Tsunami hazard in the Marmara Sea (Turkey): a numerical approach to discuss active faulting and impact on the Istanbul coastal areas He ´le `ne He ´bert a, * , Franc ¸ois Schindele ´ a , Yildiz Altinok b , Bedri Alpar c , Cem Gazioglu c a Laboratoire de De ´tection et de Ge ´ophysique, CEA, BP 12, 91680 Bruye `res-le-Cha ˆ tel, France b Faculty of Engineering, Department of Geophysical Engineering, Istanbul University, Turkey c Institute of Marine Sciences and Management, Istanbul University, Vefa, Istanbul 34470, Turkey Accepted 8 November 2004 Abstract The westward propagation of the seismic ruptures along the North Anatolian Fault (NAF) during the 20th century has increased the probability that the next rupture will be located offshore, in the Marmara Sea, in the prolongation of the 1999 Izmit earthquake faulting. Historical tsunamis have been evidenced in the Marmara Sea, e.g., those associated with the 1509, May 1766 strong earthquakes that broke submarine parts of the NAF, in the vicinity of Istanbul, and future submarine events are expected to produce tsunamis as well that could be triggered either by submarine coseismic displacements or by landsliding. The assessment of their amplitude and extent of their effects is discussed in this paper through numerical modeling based on both earthquake and landsliding sources. We use a finite difference method that yields the maximum water levels reached in the Marmara Sea and the effects (run-up values) around the coasts of Istanbul, using a set of three nested bathymetric grids with a decreasing cell size to focus on Istanbul. First, we consider probable earthquake sources along the northern branch of the emerged NAF, namely, in the eastern (C ¸ inarcik Basin; magnitude Mw 7.2), in the western (Tekirdag and Central Basins; magnitude Mw 7.5) and along the whole branch (magnitude Mw 7.6). The impact obtained in Istanbul is all the more important as a normal component in the Eastern Basin is present, and maximum water levels can reach 2 m along the shoreline, provoking locally significant inundations. These results show that the historical tsunami observations around Istanbul (1509, May 1766), provided they are directly due to earthquake sources, imply a significant normal component in the Eastern Basin whatever the extent of the rupture westward. Conversely, a source located in the Western and/or Central Basin does not provoke large waves in Istanbul (e.g., August 1766 most probably). Finally, we test three landslide sources in the southern vicinity of Istanbul and show that the damage is expected to be larger than in the case of earthquake sources for moderate to large volumes involved. This kind of source, in addition to the coseismic 0025-3227/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.margeo.2004.11.006 * Corresponding author. E-mail address: helene.hebert@cea.fr (H. He ´bert). Marine Geology 215 (2005) 23 – 43 www.elsevier.com/locate/margeo