Beyond elasticity: Are Coulomb properties of the Earth's crust important for volcano geodesy?

Abstract

Geodetic modelling has become an established procedure to interpret the dynamics of active volcanic plumbing systems and magma transfer within the crust. Most established geodetic models implemented for inverting geodetic data share similar physical assumptions: (1) the Earth's crust is modelled as an infinite, homogeneous elastic half-space with a flat surface, (2) there is no anisotropic horizontal stress to simulate tectonic stresses, (3) the source boundary conditions are kinematic, i.e. they account for an instantaneous inflation or deflation of the source. Field and geophysical observations, however, provide evidence that significant inelastic shear deformation of the host rock can accommodate the propagation of dykes and sills. We show that inelastic processes accommodating the emplacement of dykes in the brittle crust have large implications for dyke-induced surface deformation patterns. We present two quantitative laboratory experiments of dyke emplacement, during which the syn-emplacement surface deformation is monitored. In one experiment, the host material is elastic gelatine, whereas in the other experiment the host material is cohesive Coulomb, plastic silica flour. Even if both experiments produce sub-vertical dykes of similar shapes, their emplacement mechanisms and their associated surface deformation strongly differ. In the gelatine experiment, the dyke propagates as a tensile fracture in a dominantly elastic host, and the surface deformation exhibits two uplifting bulges separated by a trough parallel to, and above the apex of, the underlying dyke. Conversely, in the silica flour experiment, the dyke propagates as viscous indenter through a dominantly plastic host, and the surface deformation exhibits a single uplifting area that narrows through time. The comparison of our experiments shows that (1) plastic deformation (e.g., shear failure, compaction) of the host has large effects on dyke-induced surface deformation patterns and needs to be considered in geodetic models, and (2) dyke emplacement mechanisms matter in geodetic modelling, strongly suggesting that commonly used kinematic geodetic models such as the opening rectangular dislocation model (Okada 1985) are limited for revealing the physics and dynamics of volcano plumbing systems. Finally, our silica flour experiment shows that pure uplift geodetic signals can result from the emplacement of a dyke emplaced as viscous indenter, whereas such signals are commonly modelled using geodetic models of inflating spherical/elliptical or horizontal planar source. Our experiments call for the design of new geodetic models that account, even partly, for the plastic deformation component of the Earth's brittle crust.