Convective Melting and Water Behavior around Magmatic-Hydrothermal Transition: Numerical Modeling with Application to Krafla Volcano, Iceland

Abstract

Abstract Water is an essential component of rhyolitic magmas and nearly universally, silicic magmatism in the upper crust includes a transition from magma to water-saturated roof rocks. We have numerically simulated the effect of the addition of geothermal fluids to an intruded rhyolitic sill from the hydrothermal system contained within the porous felsitic roof rocks. Water uptake in the melt proceeds via its thermodynamically estimated saturation with a partial melt, corresponding to the fugacity of hydrothermal water in the melt-fluid zone at particular T–P-${X}_{H_2O}$ conditions. It is assumed that the exchange occurs until the melt fraction increases to the threshold melt fraction value εb ≈ 0.3–0.45. In this approximation, the amount of added water is the product of its solubility and the critical melt fraction εb. In two series of numerical experiments run at pressures of 200 and 50 MPa, the interaction of water-filled porous felsite with near liquidus rhyolite magma resulted in water absorption, induced partial melting creating a narrow several meters wide mushy zone, and sluggish convection below that distributed water across the intruded sill. At P = 200 MPa, the addition of about 1.5 wt% water results in stronger volume convection, causing the melting rate to increase to 20 m/year. However, the addition of <0.22 wt% water induced no melting on the magma/contact mush interface, and the intruded sill crystallizes without convection. We apply the results of these numerical experiments to hot and dry rhyolites of the Yellowstone hot spot track magmas and then to the 2009 AD rhyolite sampled by the IDDP-1 exploration well in Krafla (Iceland). An active contact between the hydrothermal system within felsite and hot 963°C rhyolite magma was accidentally crossed at the depth of 2100 m, with a very thin (<30 m) transition providing information for a partial verification of our theoretical model. With the parameters observed in 2009, including the water concentrations in the melt (1.8 wt%) and felsite (0.92 wt%) and the high temperature of the intruded magma (945°C–980°C), we obtained slow melting of the preheated felsite roof at a rate of about 1 m/year. This seems reasonable if the 2009 magma was intruded during the 1975–1984 Krafla Fires eruption. We additionally present new δD values (−118‰) and H2Otot (1.6–2.0 wt%) concentration and reinterpret published δD and H2Otot values for felsitic fragments from IDDP-1; we find these to be lower than the δD in the mantle-derived magmas or hydrothermal systems. We demonstrate that, in theory, the formation of a fluid with such a low δD can be provided by the addition of low-δD water from OH-bearing minerals in hydrothermally altered roof rock. This may happen during dehydration of epidote from the altered roof rocks, and, alternatively, may also proceed by the mechanism of thermal diffusion (the Soret effect) through the partially molten/hydrothermal transition zone controlled by fluid fugacity. The high-temperature gradient in the contact zone between magma and the geothermal system of about 15–17°/m with conditions at the cold end close to the critical point for the aqueous fluid further decreases the expected δD value at the hot end of the contact zone to less than −110‰.