%0 Journal Article
%@ 0148-0227
%A Parsons, Tom
%A U.S. Geological Survey, Menlo Park, California, USA,
%D 2004
%F epos:1437
%I American Geophysical Union
%J Journal of Geophysical Research
%N B5
%P 1-21
%T Recalculated probability ofM≥ 7 earthquakes beneath the Sea of Marmara, Turkey
%U https://episodesplatform.eu/eprints/1437/
%V 109
%X New earthquake probability calculations are made for the Sea of Marmara region and the city of Istanbul, providing a revised forecast and an evaluation of time-dependent interaction techniques. Calculations incorporate newly obtained bathymetric images of the North Anatolian fault beneath the Sea of Marmara [Le Pichon et al., 2001; Armijo et al., 2002]. Newly interpreted fault segmentation enables an improved regional A.D. 1500–2000 earthquake catalog and interevent model, which form the basis for time-dependent probability estimates. Calculations presented here also employ detailed models of coseismic and postseismic slip associated with the 17 August 1999 M = 7.4 Izmit earthquake to investigate effects of stress transfer on seismic hazard. Probability changes caused by the 1999 shock depend on Marmara Sea fault-stressing rates, which are calculated with a new finite element model. The combined 2004–2034 regional Poisson probability of M ≥ 7 earthquakes is ∼38%, the regional time-dependent probability is 44 ± 18%, and incorporation of stress transfer raises it to 53 ± 18%. The most important effect of adding time dependence and stress transfer to the calculations is an increase in the 30 year probability of a M ≥ 7 earthquake affecting Istanbul. The 30 year Poisson probability at Istanbul is 21%, and the addition of time dependence and stress transfer raises it to 41 ± 14%. The ranges given on probability values are sensitivities of the calculations to input parameters determined by Monte Carlo analysis; 1000 calculations are made using parameters drawn at random from distributions. Sensitivities are large relative to mean probability values and enhancements caused by stress transfer, reflecting a poor understanding of large-earthquake aperiodicity.