really very interesting, forgot to quote the abstract
[link to www.sciencedirect.com]
Journal of Volcanology and Geothermal Research
Volume 256, 15 April 2013, Pages 78–86

New insights into volcanic activity from strain and other deformation data for the Hekla 2000 eruption

...For the last eruption in 2000 we constrain the magma geometry by using a wide variety of deformation data: campaign GPS; an InSAR interferogram; dry tilt data, and borehole strain data. The dike that causes surface fissuring extends no more than ~ 0.5 km in depth, and the reservoir depth is ~ 10 km. These are connected by a conduit of small lateral extent. Data for previous eruptions are consistent with this model. We propose that the marked change in eruption interval is because this conduit remains liquid during the short interval between recent eruptions; only a small pressure increase is required to rupture the thin crustal seal. Such a state is consistent with precursory seismicity being confined to very shallow depths and may be applicable to other volcanoes that undergo abrupt changes in eruption interval.