Tracking melt injection under the Mid-Atlantic Rift near Askja, central Iceland

The crust of the large oceanic island of Iceland in the North Atlantic is formed almost entirely from successive volcanic eruptions which fill the gap caused by the separation (rifting) of the plates on which North America and Europe sit. The massive volcanism in Iceland is caused by the deep mantle there being hotter than normal, and so melting more than normal to produce large amounts of molten rock, or magma. Although the melt is formed in the deep mantle at depths of 40-100 kilometres, it is very buoyant and bleeds rapidly up towards the surface. Normally it ponds in the crust in a magma chamber about 5-6 kilometres beneath the surface until a sufficiently large volume has built up to cause an eruption. Our main objective is to investigate some exceptionally deep earthquakes occurring 15-30 kilometres beneath the surface under the active volcanic system of Askja in central Iceland. These occur at depths where the crust should be hot enough to prevent brittle failure generating any earthquakes. We first discovered these deep earthquakes in summer 2006, and will now target an array of seismometers specifically to investigate them further. Our hypothesis is that they are caused by melt moving in the deep crust. This is an exciting observation, since such deep earthquakes have only rarely been observed before, and never in a rift zone; it means that we can catch the volcano in the act of transferring molten rock through the crust. A second objective is to monitor a new area of activity, again thought to be caused by melt injection, which started in February 2007 beneath a previously seismically quiet area on the edge of the rift zone some 20 km east of Askja. Over 3000 events have occurred up to mid-August 2007, gradually decreasing in depth from 20 to 15 km and latterly extending almost to the surface. This may represent injection of molten rock into a shallow magma chamber which may result in a volcanic eruption. Previous experience suggests that activity is likely to persist for around 2 years, so we have an opportunity to track almost the entire development of a melt injection (and possibly eruption) episode. Thirdly we will map the location of the underground storage reservoirs (magma chambers), by using the fact that the crust above them is cold and brittle, therefore allowing numerous small earthquakes to occur as it fractures, whereas the hot magma chambers deform ductilely without fracturing. So by mapping where the earthquakes are absent, or where seismic waves are absorbed by passing through rock containing some melt, or rock at high temperatures, we can map the region of the magma chamber. Our seismic data will also be used to calculate the crustal thickness in this region, which is presently poorly known. Results form our studies will be integrated with information from other researchers investigating deformation in the region using satellite measurements which can detect motions at a millimetre scale, together with surface mapping of faults in the crust and evidence from detailed petrological studies of the rocks which provide control on the depths at which the molten rock has ponded in the crust during its ascent to the surface. The Askja area is inhospitable, and is easily accessible only in the summer months of July-August when snow is absent. Winter temperatures plummet to -25C. We shall install our seismometer array during summer 2008, but leave them operating remotely through the winter to gain a complete picture of the seismicity through time. The seismometers include a satellite clock receiver which provides accurately timing. Solar panels with large batteries will power the instruments through the dark winter months. Data from one of the stations will be radio-telemetered to Reykjavik and hence accessible via the internet from Cambridge, while the remainder will record data internally, to be collected in summer 2009.