been affected by diffusive water loss. In a simplified hypothetical scenario of magma ascent, the initial water concentration and magma ascent rate are inferred by numerical modeling of the diffusive water loss process. The melt inclusions in each olivine are assumed to have the same initial water concentration and magma ascent rate. The melt inclusions are assumed to be quenched after eruption (i.e., the diffusive water loss after eruption is not considered). The model results show that the melt inclusions initially had 3.9–5.9 wt% water and ascended at 0.002–0.021 MPa/s before eruption. The overall range of ascent rate is close to the lower limit of previous estimates on the ascent rate of basalts.