GNGTS 2018 - 37° Convegno Nazionale

298 GNGTS 2018 S essione 2.1 experimental approaches such as: (1) the statistical treatment of potential seismic sources, combining all the available information (seismicity, moment tensors, tectonics), and considering earthquakes occurring on all possible crustal faults and on subduction interfaces; (2) an intensive computational approach to tsunami generation and linear propagation across the sea up to an offshore fixed depth; (3) the use of approximations for shoaling and inundation, based on local bathymetry (Glimsdal et al. , 2017; Selva et al. , 2017); and (4) the exploration of several alternatives for the basic input data and their parameters which produces a number of models that are treated through an ensemble uncertainty quantification (Selva et al. , 2016), which includes a formalized elicitation (Selva et al. , 2017, TSUMAPS-NEAM TI Team, 2017). In New Zealand, as well as in Italy, the tsunami warning evacuation zones are based on S-PTHA, with a fixed average recurrence time, and a fixed level of epistemic uncertainty. The latter two choices tend to make the risk-reduction planning spatially homogeneous, i.e. with a homogeneous level of residual risk. In the last 5 years, DPC, INGV and ISPRA have been working to set-up the national tsunami early warning system. From the TSUMAPS-NEAM SPTHA, the tsunami intensity with an average return period (ARP) of 2500 yr at the 84th percentile of the epistemic uncertainty (Fig. 1) is being used as a basis for evacuation maps. These choices (ARP and percentile) for the intensity were established by a Decree recently signed by the Head of the Civil Protection Department (an implementing act of the SiAM Directive; DPCM, 2017), based on the definition of the acceptable risk level and by considering Fig. 1. A close-up view of TSUMAPS-NEAM model results in Italy (ARP 2500 yr, 84 th percentile of the epistemic uncertainty).