GNGTS 2013 - Atti del 32° Convegno Nazionale

was “expected” and such reduced effects on building structures can also be attributed to the last ten year policy of prevention and vulnerability reduction performed on the most hazardous portions of Tuscan territory. To this end, Tuscany Region Office for Seismic protection has been financing a project for estimating local seismic effects, called VEL project since 1998. It was a pioneering project accomplished by some good working strategies for microzoning studies that are still up to date and showed their efficiency where applied. Hereafter, some good practices for Italian seismic zonation are briefly reported from the writing authors’ experience in VEL projects as far as from international microzoning activities, according to the following points: 1) maximum intensity maps to be used for a preliminary zonation of the most hazardous areas; 2) integrating multidisciplinary experimental techniques to seismic characterization of sites. Point historical approach versus continuous attenuation laws. Since 1976 after Friuli earthquake, when the first microzoning studies were performed, it was quite evident that, based on historical catalogues of the seismic events, the strong earthquakes occur where they occurred in the past. Although different Magnitudes can be felt at different return periods, Signanini et al. (1983) showed that the seismic local amplification can increase the felt intensity up to 2-3 intensity degrees of MCS. After more than ten years, Favali et al. (1995), Midorikawa (2002) and lately after the 2012 Emilia Romagna earthquake, Paolini et al. (2012) focused on the maximum felt intensity maps derived by complete historical seismic catalogues as the key tool for guiding the choice of microzoning studies within the most hazardous areas in urbanized territories. Such a selection is needed because, especially in Italian territory, money is limited whereas the whole national territory is affected by earthquakes. Thus, microzoning surveys in near field areas shall be performed in those territories that suffered repeated destructions and fatalities in the past, according to the maximum intensity maps, such as the one from Boschi et al. (1995) (Fig. 2a, black circles). This latter was drawn based on thorough analysis of historical documents on past earthquakes from the years 1 BC to 1992 and it shows the expected intensity higher than VI degree in Mercalli Cancani Sieberg (MCS) scale. This choice on the intensity degree is commonly used to highlight the areas where seismic events caused from severe damages to collapse of urbanized environment (from IX to XI). As can be noted, this map shows limited areas irregularly shaped that suffered partial or complete destructions. Based on such a map, the recent “unexpected” strong events of L’Aquila and Emilia Romagna earthquakes could have been predicted (compare Fig. 1a, c with Fig. 2a). The reason of the misleading information from Italian seismic hazard map is related to the prediction of the ground motion amplitude at a site through the peak ground acceleration (PGA) attenuation laws. The estimation of these ground motion prediction equations (GMPE), whether or not truly representative of seismic shaking, is affected by too much uncertainties related to 1) the models used for seismogenic zonation of the national territory and 2) the interpolation of mean trend in PGAs databases from national and worldwide records. These uncertainties dramatically increase when the PGA estimation is referred to near field areas. This is evidence from recent studies by Faccioli (2010) shown by Fig. 2b: here, the uncertainty in predicting the GMPE based only on Italian Strong motion database (ITACA) (Luzi et al. , 2008), recorded on stiff soil V S30 =900m/s. As can be noted, in near field areas (up to 20 km far from the fault R f ), the high scatter in recorded PGA values suggest: 1) not to rely on attenuation laws and 2) that V S30 does not identify the “seismic soil types” with respect to local amplification in PGA values. Recently, Faccioli (2013) back-analyzed his GMPE by the Emilia Romagna recorded PGAs. He stated the validity of GMPEs derived only from local seismic event databases especially for hypocentral distance higher than 20 km. Thus, is it realistic to consider that PGA values continuously decrease, in the first 10-20 km far from the seismic source? The records tell us a different story (Galli and Camassi, 2009; Lanzo et al. , 2010; Bergamaschi et al. 2011; Di Giulio et al. , 2011): PGAs are distributed spot-like nearby the seismic source increasing abruptly where soft sediments lie on stiff bedrock and subsurface geometries as basin shapes are filled by soft sediments (Fig. 2c). 327 GNGTS 2013 S essione 2.2