Part of this work has focussed on quantitative field studies and modelling of the linking up of fractures into normal faults (Acocella et al. 2000) and strike-slip faults (Belardinelli et al. 2000). Part of the work, however, has focussed on the general evolution of the seismic zones, exploring the model of a stress-field homogenization being a necessary condition for the generation of large earthquakes (Guðmundsson and Homberg 1999). The work on the South Iceland seismic zone made in collaboration with Maria Elina Belardinelli and Maurizio Bonafede (Belardinelli et al. 2000) is described in Subproject 7.
Detailed, quantitative field and photogeological studies were made of the interaction and linkage of extension fractures and normal faults (Acocella et al. 2000), a fault type that is common in the South Iceland seismic zone and the Tjörnes fracture zone. 90 zones of interacting fracture segments in Holocene pahoehoe lava flows of the rift zone of Iceland were studied, each zone being located between a pair of extension fractures or a pair of normal faults, with lengths from tens of meters to several kilometers. Of all the zones, only 7% are underlapping, whereas 93% are overlapping and mostly with hook-shaped fracture pairs. The length/width ratios of most overlapping zones are from 2-6, with a mean value of 3.5. In the overlapping zones, most fracture pairs show moderate shear (strike-slip) components which are related to local variations in the extension (opening) directions. Vertical displacements on normal faults decrease as the overstep and length of overlapping zone increase; both, in turn, are proportional to the total lengths of the faults forming the pair. During their evolution, these zones develop from an underlapping stage, through an overlapping stage (the most common configuration) and, finally, to a linkage stage. The geometrical features of overlapping spreading centres at mid-ocean ridges show great similarities to those reported here. These similarities indicate that the architecture and evolution of overlapping zones are scale independent.
It is proposed that on entering crustal parts where the state of stress is unfavourable to any particular type of seismogenic faulting, the fault propagation becomes arrested (Guðmundsson and Homberg 1999). This model is supported by field and numerical studies on the propagation of fractures of various types (Guðmundsson 2000e). It follows that prior to the propagation of an earthquake fracture, the stress conditions in the zone along the whole potential rupture plane must be homogenized. The proposed homogenization of the stress field in a large rock volume as a precursor to large earthquakes implies that by monitoring the state of stress in a seismic zone, its large earthquakes may possibly be forecasted.