The puzzle of the Bardarbunga volcano, Iceland earthquake

The Bardarbunga volcano lies beneath the 500-m-thick Vatnajokull icecap, the largest glacier in Europe. Earthquakes with atypical seismic radiation have occurred beneath the Bardarbunga caldera and have been routinely reported in the Global CMT catalog. An earthquake with Mw 5.6 and a strong non-double-couple (NDC) radiation pattern occurred beneath the caldera on 29 September, 1996. A peculiarity of that earthquake was that it was the first in a sequence of seismic and magmatic events and that it was followed, not preceded or accompanied, by a major eruption which ultimately led to a breakout flood from the subglacial caldera lake. The earthquake was recorded well by the regional-scale Iceland Hotspot Project seismic experiment. One of the proposed hypotheses to explain the observed displacements and the sequence of events was the inflation of a shallow magma chamber that might have caused rupture on ring faults below the chamber.

Bardarbunga caldera is about 10km wide in diameter. The volcano lies under Vatnajokull icecap, the largest glacier in Europe. The earthquake with anomalous seismic radiation took place on 29 September, 1996.

Gjalp subglacial eruption (3 October, 1996). Photo taken by Oddur Sigurdsson, Iceland Geological Survey. See an airplane in the photo for scale.
Iceland has a heterogeneous crust, with variable thickness, and thus a 1-D structural model is not ideal for waveform modeling. We investigated the earthquake with a point-source complete moment-tensor (MT) inversion method using regional long-period seismic waveforms and a composite structural model of Iceland based on joint modeling of teleseismic receiver functions and surface-wave dispersion. When such a model is used (source-station pairs with different 1-D models used in the inversion are shown by different colors in the figure shown below on the left), the waveform modeling yields a non-double-couple solution with a strong, vertically oriented compensated linear vector dipole component and a statistically insignificant volumetric contraction (shown below on the right). The absence of a volumetric component is surprising in the case of a large volcanic earthquake that cannot be explained by shear slip on a planar fault. A possible mechanism that can produce an earthquake without a volumetric component involves two offset sources with similar but opposite volume changes (shown in a sketch at the bottom of the page). We show that although such a model cannot be ruled out, it is unlikely (Tkalčić et al., 2009).

Map showing the main tectonic and volcanic features in Iceland. Glaciers are shown in white and spreading segments in dark gray. Volcanoes are shown with thin lines. The Bardarbunga earthquake is shown with the white star. Triangles are locations of the eleven Iceland Hotspot Project broadband stations used in the moment tensor inversion. Lines indicate the locations of the six different one-dimensional models used between the source and stations.

Full moment-tensor point-source inversion results for the Bardarbunga event. Three-component displacement seismograms (radial, vertical and transverse, from left to right) are shown by solid lines and compared to one-dimensional synthetic seismograms (dashed lines). The lower-hemisphere projection of the P-wave radiation pattern is shown at right. The strike, rake and dip of the two nodal planes of the best DC solution, as well as the scalar seismic moment, moment magnitude, and percentage DC, CLVD and isotropic, are given numerically at right.

Sketch of tested models with two compensating sources reproducing a vertically oriented CLVD in dilatation (where volumetric exchange is equal and opposite in sign): (a) a complex magma chamber; (b) an implosive source and a vertical, horizontally opening crack above.

WORK IN PROGRESS: In order to investigate the hypothesis of a rupture occurring on a ring fault, we simulated different caldera geometries and rupture scenarios on the walls of a conical surface. We obtained excellent fits for ruptures extending along one-half perimeter of the caldera at a super-shear velocity, but could not determine the location of the initiation point nor the rupture propagation direction (Tkalčić et al., in preparation). If studied in different frequency bands however, the point source MT inversion fails to simultaneously explain the observed data, and this indicates the presence of finite-source effects. Using a 3D model of the Icelandic crust and upper mantle, we perform a probabilistic finite source inversion (Fichtner and Tkalčić, submitted). One of the most robust outcomes of this is a well-constrained source duration with approximately equal amount of energy radiated by individual segments. This indicates that the caldera dropped coherently as a single block. We speculate that the earthquake accompanied a small-scale eruption that went unnoticed prior to the caldera drop caused the earthquake. The caldera drop could have increased the pressure in the magma chamber thus inducing the principal eruption.

This is an electronic version of an article published by Bulletin of the Seismological Society of America ; Copyright (2009) Seismological Society of America:
Tkalčić, H., D.S. Dreger, G.R. Foulger and B.R. Julian, The puzzle of Bardarbunga vocano, Iceland earthquake, Bull. Seismol. Soc. Am., 99:3077-3085, 2009.

Download this paper

Download the AGU 2009 invited presentation here

also available on the Seismological Society of America online delivery service,
accessible via the journal's Website at:

back to homepage