A two‐magma chamber model as a source of deformation at Grímsvötn Volcano, Iceland

Abstract

Grímsvötn Volcano is the most active volcano in Iceland, and its last three eruptions were in 1998, 2004, and 2011. Here we analyze the displacement around Grímsvötn during these last three eruptive cycles using 10 GPS stations. The observed displacements in this region generally contain a linear component of tectonic and glacio‐isostatic origin, in agreement with the previously estimated values of plate motions and vertical rebound. Larger amplitude deformation observed close to Grímsvötn at the GFUM continuous GPS station clearly reflects a major volcanic contribution superimposed on a tectonic component. We estimate and subtract the tectonic trend at this station using regional observed displacement. The direction and pattern of the residual volcanic displacement (for coeruptive and intereruptive periods) are consistent for all three of these eruptive cycles. The posteruptive inflation is characterized by an exponential trend, followed by a linear trend. In this study, we explain this temporal behavior using a new analytic model that has two connected magma chambers surrounded by an elastic medium and fed by a constant basal magma inflow. During the early posteruptive phase, pressure readjustment occurs between the two reservoirs, with replenishment of the shallow chamber from the deep chamber. Afterward, due to the constant inflow of magma into the deep reservoir, the pressurization of the system produces linear uplift. A large deep reservoir favors magma storage rather than surface emission. Based on displacement measured at GFUM station, we estimate an upper limit for the radius of the deep reservoir of ∼10 km.

Assimilation of Deformation Data for Eruption Forecasting: Potentiality Assessment Based on Synthetic Cases

“…We use the two-magma reservoir model proposed by Reverso et al (2014). This model consists of two reservoirs embedded in an elastic medium and connected by a hydraulic pipe.…”

“…The good point is that the temporal evolution of the displacement field should help deciphering between those two cases, considering similar magma and crustal rheologies, the pressurization of a small chamber being much quicker than for a large one. One limitation of this approach, as recently demonstrated by Segall (2016), is that the temporal evolution actually results from a convolution between the history of a pressure source and the magma and crustal rheology (Reverso et al, 2014). It follows that additonal observations such as gravity data might be useful to discriminate both effects.…”

“…They provide an interpretation of the temporal evolution of geodetic data as well as seismic observations considering either the rheology of the encasing medium (Nooner and Chadwick, 2009;Carrier et al, 2014;Got et al, 2017) or the evolution of the magma inflow at the bottom (Lengliné et al, 2008;Pinel et al, 2010;Reverso et al, 2014).…”

“…The latter type of models has proven to be useful in recovering information of the deepest part of the magma plumbing system, such as the size of the deep storage zone as well as the bottom magma inflow (Reverso et al, 2014), which is always quite difficult to constrain. Segall (2013) has nicely demonstrated the interest to combine deformation data and a physics-based model of the plumbing system with a Bayesian-based approach in order to forecast eruptions.…”

“…We developed an efficient approach to assimilate GNSS and InSAR data into a twomagma reservoir model, which is a simple and generic dynamical model for the magma plumbing system, in order to forecast the overpressures and to constrain two model parameters related to the deepest part of the reservoir system. We begin by briefly discussing the two-magma reservoir model of Reverso et al (2014) in Section 2, then reviewing the fundamentals of the EnKF approach in Section 3, followed by the experiment setup and the step-by-step implementation of our EnKF strategy in Section 4. We then present two synthetic cases (i.e., case A and case B) when considering the magma inflow at the bottom of the system as well as the size of the deepest reservoir as uncertain and given two kinds of initial conditions about the uncertain parameters (i.e., biased or unbiased distribution).…”