EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Post-eruptive volcano inflation following major magma drainage: Interplay between models of viscoelastic response influence and models of magma inflow at Bárðarbunga caldera, Iceland, 2015-2018

Siqi Li1, Freysteinn Sigmundsson1, Vincent Drouin2, Michelle M. Parks3, Kristín Jónsdóttir3, Benedikt G. Ofeigsson3, Ronni Grapenthin4, Halldór Geirsson1, and Andy Hooper5
Siqi Li et al.
  • 1Nordic Volcanological Center, Institute of Earth Science, University of Iceland, Reykjavik, Iceland (
  • 2Iceland GeoSurvey (ÍSOR), Reykjavik, Iceland
  • 3Icelandic Meteorological Office, Reykjavík, Iceland
  • 4Geophysical Institute and Dept. of Geosciences, University of Alaska Fairbanks, Fairbanks, USA
  • 5COMET, School of Earth and Environment, University of Leeds, Leeds, UK

Unrest at Bárðarbunga after a caldera collapse in 2014-2015 includes elevated seismicity beginning about six months after the eruption ended, including nine Mw>4.5 earthquakes. The earthquakes occurred mostly on the northern and southern parts of a caldera ring fault. Global Navigation Satellite System (GNSS, in particular, Global Positioning System; GPS) and Interferometric Synthetic Aperture Radar (InSAR) geodesy are applied to evaluate the spatial and temporal pattern of ground deformation around Bárðarbunga caldera outside the icecap, in 2015-2018, when deformation rates were relatively steady. The aim is to study the role of viscoelastic relaxation following major magma drainage versus renewed magma inflow as an explanation for the ongoing unrest.

The largest horizontal velocity is measured at GPS station KISA (3 km from caldera rim), 141 mm/yr in direction N47oE relative to the Eurasian plate in 2015-2018. GPS and InSAR observations show that the velocities decay rapidly outward from the caldera. We correct our observations for Glacial Isostatic Adjustment and plate spreading to extract the deformation related to volcanic activity. After this correction, some GPS sites show subsidence.

We use a reference Earth model to initially evaluate the contribution of viscoelastic processes to the observed deformation field. We model the deformation within a half-space composed of a 7-km thick elastic layer on top of a viscoelastic layer with a viscosity of 5 x 1018 Pa s, considering two co-eruptive contributors to the viscoelastic relaxation: “non-piston” magma withdrawal at 10 km depth (modelled as pressure drop in a spherical source) and caldera collapse (modelled as surface unloading). The other model we test is the magma inflow in an elastic half-space. Both the viscoelastic relaxation and magma inflow create horizontal outward movements around the caldera, and uplift at the surface projection of the source center in 2015-2018. Viscoelastic response due to magma withdrawal results in subsidence in the area outside the icecap. Magma inflow creates rapid surface velocity decay as observed.

We explore further two parameters in the viscoelastic reference model: the viscosity and the "non-piston" magma withdrawal volume. Our comparison between the corrected InSAR velocities and viscoelastic models suggests a viscosity of 2.6×1018 Pa s and 0.36 km3 of “non-piston” magma withdrawal volume, given by the optimal reduced Chi-squared statistic. When the deformation is explained using only magma inflow into a single spherical source (and no viscoelastic response), the optimal model suggests an inflow rate at 1×107 m3/yr at 700 m depth. A magma inflow model with more model parameters is also a possible explanation, including sill inflation at 10 km together with slip on caldera ring faults. Our reference Earth model and the two end-member models suggest that there is a trade-off between the viscoelastic relaxation and the magma inflow, since they produce similar deformation signals outside the icecap. However, to reproduce details of the observed deformation, both processes are required. A viscoelastic-only model cannot fully explain the fast velocity decay away from the caldera, whereas a magma inflow-only model cannot explain the subsidence observed at several locations.

How to cite: Li, S., Sigmundsson, F., Drouin, V., Parks, M. M., Jónsdóttir, K., Ofeigsson, B. G., Grapenthin, R., Geirsson, H., and Hooper, A.: Post-eruptive volcano inflation following major magma drainage: Interplay between models of viscoelastic response influence and models of magma inflow at Bárðarbunga caldera, Iceland, 2015-2018 , EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-19030,, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 04 May 2020
  • CC1: Questions and answers from the live chat during EGU2020, Michael Heap, 11 May 2020

    Q: Oh no, Bardarbunga uplifting again! Your estimated point source depth is surprisingly shallow, less than 1 km? Is it centered at the caldera, or near the rim? If it is so shallow, what is it?

    A: Thank you for your question. The estimated source (inflow) at less than 1 km is within the caldera, about sea level

    Q: What viscoelastic configuration (model) do you use? And do you model your source with a pressure or volume change? (as this will influence whether you see VE creep, relaxation or recovery)

    A: We use software called RELAX to have a two-layer model, a 7 km elastic layer + viscoelastic below that. We use volume change

    Q: Are there gravity data available for this area? Could it improve the modeling?

    A: There are gravity data measured. I think Magnus Tumi Gudmundsson has one presentation on this topic earlier on this conference. It helps to constrain our model for sure. The suggestion from gravity is similar - a shallow source. But we need to understand if continuous uplifting several meters for a few years is realistic.

    Q: How did you decide on the thickness of the upper elastic part of your model?

    A: We have seismicity studies in the area suggesting the depth of the brittle-ductile boundary depth between 4-6 below sea level. That corresponds to our modelled elastic layer thickness.

    Q: What factors did you consider for using a Maxwell rheology (a fluid) for the viscolestic portion of the crust?

    A: There are studies on Glacial Isostatic Adjustment studies over Iceland using viscoelastic rheology to study the area. So here we consider using Maxwell rhelogy.

    Q: Thanks. In the abstract you say there have been several M4.5+ events, how do you think they contribute to the observed deformation?

    A: It suggests the Bardarbunga is in an unrest stage. The earthquake locations can provide more information for us, for example, if there are caldera ring fault slip in this area and can they influence our modeling. Seismicity studies suggest that caldera ring fault is moving - so we also need to consider that in our study.

    Q: Very interesting research! How much do the results of your joint model vary with varying viscosity? Is 5e18 the best fit? And what about the effect of the thickness of the elastic layer?

    A: We did a grid search for viscosity. Based on our data, it seems 3.1e18 is the optimal viscosity. We didn't vary the thickness of the elastic layer - we will consider that in future studies.

    Q: Thanks, but could some of the deformation be coseismic deformation?

    A: I won't rule that out. But our closest continuous GPS station is around 3 km from the caldera rim and we didn't see any co-seismic deformation signal from there. We have also study the possibility to have sill source inflation + fault slip, which can also fit our observation (not presented in the presentation).