To gain a better insight of the hydrogeology and the location of the main tectonic faults of Stromboli volcano in Italy, we collected electrical resistivity measurements, soil CO2 concentrations, temperature and self-potential measurements along two profiles. These two profiles started at the village of Ginostra in the southwest part of the island. The first profile (4.8 km in length) ended up at the village of Scari in the north east part of the volcano and the second one (3.5 km in length) at Forgia Vecchia beach, in the eastern part of the island. These data were used to provide insights regarding the position of shallow aquifers and the extension of the hydrothermal system. This large-scale study is complemented by two high-resolution studies, one at the Pizzo area (near the active vents) and one at Rina Grande where flank collapse areas can be observed. The Pizzo corresponds to one of the main degassing structure of the hydrothermal system. The main degassing area is localized along a higher permeability area corresponding to the head of the gliding plane of the Rina Grande sector collapse. We found that the self-potential data reveal the position of an aquifer above the villages of Scari and San Vincenzo. We provide an estimate of the depth of this aquifer from these data. The lateral extension of the hydrothermal system (resistivity ∼15–60 ohm m) is broader than anticipated extending in the direction of the villages of Scari and San Vincenzo (in agreement with temperature data recorded in shallow wells). The lateral extension of the hydrothermal system reaches the lower third of the Rina Grande sector collapse area in the eastern part of the island. The hydrothermal body in this area is blocked by an old collapse boundary. This position of the hydrothermal body is consistent with low values of the magnetization (<2.5 A m−1) from previously published work. The presence of the hydrothermal body below Rina Grande raises questions about the mechanical stability of this flank of the edifice.
Crustal accretion at divergent plate boundaries typically occurs via the periodic intrusion of dikes, but their emplacement and the associated deformation are rarely observed. The few existing observations at subaerial rifts show that these diking events are followed by a decadal‐scale period with extension rates faster than the secular divergent plate motion. This transient accelerated deformation has been explained by continued subsurface magma injection or by relaxation, in the viscoelastic mantle, of the stress changes imparted by dike opening. For the first time, GPS measurements were collected within a few months of a rifting event at a major plate boundary, the September 2005, 60 km‐long dike intrusion in the Dabbahu segment, Afar, Ethiopia. Extension rates for the first 3 years greatly exceed the plate motion (Nubia‐Arabia) secular divergence rate, even at sites located more than 60 km from the rift axis. Here we show that these observations are consistent with stress relaxation in a viscoelastic upper mantle with a viscosity of about 5 × 10 18 Pa·s overlain by a 12–14 km‐thick elastic crust. The alternative model of continued diking requires continuous opening well below the Moho and is therefore unlikely. Instead, magma injection in Afar since June 2006 has taken the form of smaller discrete diking events, tapping into a mid‐crustal melt reservoir under the segment center.
A 60-km-long dyke intruded the Dabbahu segment of the Nubia—Arabia Plate boundary (Afar, Ethiopia) in 2005 September, marking the beginning of an ongoing rifting episode. We have monitored the continuing activity using Satellite Radar Interferometry (InSAR) and with data from Global Positioning System (GPS) instruments and seismometers deployed around the rift in response to the initial intrusion. These data show that a sequence of new dyke intrusions has reintruded the central and southern section of the Dabbahu segment. The first was in 2006 June and seven new dykes were emplaced by the end of 2007. Modelling of InSAR data indicates that the dykes were between 0.5 and 2 m wide, up to ∼10 km long and confined to the upper 10 km of crust. An intrusion in 2007 August was associated with a 5-km-long basaltic fissural eruption. During the new dyke injections, InSAR and GPS data show no subsidence at either of the volcanoes at the northern end of the segment, which partly fed the 2005 September dyke. Seismicity data imply that the dykes were probably fed from a source near the Ado'Ale Silicic Complex at the centre of the segment, but the lack of significant subsidence there implies that the source is very deep, or that there was minimal deflation at shallow magma sources. The new dykes are concentrated in an area where the 2005 dyke did not produce significant opening, implying that residual tensile tectonic stresses are higher in this location and are focusing the later intrusions. The sequence of dyke intrusions observed so far is similar to those seen in Iceland during the Krafla rifting episode, which lasted 9 yr from 1975 to 1984. It is likely that, with a continued magma supply, dykes will continue to be intruded until the tectonic stress is fully relieved. As observed at Krafla, eruptions are likely to become more common before the rifting episode is concluded.