Constraining the limits to magma chamber evacuation during explosive eruptions

Only a fraction of the magma generated in the earth finds its way to the surface during volcanic eruptions, while most of it will cool down and crystallize at different depths in the crust. Of particular interest is the pre-eruptive level, typically between 10km to 2km. Here, understanding the ratio erupted vs non-erupted magma has implications for volcanic eruption forecasting, long-term magmatic evolution, pluton formation, volcanic cyclicity, and post-eruptive geophysical monitoring. With a special focus on crystal-rich or mushy magmas, we address this problem by exploring, once a conduit reaches the surface, how efficiently the magma reservoir gets depleted and what regions of the reservoir are affected. We address those questions here using an unstructured finite element code, Gridap, written in Julia (Badia et al. 2020).Results show that several modes of magma advection exist including the classical pipe flow mode where a new batch of magma added to a mush chamber moves through a dike to the surface. Yet, several other modes also exist, which include a Stokes flow mode where magma does not make it to the surface despite a pre-existing open connection, and various intermediate modes. We use the numerical simulations to determine how magma rising speeds depends on the material and geometrical parameters such as magma and mush viscosities, or sizes of the magma batch, mush chamber or dike widths. As the numerical simulations cannot be performed for the full range of realistic magma viscosities, we use them to derive scaling laws for each of the mechanical deformation modes. These scaling laws can be used to extrapolate results to natural conditions, and highlight the key controlling parameters that determine whether melt buoyancy will result in an eruption or not. Importantly, it shows that there are physical limits to the volume of magma that can be erupted from a newly added batch of magma in a mush chamber. We will discuss the application of the results to natural cases. References citedBadia, S., & Verdugo, F. (2020). Journal of Open Source Software, 5(52), 2520.