Department of Geology and Geophysics, The University of Edinburgh, Grant Institute, West Mains Road, Edinburgh EH9 3JW.

Almost all rocks, including most igneous rocks, contain fluid-saturated intergranular microcracks which are aligned by the stress field. The microcracks are the most compliant elements of the rockmass, and the immediate response of rock to changes of stress (and most other conditions) is to modify the geometry of the microcracks. Seismic shear-waves are very sensitive to aligned microcracks and display shear-wave splitting, analogous to the bi-refringence of light in crystals, when propagating through distributions of aligned cracks, and stress-induced changes in crack geometry can be monitored with shear-wave splitting. Such shear-wave splitting is seen in almost all igneous, metamorphic, and sedimentary rocks, regardless of the porosity or crack density. In particular, the polarizations of the shear-waves are aligned with the stress field, and changes in shear-wave splitting have been recognized before those earthquakes where appropriate measurements are available. [Note that to record interpretable shear-waves at the surface, raypath incidence-angles must be within the shear-wave window (less than 45 degrees from the vertical).]

PRENLAB SubProject 3 uses shear-wave splitting to monitor crustal deformation in Iceland, where seismic records are picked up via the Internet. Almost all available shear-wave recordings have been analyzed for the period September 1996 to April 1997. In that period, there is only adequate data (sufficient small earthquakes within the shear-wave window) for four stations (BJA, KRI, KRO, and SAU) in the South of Iceland Seismic Zone (SISZ), with marginal data at ASM, HAU, SAN, SOL in the SISZ, and at station GRI on the island of Grimsey, North of Iceland. In general, the observed shear-wave splitting has similar characteristics to that seen elsewhere around the world, except that the normalized values of the time delays between the split shear-waves in Iceland (5 to 20ms/km) are higher than those usually observed elsewhere (1 to 8ms/km). This is almost certainly caused by the high heat flow in Iceland. We have not detected any significant temporal variations during this period (except that the SISZ stations show a slight decease in time delays that may indicate stress relaxation following the Vatnajokull eruption, see below).

Note that the Vatnajokull eruption was on 30th September 1996. The only station which we have analyzed for more than a few weeks before the eruption is station SAU at the East end of the SISZ, about 160km SSW of the fissure eruption. For five months before the eruption, the shear-wave splitting at SAU shows similar directional variations of time delays between the split shear-waves as those typically seen before earthquakes. These can be interpreted as the microcracks swelling (increasing in aspect ratio, or dilatancy) as stress increases in a direction parallel to the cracks. (There is now a comprehensive theoretical model supporting this interpretation.) The hypothesis is that the increase of stress is caused by the aseismic injection of magma into the lower crust before erupting through the upper crust. This suggests that the approach of volcanic eruptions as well as earthquakes can be monitored by analyzing shear-wave splitting.

The Vatnajokull eruption was a massive event marking a spreading episode of the North Atlantic Ridge and the effects would be expected to be visible at all the SISZ stations, where there is adequate data coverage within the shear-wave window. Consequently, one of the future priorities in this subproject will be analyzing shear-wave splitting before the eruption at, particularly, the SISZ stations. In general, this subproject, and the additional developments at Edinburgh, confirm that deformation of the rockmass, in particular, increasing stress before earthquakes and eruptions can be monitored by analyzing shear-wave splitting.