| Popis: |
As with previous discussions submitted by the lead author of this comment (Wang, 2005), Wang and Zhou (2007) are not really commenting directly on the contents of our review article (Bommer and Abrahamson, 2006) but are using the opportunity to present a fundamental objection to probabilistic seismic-hazard analysis (PSHA). However, because the objective of our article was to clarify a fundamental aspect of PSHA that is frequently handled incorrectly in current practice, we welcome the opportunity to also respond to the arguments put forward in this comment. These arguments against the basic formulation of PSHA are likely to appear in print somewhere at some point, and we believe it is useful to explain why they are erroneous. Before addressing the core argument of Wang and Zhou (2007), it may be useful to clarify the issue of the dependence of the ground-motion variability on magnitude and distance. Most ground-motion prediction (attenuation) equations produced up until about 10 years ago modeled the variability as being homoscedastic, which simply means that the standard deviation ( σ ) of the lognormal residuals is constant for all values of magnitude and distance. The variation of σ with magnitude—called heteroscedastic variability—is discussed by Youngs et al. (1995) and has been found in many strong-motion prediction equations (e.g., Abrahamson and Silva, 1997; Campbell, 1997; Campbell and Bozorgnia, 2003; Ambraseys et al. , 2005; Akkar and Bommer, 2007a). Whether or not the apparent dependence of σ on magnitude is genuine or an artefact of the datasets that have been used for regressions, however, is not the main issue … |
