Three-phase flow dynamics in the lava lakes at Mount Erebus, Antarctica

Abstract:

Long-lived, persistently active lava lakes expose the top of a convecting magma column to direct observation and offer a unique window into the cryptic magmatic plumbing system at depth. In this paper, we focus on the lava lake at Mount Erebus, a large intraplate stratovolcano at Ross Island, Antarctica, to gain new insights into the multi-phase interactions between gas bubbles, crystals and magmatic liquid in basaltic volcanoes.

Early studies of magmatic convection have considered multi-phase magmas as perfectly homogeneous mixtures. The high proportion of erupted gas relative to magma, however, suggests that gas separates from the flow and drives eruptive activity. Similarly, the large size (up to 10cm) of the megacrysts that make up 97% of the crystal cargo at Erebus begs the question whether these crystals are likely to remain entrained and how crystal segregation in the lava lakes and conduit alters eruptive behavior.

We study the multiphase behavior of magmatic convection at Mount Erebus through two dimensional numerical simulations. Our model was developed with Mount Erebus in mind, but we argue that it could also serve as a virtual laboratory for studying multiphase flow in other basaltic systems. To accurately capture the deformability, breakup and coalescence of large gas bubbles, we track the gas-liquid interfaces with level-set functions. The crystal phase is incorporated using distributed Lagrange multipliers. We discretize the multiphase Stokes and energy equation through an iterative finite difference method that captures the potentially discontinuous jumps in the pressure, stresses, density and viscosity through a Ghost-Fluid approach. We have benchmarked and validated our numerical approach against analytical results and laboratory experiments.

We synthesize observations of thermal flux, seismic behavior, geodesy and geochemistry to deduce constraints on the mass flux, conduit dimensions, reservoir size, and crystal growth as a basis for our numerical experiments. We then compare our model results against surface observations of convective speed and eruptive bursts. We find that phase separation and the dynamic interaction between the phases play an important role in reproducing surface observations.