Discussion on uncertainties, attenuation of ground motion and aseismic design criterion

The usually calculated exceedance probability curves of ground motion are reduced as cumulative probability curves of normal distribution on some assumptions in this paper. These curves clearly displayed the variation of curve shape with the variance of attenuation relation and its physical implication. Similar investigation is also made on uncertainties of hypocenter locations and magnitudes of future earthquakes. The larger the uncertainty, the flatter the exceedance probability curve is. In a broad sense, the less people know future earthquakes, the flatter is the curve. Due to the rich or poor data sets available, the uncertainties of attenuation relation are different from region to region. The acceleration attenuation relations in these regions with no enough strong earthquake records can be obtained by conversion from other region, but with larger uncertainty. The uncertainty contains systematic difference between the two regions in addition to common stochastic error. A method to check and adjust the converted attenuation relation by using the local data is proposed in this paper. If the uncertainty of attenuation relation is too large, the result of seismic hazard assessment may be unaccepted sometimes. In order to realize that the structures do not collapse in large earthquakes, be repairable under moderate ones and without destruction in small earthquakes, this paper suggests that it may be reasonable to get the first and the third levels of aseismic design parameters by some empirical relation on the basis of the second level not by fixing the risk probability levels (63%, 10% and 3%). A particular aseismic design criterion is not only a balance between economic cost and seismic risk but also suited to the human’s knowledge of nature.