GNGTS 2017 - 36° Convegno Nazionale

A reappraisal of the 1783 Scilla landslide-tsunami: numerical techniques for hazard assessment F. Zaniboni, G. Pagnoni, S. Tinti Dipartimento di Fisica e Astronomia, Alma Mater Studiorum – Università di Bologna, Italy The reconstruction and simulation of tsunami historical events by means of numerical techniques aims at two main purposes: on one side, the availability of reports and measures can contribute to create a set of observed data useful for the comparison with the simulated result. When such data are complete and detailed both before and after the event, they can constitute a valid benchmark to assess the adequacy and the accuracy of the methodology adopted. On the other side, such reconstruction can provide precious information on the potential hazards that can affect a certain coastal stretch, especially in areas where anthropization has increased consistently in the last decades, increasing the vulnerability and exposures to hazardous coastal phenomena. Atragic event resuming all these considerations is surely the landslide-generated tsunami that hit the coasts around Scilla (on the Tyrrhenian coast of Calabria) during the night of 6 February 1783. In that period, a well-known and extensively studied seismic sequence interested southern Calabria (Jacques et al. , 2001), starting from the Mw=7.0 earthquake of February 5 (CPTI11 catalogue; Rovida et al. , 2011). Many villages were destroyed and about 30,000 people died (Guidoboni et al. , 2007) during the whole series of earthquakes. After the strong earthquake of February 5, most of the people living in Scilla had gathered in the beach of Marina Grande to escape building collapses; during the night between 5 and 6, an aftershock mobilized a huge mass 1 km west of Scilla, probably already destabilized by the main event. The sliding body crashed into the sea and generated a tsunami that within 1 minute hit the Marina Grande beach, causing 1,500 casualties and severe building damages (Graziani et al. , 2006). Some surveys were carried out in the aftermath of the event, describing accurately the catastrophic consequences of the waves and the characteristic of the tsunami source. As concerns the tsunami effects, they were noticed and reported for a stretch of about 40 km along the Calabria coast, between Nicotera and Reggio Calabria. Apart from the disastrous effects in Scilla, where the run-up reached around 9 m, the most evident consequences were found on the other side of the Messina Straits, about 10 km far, in the Sicilian area of Capo Peloro. Here the sea penetrated by hundreds of meters, sweeping people and small houses. Also in the Messina harbor, 20 km southward, some effects were reported. With regards to the tsunami source, its traces are visible even today: there is a large, 500 m wide, scar along the seaward flank of Mt. Pacì, starting from around 400 m a.s.l. and extending downslope also under the sea. Recent submarine investigations have described some underwater deposits, at about 300 m below sea level, characterized by huge blocks that have been associated to the 1783 landslide. Their volume ranges 3 Mm 3 , covering an area of about 1 km 2 (Bosman et al. , 2006). The morphological evidence allows for a good reconstruction of the sliding body: in the most plausible hypothesis, the mass reaches almost 6 Mm 3 , with the toe located at about 100 m a.s.l., meaning then a purely subaerial initial collapse. This is the scenario explored in the work by Mazzanti et al. (2011), that was the first published attempt to apply numerical simulation to the 1783 landslide and tsunami. In the work by Zaniboni et al. (2016), the same hypothesis is assumed for the tsunami source. The effects of the generated waves are evaluated in the near field, i.e., around Scilla, with a detailed reconstruction of the coastal morphology and a very good agreement between the simulated and observed run-up and inundation. The numerical codes were developed by the University of Bologna Tsunami Research Team. The landslide motion is computed through the code UBO-BLOCK1, that considers the sliding body divided into a set of interacting deformable blocks. In this way, the complete time history of the shape changes and of the dynamics of the landslide is obtained (for further details see Tinti et al. , 1997). Such data provide the GNGTS 2017 S essione 2.1 269