Modelling three-phase magma dynamics during assimilation: Insights into the formation of low-δ18O rhyolites at Krafla, Iceland

The discovery of ¹⁸O-depleted igneous rocks at Krafla, Iceland, suggests that the system interacted with crustal rocks that experienced high-temperature hydrothermal alteration by a meteoric fluid to deviate from the expected mantle signature (δ¹⁸O = 5.5 ‰). Such assimilation is documented in low-δ¹⁸O settings worldwide, however, the mechanisms of this dynamic process remain poorly understood.  Due to intense drilling activity and exploration at Krafla, both hydrothermally altered crustal rocks and parental magma are comparably well characterized, making Krafla a great case study for the application of a numerical model that can further advance the understanding of the formation process of low-δ¹⁸O magmas. In this study, we use a new three-phase two-component thermo-chemical-mechanical model to simulate the effect of variable crustal compositions on the assimilation process and the magma chamber dynamics.  We define the simplified square-shaped magma chamber (10 x 10 m) of magma with initially basaltic composition (1250 °C) that assimilates the crustal rock (500 °C) at the top and bottom. Our results indicate that convective behaviour and the formation of cumulate layers can significantly hinder the assimilation process. While the crystal settling Stokes speed scale is the dominant driver for the formation of this boundary layer, depending on the assimilation timescales, the mushy chamber margins are able to grow to sufficient thickness to prohibit additional assimilation of low-δ¹⁸O crustal material. Density and buoyancy contrasts produce three types of convection: chamber convection, layered convection and plume driven convection. Final magma compositions in our preliminary model outputs range from mafic to intermediate but are not able to reach the felsic compositions encountered at Krafla. This suggests that evolution towards the erupted low-δ¹⁸O rhyolitic products involved multiple stages or included additional factors not yet accounted for in our model. Further refining of this and similar thermo-chemical-mechanical model setups may provide important new insights into the assimilation dynamics in the Krafla volcanic field and other low-δ¹⁸O settings worldwide.