The effect of weak Mohr–Coulomb layers on the onset of caldera collapse - A limit analysis modelling approach

Abstract

Calderas are subcircular volcanic depressions that can occur due to drainage of a subsurface magma reservoir. Numerous models simulating the initiation and growth of caldera collapse consider homogeneous overburden of the magma reservoir. This study describes plastic models implementing limit analysis to investigate the effects of weak layers (low cohesion and low friction) on caldera formation and structure. Our models show that the presence of weak layers within the crust favours the onset of caldera collapse, as it reduces the critical magma underpressure within the magma chamber to initiate roof failure. This effect is more pronounced with greater cumulated thickness of weak layers. In homogeneous models, the onset of caldera collapse is accommodated by a simple, localized outward dipping reverse damage band (caldera fault), whereas in layered models caldera collapse is accommodated by more complex damage structures. Weak layers confine damage underneath the layers, limiting the growth of the caldera fault toward the surface. Calculated stress trajectories are rotated across weak layers, showing that weak layers act as stress barriers. The effect of weak layers is stronger when the layer is closer to the magma reservoir, where layer-parallel damage form underneath the layer, interpreted as a potential detachment level. Multiple layers trigger more distribution of the damage and several layer-parallel damage bands. The subsurface distribution of damage due to weak layers may lead to more distributed surface subsidence, enhancing sagging before a caldera fault reaches the surface. Finally, internal detachments due to weak layers are likely important for observed episodic transient subsurface collapse episodes before collapse occurs at surface. All in all, our models that implement plastic deformation predict significant stress perturbations as a result of varying Mohr–Coulomb properties only. Our study thus shows that widely used static elastic models are not sufficient for physically relevant stress analyses of geological systems. In addition, our study shows that plastic (or elasto-plastic) models are necessary to predict the location and extent of inelastic damage accommodating volcano deformation and failure.