In all models, the pre-seismic stress level for most main shocks is high and fairly stable. This is also true for the events at the end of the sequence, namely those of 1912, of June 20, 2000, and - with minor reservations - of June 17, 2000. It indicates that the rather simple model can already explain the main features of the behaviour of the SISZ. This is especially astonishing, when the fact is kept in mind, that most (all but one) events used are not instrumentally recorded. Before the June 2000 events the SISZ seems to have been prepared for rupturing at the specific locations. This fits very well to the findings from stress measurements in wells (see Subproject 4) that E-W left-lateral shear stress was acting on the fault zone. To improve the model further, the initial unknown stress field of 1706 could be reduced in the eastern part and the central part, where the first events did not occur before 1732 and 1734, respectively. Another improvement might be to include more basal drag as the source of plate tectonic stress increase as compared to ridge push.
Nevertheless, the problem remains, why some events, as those of September 6, 1896, or June 21, 2000, did not occur earlier (i.e. at lower stress), just passing the "threshold in pre-seismic stress" (here e.g. the average pre-seismic stress (cf. Figure 59). Only if this would be the case, a prediction of the occurrence time might be at reach.
Even though the earthquake rupture planes strike N-S, the stress changes calculated here affect the whole area of the SISZ.
A tendency with time towards higher values of pre-seismic stress was found. It is an indication that the stress increase due to rifting might have been assumed too high, i.e. not all of the stress increase due to the spreading rate of 2 cm/year (assumed to be constant between 1706 to 2000) was released by earthquakes. The assumption that only half of the accumulated stress is seismically released led to a rather constant pre-seismic stress level.
The variation in model parameters does not lead to totally different results, i.e. the model is rather stable.
In general, the models go beyond the standard earthquake moment release and hazard analysis as they include the spatial location and extension of the events, quantify the amplitudes of stress release and that of plate motion on the faults, as well as providing an extrapolation to the present stress situation. This report will be submitted for publication in an appropriate form.