Disentangling Ice and Magma Induced Surface Movements at Katla and Eyjafjallajoekull Volcanoes using InSAR and GPS
Spaans, Karsten¹; Hreinsdóttir, Sigrún²; Hooper, Andy¹; Ofeigsson, Benedikt^3
1School of Earth and Environment, University of Leeds, UNITED KINGDOM; ²Institute of Earth Science, University of Iceland, ICELAND; ³Icelandic Meteological Office, ICELAND

We study surface deformations around Katla and Eyjafjallajoekull volcanoes, south Iceland, using InSAR and GPS measurements between 2001 and 2009. We rule out the previously suggested possibility of magma induced inflation of Katla after 2004. In stead, we detect movements related to the thinning of ice caps. At Eyjafjallajoekull, we find an area of movements towards a central point, pointing to contraction of a cooling magma volume intruded in 1994 and 1999. We also see deformations at two close-by volcanoes, subsidence at Torfajoekull and uplift at Hekla.

Eyjafjallajoekull volcano is well known for its eruption in 2010, which disrupted flight traffic in Europe for several weeks. Abutting it to the east is Katla, one of the more active volcanoes in Iceland, with twenty confirmed eruptions since the settlement of the island 1100 years ago. The last confirmed eruption of Katla was in 1918, however since then several large joekulhlaups (glacial outburst floods) might have been caused by eruptions that did not break through the icecap covering the volcano, which is ~30 km in diameter and up to 600 m thick. Both volcanoes have been monitored using levelling and GPS, and since the beginning of the 2000's continuous GPS stations have been in operation. Complicating the monitoring of the volcanoes using surface deformation measurements is the fact that multiple source of deformation, like plate spreading, thinning of ice caps and magma movements, affect the area. Between 1999 and 2004, GPS stations placed on nunataks sticking out of the ice cap showed signs of inflation. Two continuous GPS stations on the south edge of the ice cap also showed movement away from the caldera. After 2004, the same two continuous stations continued to show movement away from the caldera, especially compared to a third GPS station 30 km away from the volcano, on the south flank of Eyjafjallajoekull. At the time, this was interpreted as a possible sign that Katla was still inflating. However, the long time span over which the movement occurred raised doubt to this explanation. Interferometric synthetic aperture radar (InSAR) is a differencing technique that allows for high accuracy, densely sampled deformation measurements. The dense sampling reveals the pattern of deformations, adding crucial constraints on the source of the signal. Here we use both GPS and InSAR results to show the surface displacements at Katla and Eyjafjallajoekull between 2001 and 2009, as well as provide insight into their sources.

We used 23 Envisat SAR images acquired between 2003 and 2009, covering Katla and Eyjafjallajoekull. We processed these images using the StaMPS InSAR time series processing methodology, altered to allow the use of several images affected by snow cover, which would normally reduce the quality of the point selection and velocity estimates. The resulting velocity map showed a long wavelength signal obscuring most regional deformation signals. This long wavelength signal was caused by uplift of the surface due to the thinning of the Vatnajoekull ice cap, to the north-east, and to a lesser extent, thinning of the Myrdalsjoekull ice cap on Katla itself. Furthermore, there is a ramp present in Envisat interferograms that increases systematically with time, adding to the long wavelength signal. We dealt with the ice thinning signal by using a model of the surface displacement resulting from ice unloading, constrained by GPS [Schmidt et al. 2010]. The systematic orbital ramp was estimated from the residual velocity field as a bilinear ramp. The resulting residual velocity field showed only regional signals, allowing for a better overview of activity around the volcanic systems in the area. We also used GPS data around Eyjafjallajoekull and Katla, which were processed using the GAMIT-GLOBK software relative to the ITRF 2008 reference frame using over 200 global reference stations. Together, the two techniques give an excellent overview of deformations that occurred between 2001 and 2009.

The figure below shows a close view of the residual velocity in the Katla and Eyjafjallajoekull area, with horizontal GPS velocities plotted on top. From this figure, we see no evidence of magma accumulation beneath Katla, as was previously hypothesised. Some residual velocities are present close to the icecap, but the shape and the fact that they follow the edge of the ice closely indicates that these movements are related to thinning of the ice cap which are not captured by the model. The same can be said for the horizontal movements of the GPS, most of which seem in line with movements associated with ice mass loss from Vatnajoekull and Myrdalsjoekull. This does not hold for several stations around Eyjafjallajoekull however. Three GPS vectors south of the volcano, and one to the north, show a movement towards a central point, indicating a contraction signal. The InSAR shows a similar signal, a small area of movement away from the satellite. This area coincides with the location of two intrusions underneath Eyjafjallajoekull in 1994 and 1999. We therefore infer that the deformation seen at Eyjafjallajoekull was most likely caused by a contraction of a cooling magma volume intruded in the 1990's. This also explains why the two continuous stations on Katla's southern flank were seen moving away from the volcano after 2004, while the third located south of Eyjafjallajoekull did not show this motion. At the third station, movement caused by ice thinning was likely counteracted by displacements related to the contracting magma volume. Two noteworthy signals unrelated to Katla or Eyjafjallajoekull are present in the residual velocity field, after removal of the long wavelength signal. These signals are outside of the figure shown below. To the north of Katla, a movement away from the satellite of about 1 cm\yr can be seen in the Torfajoekull volcanic system. Torfajoekull has been known to subside for at least 2 decades, likely due to the cooling of a large magma body underneath its central volcano. To the north-west of Katla, a subtle but broad signal towards the satellite of about 5 mm\yr is present. This is signal is just to the south-east of Hekla, a volcano that has been inflating since its last eruption in 2000.

Figure: Line-of-sight velocity map of the area around Katla and Eyjafjallajoekull. Ice thinning model and bilinear ramp have been removed. Red arrows give the horizontal components of the GPS velocities. Thin black lines and hatched lines give the outline of central volcanoes and caldera in the area.