Seismic risk in Turkey, the Aegean and the eastern Mediterranean: the occurrence of large magnitude earthquakes

• Published in: Geophysical Journal of the Royal Astronomical Society Vol. 78; no. 2; pp. 475 - 506
• Main Authors: Burton, Paul W., McGonigle, Robert, Makropoulos, Kostas C., Üçer, S. Balamir
• Format: Journal Article
• Language: English
• Published: Oxford, UK Blackwell Publishing Ltd 01-08-1984
• ISSN: 0956-540X; 0016-8009; 1365-246X
• DOI: 10.1111/j.1365-246X.1984.tb01961.x

The file of Turkish seismicity developed by Kandilli Observatory, Istanbul, for earthquakes to 1970 is extended here up to 1978 using 1SC and PDE data. Entries into this file are maintained on the surface wave magnitude scale Ms, and conversion of body wave magnitude mb to Ms has been carried out where necessary using a formula derived for Turkish earthquakes. Completeness analysis suggests that magnitudes Ms> 4.5 may be used for statistical evaluation of seismic risk. This file is analysed by a range of methods to provide a suite of risk forecasts. Forecasting results from least squares and maximum likelihood estimates of the whole process Gutenberg-Richter cumulative frequency law of earthquake magnitude occurrence, and from the part process of Gumbel's first extreme value distribution, all show small systematic differences in forecasts, but all three methods lead to magnitude forecasts which fall well within the range of standard deviation. However, these forecasts are obtained by subdividing Turkish seismicity in a cellular manner, and many of these cells show curvature of the earthquake frequency magnitude distribution curve concave with respect to the origin: Gumbel's third asymptotic distribution of extreme values is chosen as an appropriate statistical description. Approximate upper bounds ω to earthquake surface wave magnitude occurrence are evaluated and estimates of largest magnitudes expected over an interval of 75yr are forecast with uncertainties. Values of ω are asymptotic, uncertain, and theoretically correspond to infinite return periods. Strain energy release diagrams are then invoked to estimate empirically the large magnitude M3 which is equivalent to the total strain energy which may be accumulated in a region. This equivalent magnitude M3 is consistently less than ω and there is a finite ‘waiting time', typically ranging from about 15 to 70 yr, during which the energy equivalent to M3 may be accumulated. Combination of the Gumbel III earthquake occurrence statistics for each cell with an inferred intensity attenuation law leads to a suite of perceptibility curves which give the probability of perceiving specific intensity levels from each possible earthquake magnitude up to the local upper bound magnitude ω. This family of curves generated for an individual cell is seen to be nested, and shows a peak probability for each intensity level which typically occurs at similar magnitudes defined as the ‘most perceptible’ earthquake. These ‘most perceptible’ earthquakes show values which range from an Ms of about 5.5 to 7.5 for regional seismicity cells in Turkey; local values may be used as a criterion for choosing engineering design time histories. The seismic risk parameters from the cellular analysis of seismicity are interpreted as contoured seismic risk maps. These maps show that contoured values of the maximum strain energy earthquake M3 usually exceed the 75 yr earthquake by about one-half magnitude for similar geographic locations, but there is overall similarity in contour shape and distribution of highs of seismic risk. When regional variations in attenuation are included for calculations of perceptibility the overall contour shape changes slightly, and the net effect is to emphasize the relative risk associated with the North Anatolian Fault zone. The features of the contoured perceptibility map are compatible with the existing Earthquake Zoning Map of Turkey which is entirely based on observed felt effects in Turkey.