Seismic hazard assesment and zoning in Java: new and alternative probabilistic assessment models

Burton, Paul W. and Weatherill, Graeme and Karnawati, Dwikorita and Pramumijoyo, Subagyo (2008) Seismic hazard assesment and zoning in Java: new and alternative probabilistic assessment models. In: International Conference on Earthquake Engineering and Disaster Mitigation 2008, Indonesia.

Full text not available from this repository.

Abstract

Applying alternative and different approaches to seismic hazard assessment is instructive. It allows learning from the different outcomes of the different approaches. These outcomes may be mutually reinforcing or diverge, suggest further study and research is needed, or provide new insights into old problems. Herein Java island-scale seismic hazard will be considered by applying different probabilistic approaches to hazard assessment. Results from two distinct methods are provided for Java: 1) primary zoning using K-means partitioning of seismicity into spatial clusters (progressed into zones) which are then developed into seismic hazard maps using Monte Carlo earthquake catalogue simulation, and 2) extreme value analysis applied at a matrix of points throughout a zoneless Java. The latter approach has been used before, the former adopts seismicity partitioning into spatial clusters prior to Monte Carlo modelling and is novel. The earthquake catalogue analysed is NEIC (1973-2006). This catalogue is homogenised to the moment magnitude scale Mw and Poisson declustering of fore- and after-shocks applied. The completeness threshold is around 4.9 Mw . Shallow earthquakes down to 80 km depth contribute most to the hazard and are partitioned into 1 to K trial clusters of seismicity by minimising the total within cluster distance from seed centroids. Repeated trials produce an optimum partition. A variety of indices can be invoked to try to quantify cluster quality for a given K; in addition to this, it is decided to seek the best value of K by testing the influence of K on ensuing seismic hazard analyses. Monte Carlo synthesis generates synthetic catalogues for each K value, from which peak ground acceleration (PGA) hazards are calculated and compared against results from the observed catalogue to choose acceptable K values. To summarise the results, seismic hazard maps are constructed for two acceptable values of K (8 and 27) for Java from the Poisson declustered catalogue of shallow earthquakes using the Boore, Joyner, Fumal attenuation law. Not surprisingly the smaller value of K with 8 clusters (progressed to zones) produces the smoother hazard map. All of the maps indicate highest hazard around the Sunda Strait and a general expectation in Java Island of 100-300 cm/s^-2 with one-in-ten chance of exceedance in 50 years.

[error in script]
Item Type: Conference or Workshop Item (Paper)
Subjects: Methodology > Method and procesing > Collective properties of seismicity > Clustering and migration
Methodology > Method and procesing > Probabilistic seismic hazard analysis - stationary > Aggregated solution
Project: IS-EPOS project