New Perspectives in Developing a Ground Motion Prediction Equation for the Low-Seismic Region of Australia Using a Hybrid Approach

Abstract eng:
Ground motion prediction equations (GMPEs) are fundamental tools in conducting both deterministic and probabilistic seismic hazard assessments. These equations are developed to predict the intensity of ground shaking for a given earthquake, particularly the combination of magnitude and source-site distance. GMPEs are furthermore central in developing earthquake loading codes used for the design of large engineered structures. A major revision of Australia’s National Seismic Hazard Assessment is currently underway. It is therefore timely in developing a new hybrid based approach to GMPE, particularly important for Australia that experiences low to moderate seismicity and lacks vast amounts of strong motion records required in an empirical approach. A key aspect of any GMPE is the attenuation behaviour. Attenuation behaviour of ground shaking due to an earthquake is a complicated interaction between rupture propagation, direction and energy release, as well as material it passes through. These basic properties can be modelled to form a GMPE for a given region, mechanism and method of modelling. Typically, such attenuation factors can be classed into regional (properties of seismic waves generated at source), local (extent of amplification and attenuation) and site (filtering mechanics of the bedrock layers) factors. In low to moderate seismic regions, most GMPEs use any available data, but also heavily rely on stochastic approaches to generate sufficient data for modelling. In Australia three existing GMPEs have been developed; the Liang model [1] specifically for south-western Western Australia; Somerville models [2], one for Cratonic Australia and the other for NonCratonic Australia; and Allen model [3] for south-eastern Australia. These models use a stochastic approach and therefore lack empirical data. Another approach applied in Australia is the Component Attenuation Model (CAM). CAM is a framework by which a generalised attenuation model is derived from stochastic data of seismological properties rather than recorded earthquake data. This model comprises a series of component factors that represent effects of the source, wave travel path and material it passes through. The CAM technique has been successfully compared with real earthquake data from PEER strong-motion databases, as well as pilot case studies in parts of Australia and Southeast Asia. The purpose of this paper is to model the GMPEs using a hybrid GMPE approach (investigated by Campbell [4]) as well as CAM approach, then compare Australian GMPEs and PEER-NGA models with available Australian recorded data. Discussion focusses on the models with better comparisons to Australian data and explores the underlying assumptions or requirements used in each case. Analysis shows that the attenuation is over estimated by the models in the 10 – 100 km range and the 5% damped acceleration spectra are inconsistent in the 0.3, 0.5 and 1second periods. As this range of distance and period of structures are highly crucial for major Australian cities there is a requirement to produce an improved GMPE model for Australia.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket