Mathematical modelling of a potential tsunami associated with a late glacial submarine landslide in the Sea of Marmara

Abstract

Potential tsunami waves were modelled on the basis of the morphology and geological setting of a late glacial submarine landslide localized in the north-eastern sector of the Sea of Marmara, using a three-dimensional algorithm with the purpose of assessing the future risk of tsunamogenic landslides in the region. The landslide occurred off the Tuzla Peninsula on the north-eastern slope of the Cinarcik Basin, the easternmost of the three deep Marmara basins. The mass movement appears to be related to the Main Marmara Fault that passes below the toe of the failed mass. Observations from earlier manned submersible dives suggest that the initiation of the slide was facilitated by secondary faults associated with the Hercynian orogeny and involved Palaeozoic shales dipping southwards towards the deep basin. Radiocarbon dating of core material, together with the well-dated Marmara sapropel above the chaotically mixed landslide surface, reveal that the latest landslide event occurred about 17 14C ka b.p. The uppermost scar of the landslide is found at 250 m and its toe at about 1,200 m below the present sea level. At the time of the slide, the Marmara Sea Basin was lacustrine, with its water level at −85 m. In plan view the landslide has a distinctively triangular shape and the lateral extent of its toe is about 10 km. Multibeam bathymetric data indicate that the sliding motion probably occurred in two phases: a slower phase affecting the eastern part, characterized by an undulating surface, and a more rapid phase affecting the western part that possibly created tsunami waves. In the seismic sections, older failed slide masses can be clearly identified; these were probably displaced during marine isotopic stage 6 (∼127–160 ka b.p.). The front of this buried material is located more than 1.5 km further south of the fault. We used a three-dimensional, Green’s function-based potential theory approach, rather than shallow-water equations commonly used in conventional tsunami simulations. The solution algorithm is based on a source-sink formulation and an integral equation. The results indicate that the maximum height of the tsunami in the Cinarcik Basin could have reached about half the average thickness of the sliding mass over a lateral extent of 7 km. Assuming an average thickness of 30 m for the landslide, and considering that the water level at 17 ka b.p. was at about −85 m, the modelling shows that the maximum wave height generated by the slide would have been about 15–17 m.