Modeling Deformation Episodes at Large Silicic Systems using Poroelasticity: Examples from Long Valley Caldera and Laguna del Maule Volcanic Field.

Abstract:

The magma reservoir underlying large silicic systems presumably consists of a small fraction of interstitial melt residing within a rigid framework of crystals, called a “crystal mush” (e.g., Hildreth, 2004). For those systems, the magma would remain mushy during the long time intervals between eruptions (~10,000 to 100,000 years). Over shorter timescales (~10 years), episodes of unrest are characterized by surface uplift. To describe the time-dependent deformation field observed during volcanic unrest, we propose a new conceptual model. In a poroelastic medium, the flow of fluids within the pores is coupled to the deformation of the solid rock matrix (e.g., Wang, 2000). In our model, recurrent injections of mafic magma into the reservoir produce deformation within the surrounding crust. To solve the governing equations of poroelasticity, we apply the Finite Element Method (FEM) as implemented in the numerical modeling software COMSOL Multiphysics (v5.3). In the 2D axisymmetric configurations, a time dependent mass flux is imposed at the base of a porous domain, representing the magma reservoir. We validate the numerical solution by comparing it to analytical solutions for two models: a pressurized spherical cavity within an elastic half space (Mogi, 1958), and a single-phase fluid flowing through a conduit into a cavity (Le Mével et al., 2016). The key governing parameters include: the fluid viscosity, reservoir permeability, reservoir dimensions as well as the temperature and rheology (elastic or viscoelastic) of the matrix and surrounding crust. Here we consider: (1) Long Valley caldera (California) which has experienced five major episodes of uplift since 1975, as recorded by several geodetic techniques; and (2) Laguna del Maule volcanic field in the Southern Andean Volcanic Zone of Chile, where the rate of vertical uplift has exceeded 200 mm/yr for the past 10 years. For both cases, we estimate the best fitting model parameters to reproduce the geodetically measured spatial and temporal pattern of deformation. These new models lead to realistic estimates of the volume of magma intruded into the reservoir over each episode of unrest.