Modélisation numérique des processus magmatiques

***WHAT IS PHYSICAL VOLCANOLOGY ?

The surprising diversity of magmas erupted at the Earth’s surface is caused by the chemical evolution taking place because magmas are cooling within the Earth’s crust and crystallize, encounter new magmas and mix, and interact with the wall rock. Our broadest scientific objective is to unravel, from a physical standpoint, the complex interplays between crystals, exsolution bubbles, and melt that affect the pressure/temperature path of magmas and condition the eruptive behavior of volcanoes at large.

Physicists tracking magmatic processes are mostly concerned with the location and motion of the three main phases (melt, crystals, and gas bubbles) composing the magma. These phases can be trapped into mobile or immobile regions, and physicists define magma chambers as regions of mobile magma available for eruption. Magma mobility is linked to rheology, which depends not only on quantities such as crystal content or melt chemistry, but also on the stress state the magma is subjected to. Thus, for physicists, a magma chamber can significantly change in shape and size during the course of an eruption. These three visions are illustrated by the outcomes of studies of the 1991 eruption of Mt Pinatubo :

The many faces of the magma chamber feeding the 1991 eruption of Mt Pinatubo. Geochemistry of the erupted products led to the identification of the storage conditions of a host magma and input of another, fresh magma (Hammer & Rutherford, 2003). Geophysics located the many volcano-tectonic quakes generated during the eruption, which led to the identification of an aseismic zone (red) that may contain the host magma (Mori et al, 1996). Physics integrated these elements into a scenario of a chamber filled by highly viscous host magma and being remobilized by fresh magma, which led to the identification of a plausible reheating mechanism (Burgisser & Bergantz, 2011).

***PHYSICS OF MAGMA ASCENT

Our modeling approach considers magma as a mixture of crystal, gas bubbles and silicate liquid with independent motions. It is at the forefront of fluid dynamical simulations of volcanic processes and has successfully been applied to the simulation of pyroclastic density currents, magmatic conduits, and magmatic chambers. We use mainly two in-house fluid dynamical models to that end. One is based on the numerical model Multiphase Flow with Interphase eXchange (MFIX) and the other has successively been developed in Collier & Neuberg (2006) and Collombet (2009).

Multiphase simulations (crystal + melt) of the convection of the long-lived lava lake at Erebus volcano, Antarctica. After 15 years (right), in steady state, thermal convection is only able to suspend 20 vol.% crystals (yellow color), which is smaller than the 35 vol.% observed (green color, initial conditions at year 0, left). This suggests that the gases are needed to accelerate convection and suspend more crystals.

	Principle of simulation of magma ascent in a volcanic conduit. (Collombet, 2009). The magma rises in the conduit and becomes richer and richer in gas bubbles. These gases accumulate in the magma, possibly leaking into the surrounding rocks. If gas loss is too slow, the magma fragments and an explosive eruption occurs.

Magma rheology is controlled by the amounts of crystals and volatiles present in the magma and the strain rates magma is subjected to. These controls have been approached separately, and our aim is to bring them together. The resulting simulation outputs need to be confronted to natural data. The proposed natural target is Merapi volcano in the framework of DOMERAPI, a multidisciplinary research project that seeks to integrate geophysics, geochemistry, and physical volcanology into a deterministic model of the behavior of this active volcano.

Example of coupling between magma flow and the elastic deformation, which can be monitored around a volcano (Albino et al. 2011).

***VOLATILE CHEMISTRY

Magmas show such complex degassing patterns that it is impossible to predict the gas composition at vent without a thermodynamical model. Our chemical model of degassing allows us to calculate the evolution during decompression of the volatile composition of gas and melt for the S-O-H-C-Cl-Fe system in rhyolitic, basaltic, and phonolitic melts :

Inverse modeling of volcanic gas composition measurements. Measures at the surface of the lava lake of Erebus volcano, Antarctica, show strong variation in gas composition between quiescent convection (triangle) and Strombolian explosions (star, blue shaded area shows variability between explosions). Schematics (right) depict various physical scenarios, where explosions are caused by fast gas slugs or slow bubble swarm, and quiescent degassing is fed by a pulsatory or still magma column. Depending on the scenario, the model predicts different evolutions of volatile distribution at depth (left).