Reawakening of a volcano: Activity beneath Eyjafjallajökull volcano from 1991 to 2009

The ice-capped Eyjafjallajökull volcano, south Iceland, had been dormant for 170 years when the first signs of reawakening of the volcano were captured by seismic and geodetic measurements in 1994. These were the first clear observed signs of unrest followed by 16 years of intermittent magmatic unrest culminating in 2010 when two eruptions broke out on the flank and at the summit. We analyze seismic data from 1991 through 2008 and GPS data from 1992 to May 2009 to infer magma movements beneath the volcano. The relocated earthquakes reveal an overall pipe-like pattern northeast of the summit crater, sporadically mapping the pathway of magma from the base of the crust towards an intrusion in the upper crust. During the study period, three major seismic swarms were recorded. Two of them, in 1994 and 1999-2000, occurred in the upper and intermediate crust and accompanied crustal deformation centered at the southeastern flank. No uplift was detected during the 19- to 25-km-deep 1996 swarm, near the crust-mantle boundary, but the horizontal, ~ E-W oriented T-axes indicate a period of tension/opening, suggesting magma intruding up into the base of the crust. The GPS measured deformation during 1999-2000 can be modeled as intrusion of a horizontal, circular sill with volume of 0.030 ± 0.007 km3 at 5.0 ± 1.3 km depth. The less constrained 4.5- to 5-km-deep sill model for the 1994 episode indicates a three times smaller intruded volume (0.011 km3) than during 1999-2000. In the years between/following the intrusions, contraction was observed at the southeastern flank. The contraction from 2000.5 to 2009.3 can be fitted by a circular sill model with a volume contraction of − 0.0015 ± 0.0003 km3/year at 5.5 ± 2.0 km depth. The less well constrained model for 1994.7 to 1998.6 gives a volume contraction of -(0.0009-0.0010) km3 at a fixed depth of 5 km. The accumulated volume changes (~− 0.013 km3 for the second period, ~ 0.0037 km3 for the first period) are much larger than expected due to solidification and cooling of magma alone and might partly be explained by the underestimated volume of intruded magma, mass loading effects within the crust due to the intruded magma and possibly, and to less extent, degassing (CO2).