Student: Luis Fernando Restrepo Velez
Supervisor: Dr. J.J. Bommer
Attenuation relationships for the estimation of strong-motion parameters for a particular earthquake scenario are fundamental to seismic hazard assessment. These relationships are obtained from regression analysis on recorded values of the parameter of interest and they generally have a relatively large degree of scatter that is usually characterized by the standard deviation of a lognormal distribution of the residuals. The scatter in the attenuation relationship exerts a pronounced influence on the results obtained from seismic hazard assessments.
In disaggregation of PSHA to determine hazard-consistent design earthquake scenarios, the scatter, defined by a certain number of standard deviations above the median, becomes part of the scenario together with the magnitude and distance. In current disaggregation practice the number of standard deviations in the design scenario is added to the median values of all of the ground-motion parameters that are required without consideration of how likely it is that the ground motion will yield such high values of all the parameters simultaneously.
This project explores the nature of the scatter in attenuation relationships and shows that the scatter is not genuinely represented by the lognormal distribution of the residuals, particularly for the most extreme outliers. The covariance of the scatter is discussed, along with the existence of physical upper
bounds for strong-motion parameters.
You may download (1900 kB) a digital version of this MSc dissertation here.