Ground motion prediction with the empirical Green's function technique: and assessment of uncertainties and confidence level

摘要

An assessment of uncertainties for ground motion predictions with the aid of the empirical Green's function (EGF) technique is presented. The main input parameters were identified, and their respective uncertainties were assessed by means of an international expert inquiry. The repercussion of these 'input uncertainties' on the final ground motion estimates were investigated by means of the Latin Hypercube Sampling technique. The mean ground motion estimates (response spectra) and their standard deviations were compared with results obtained from empirical 'attenuation laws'. The most sensitive input parameter turned out to be the seismic moment corresponding to the EGF. In general, if the source parameters are well determined, equivalent uncertainties, statistically speaking, can be expected from the EGF technique and from the application of attenuation laws. Therefore, if EGFs with well known source parameters are available, the EGF technique seems to be preferable: site effects are automatically taken into account, and physically realistic acceleration time histories can be obtained. However, further investigations on the reliability of the EGF technique should be performed, and finally, it is recalled that the EGF technique is based on the assumption of linearity. If conditions are such that this assumption cannot be maintained, the EGF technique should be combined with non-linear geotechnical methods.