Earthquake hazard parameters in Crete Island and its surrounding area inferred from Bayes statistics: An integration of morphology of the seismically active structures and seismological data

• Published in: Pure and applied geophysics 2003-08, Vol.160 (8), p.1517-1536
• Main Authors: TSAPANOS, Theodoros M, CHRISTOVA, Cenka V
• Format: Article
• Language: English
• Subjects: Bayesian analysis; Data analysis; Earth sciences; Earth, ocean, space; Earthquakes; Earthquakes, seismology; Engineering and environment geology. Geothermics; Exact sciences and technology; Geophysics; Hazards; Internal geophysics; Islands; Morphology; Natural hazards: prediction, damages, etc; Parameter estimation; Seismic activity; Seismic engineering; Seismic hazard; Seismic phenomena; Seismology; Spatial distribution; Statistics
• ISSN: 0033-4553; 1420-9136
• DOI: 10.1007/s00024-003-2358-4

-- The study addresses the evaluation of earthquake hazard parameters such as maximum regional magnitude (M^sub max^) and the slope of Gutenberg-Richter law β (where b=β log e) for the Hellenic Wadati-Benioff zone and the overriding lithospheric plate in the area of Crete and its surroundings. The seismicity of the area is divided in a cellular (1.0° × 1.0°) manner allowing analysis of the localized earthquake hazard parameters and graphical representation of their spatial variation. Our approach incorporates the recently updated earthquake catalogue for Greece and the adjacent areas, the consideration of the morphology of the deep seismically active structures in the studied area and use of a probabilistic procedure for estimating the earthquake hazard parameters.¶One of the main inconsistencies in the earthquake hazard assessment is the estimation of the maximum magnitude and the related uncertaint y. The Bayesian approach, applied in the present, is a straightforward technique for evaluating the earthquake hazard parameters and is based on the following assumptions: Poissonian character of seismic events flow, a frequency-magnitude law of Gutenberg-Richter's type with cutoff maximal value for estimated parameter and a seismic catalogue, having a rather sizeable number of events (i.e., 50 events at least per cell). For five cells in which the number of events is less than 50, an effort is made to produce synthetic data. The re-assessed parameters obtained from the synthetic data show no significant difference and the real data (of the five cells) are finally taken into account although the estimated uncertainty is high.¶For four random cells we constructed hazard curves showing the probabilities that a certain magnitude M will be exceeded in one year and the return periods (in years) that are expected for a given magnitude. These are particularly useful for the mapping of earthquake hazard in regions of either low or high seismic activity, as is Crete and the adjacent area.¶The obtained results show that the W and E parts of both subducting and overriding plates differ in the spatial distribution of all the estimated earthquake hazard parameters. The M^sub max^ distribution indicates strong coupling between the western portions of the interacting plates (M^sub max^ > 6.3) to the south of 36°N. The smaller values of M^sub max^ (M^sub max^ < 6.3) estimated in the SE part of the studied area indicate weak coupling between the eastern portions of the subduct