Effects of a volatile phase on the Creation and Dynamics of Lower Crustal Magma Reservoirs: A Three-Phase Numerical Model Study

We use a one-dimensional numerical model to investigate how the presence of a volatile phase such as water impacts the creation and dynamics of a magma reservoir in the mid- to lower crust. We assume that the reservoir is created and sustained by the repeated intrusion of mantle-derived basalt sills containing a few wt% volatiles. Our numerical model solves the equations governing heat transfer by conduction and advection; two- and three-phase flow of the solid, melt and volatile phases assuming porous flow and compaction at low melt fraction and hindered settling at high melt fraction; and handles phase exchange using a two-component (SiO2 and volatile) chemical model fitted to experimental melting data. The solidus and liquidus temperatures are functions of bulk SiO2 content and melt volatile content. The volatile exchange between solid and melt phases is modelled using a partition coefficient and the maximum volatile content of the solid phase is capped based on mineralogical data. The melt phase has a maximum volatile saturation above which a free volatile phase is formed. There is no direct transfer of volatiles from the solid to the volatile phase.We find that the highly non-linear coupling between melt volatile content and the solidus and liquidus temperatures, flow of melt and (when present) volatile phases, and the exchange of latent heat, gives rise to complex emergent behaviour. We do not observe a simple, upwards propagating volatile front below which melt is always present in a stable mush column as has been proposed in some previous studies. Rather, we observe the formation of transient, high melt fraction, evolved and volatile-rich layers interspersed with refractory and volatile-poor mush. The high melt fraction layers can propagate upwards, merge and split. The layers are typically thin and likely below geophysical resolution. The dynamics of mush reservoir growth are strongly influenced by the fertility of the overlying crust: if the crust can melt in response to the addition of heat and volatiles, then the top of the reservoir migrates upwards until it reaches mid-crust depth, creating a reservoir that spans the lower- to mid-crust and hosts numerous melt-rich layers. Our results so far suggest a highly dynamic magmatic system characterized by significant melt fraction variations in time and space. Much of this dynamic complexity is lost in geophysical images that are interpreted to suggest reservoirs have low and uniform melt fraction.