SOME OBSERVATIONS WHICH CAN BE EXPLAINED BY THE MODEL

• The energy released in earthquakes in the SISZ indicates that the accumulated strain energy in the area around the SISZ is larger than expected if the strain build-up process in the area was only due to transversal plate motion of 2 cm/year  Stefansson/Halldorsson:1988. Heat energy from below the crust must play a significant role in the strain energy build-up.

• In the middle of the SISZ compressional stress is close to 45$^{\circ}$ E of North according to the shear wave splitting polarizations at the stations Sau and Sol in the SISZ  Stefansson/etal:1998. This is interpreted as the prevailing pore direction. It seems that the pores exist all through the crust and even below the seismogenic crust  Crampin/Zatsepin:1996. The fault plane solutions of small earthquakes in the SISZ in recent years express N-S vertical fault planes with RL motion  Stefansson/etal:1993. The compressional stress driving such earthquake slips is optionally 25-30$^{\circ}$ E of N, which might seem to oppose the stress orientation obtained from the shear-wave splitting observations. But it is not in opposition as earthquakes occur on preexisting faults even if the stresses are 15-20$^{\circ}$ from the optimal directions for the faulting. Most of these small earthquakes are close to the base of the seismogenic crust. They express starting from below of a N-S cut in the seismogenic crust. Such a cut is a candidate for a later large earthquake fault.

• The large historical earthquakes in the zone do not repeat the same N-S surface faults as were observed to have moved in older historical earthquakes  Einarsson/Eiriksson:1982. It is likely and not possible to exclude that there are repetitons on a longer time scale than the 300 years which we have a good information on.

• The migration of earthquake activity from a large earthquake in the eastward part of the zone, where the seismogenic crust is thick, to west where smaller earthquakes are triggered  Einarsson/etal:1981

• Clustering of microearthquakes and lack of large historical earthquake moment release coincide in the zone Stefansson/etal:1993. Clustering of earthquakes at about 20.4$^{\circ}$W in the zone near the base of the seismogenic crust coincides with lack of large earthquakes since at least 1700. This site is proposed to be a candidate for the next large earthquake in the zone. The next clustering to the west of microearthquake activity is at 20.8$^{\circ}$W, however, considerably shallower. This site has also little historical moment release in large earthquakes. This site is a candidate for a secondary earthquake following the large one further to the east.

• The magnitude 5.8 Vatnafjöll earthquake in 1987 near the eastern end of the SISZ (Figure /reffig:1) expressed a slightly east of north oriented right- lateral strike-slip fault similar to the SISZ earthquakes. Extension of the fault was monitored by strainmeters hours to a day after the earthquake. It was modelled  Agustsson/etal:1996 as caused by influx of fluids into the earthquake fault from below following the earthquake. The modelling of the Vatnafjöll earthquake being a N-S strike-slip earthquake near the eastern end of the of the SISZ is suggestive for what happens in the source area after a large earthquake in the SISZ.

• Recent activity in the western prolongation of the SISZ (Figure 1) is indicative for what may happen in larger events in the SISZ. The earthquake sequence on November 13-15, 1998, in Olfus occurred on a 10 km long and 2-3 km wide E-W zone, Figure 1,  RognvaldssonHen2/etal:1998. In this area the seismic activity is divided between this zone and activity farther to the north. The E-W Ölfus zone, however, is an almost direct continuation of the SISZ, although narrower, and taking up on a longer time scale probably only a fraction of the transversal and opening motion which is taken up by the SISZ.

  The seismic activity of November 13-15 migrated along the zone expressed in small earthquakes on E-W oriented faults and expressed in 2 large earthquakes (M 4.5-5) on N-S faults at two locations along the zone. GPS measurements show general expansion of the area together with E-W transverse motion (Árnadóttir 1999, personal communication) during a period spanning the earthquake. This characteristics of activity is readily explained by the above hypothetized model. As this zone is a narrow prolongation of the SISZ it is tempting to assume that this activity expressed comparable characteristics as would be expressed in SISZ earthquakes.

• Changing stress directions in the historical geology. Geological studies of paleostresses and mechanisms of earthquakes in recent years suggest two stress regimes which appear as opposite in the SISZ. The primary stress field is generally in agreement with present day seismicity. Two explanations are suggested for the secondary stress field, i.e. that it may be partly due to stress release and rebound and partly due to dike injections in the adjacent riftzone segments in South Iceland  Bergerat/etal:1998. The above hypethetized model implies a short lived and local secondary stress field due to expansion of faults following strike-slip earthquakes in the SISZ  Stefansson/Halldorsson:1988.