There are two types of uncertainty in the probabilistic seismic hazard analysis (PSHA), namely：I – the uncertainty connected with the effect of future earthquake (source scaling, attenuation relation and local site effect), II – the uncertainty due to unpredictable nature of future earthquake (when, where, how large). The first uncertainty is caused by uncompleted knowledge and may be reduced by using additional data (see the Part I of this report). At present there is a common opinion that the only way to reduce the second type of uncertainty is the development of the several models which approach the reality (for example, a Monte Carlo method to generate a synthetic earthquake catalogue). In most cases the definition of the earthquake source zone parameters (configuration, maximum possible magnitudes, and earthquake recurrence) is carried out on the basis of the historical data, i. e. using the events that were already occurred. By the other words, statistical properties of the known seismicity are assumed to be stationary in time and space, and therefore should apply in the foreseeable future. There are many examples, including the case of the Chi-Chi earthquake, when the erroneous estimation of these parameters led to the grave consequences. Thus, there is a necessity to estimate (even if roughly) the place and energy of the possible future events.

Procedures of seismic hazard assessment are based on ground motion attenuation relationships which can be derived from statistical analyses of recorded ground motions or, in conditions of limited strong-motion records, can be obtained from the available literature sources. Ideally, these attenuation models should consider regional earthquake source and propagation path effects and local site response peculiarities. At present, these is no doubt that the relationships are different for different seismic regions, and “region, and site-specific” models should be developed. However, the empirical databases usually consist on the recording which were obtained during the events of small and moderate magnitudes. Therefore, a question is arising-“is it possible to use these relationships in the case of large events”? The lack of strong motion data close to large earthquakes results in using of several seismological models. An important criterion for any simulation procedure is its validation against recorded ground motion. Every strong earthquake provides an unique opportunity both to verify the accepted attenuation models, and to update empirical relationships.