GNGTS 2013 - Atti del 32° Convegno Nazionale

produced by the destructive earthquakes occurred in the last decade. Strong shocks have hit zones where the expected shacking was relatively low (e.g., 2003, Algeria and 2004, Morocco). Moreover, hazard maps elaborated after strong shocks often show a considerably higher hazard with respect to the maps drawn before the shock (e.g., 2010, Haiti). For the 2010 and 2011 earthquakes in the South Island of New Zealand, the observed ground motion (PGA up to 2.2 g) largely exceeded the values predicted by PSHA maps. Finally, the 2011 Tohoku (Japan) earthquake hit a zones that in the official seismic hazard map was not indicated as particularly risky (e.g., with respect to the Tokai zone located southwest of Tokyo metropolitan area). The PSHA approach implies the use of modelled seismic sources, defined in terms of location, geometry and maximum expected magnitude. However, for areas such as the Italian region most of available data on seismic history are represented by macroseismic (not instrumental) information. This fact may explain the discrepancies between hazard maps, expressed in term of expected MCS intensity, elaborated by different methodologies (PSHA and site approach, e.g., Gomez Capera et al. , 2010 and references therein). As discussed above, the PSHA is based on a series of operative steps, each needing a number of choices about model parameters. This implies the intervention of an operator, whose personal knowledge and belief may significantly affect the final result. Indeed, elaborations independently carried out by distinct research groups may predict very different levels of ground shaking for the same site (e.g., Stucchi et al. , 2011; Stein et al. , 2012). The ground motion predicted by probabilistic hazard maps refers to the behaviour of stiff soil or hard rock. However, it well known that local structural, stratigraphic and topographic features (detectable by accurate geological and geophysical survey only) may considerable affect the amplitude of seismic shacking. As occurred for the 2009, L’Aquila and 2012, Emilian Fig. 1 – Seismogenic zones of northern and central Italy, as defined by the ZS9 seismic zone model which has been adopted to compute seismic hazard by the PSHA approach (Gruppo di Lavoro MPS, 2004). The regional borders of Emilia-Romagna, Toscana, Umbria and Marche are shown in blue. Squares indicate the epicenters of shallow earthquakes (modified after Meletti et al. , 2008). 476 GNGTS 2013 S essione 2.3