Probabilistic spectral seismic hazard assessment for Italy

BOLLETTINO DI GEOFISICA TEORICA ED APPLICATA VOL. 40, N. 1, pp. 31-51; MARCH 1999 Spectral probabilistic seismic hazard assessment for Italy A. REBEZ (1), L. PERUZZA (2) and D. SLEJKO (1) Osservatorio Geofisico Sperimentale, Trieste, Italy C.N.R. Gruppo Nazionale per la Difesa dei Terremoti at O.G.S., Trieste, Italy (1) (2) (Received January 25, 1999; accepted March 26, 1999) Abstract. During the last decade the Italian "Gruppo Nazionale per la Difesa dai Terremoti" has undertaken a project for assessing seismic hazard on probabilistic bases in the national territory, to be used as scientific background for the revision of the current seismic zonation; the seismic hazard was expressed there in terms of peak ground acceleration and macroseismic intensity. A more detailed analysis is now performed. Probabilistic spectral seismic hazard maps at different frequencies of engineering interest demonstrate that the soil condition is a first order factor influencing the level of the expected shaking, with an average increase of about 0.2 g for the prediction related to a 475-year return period, when passing from rock to soft soil. The attenuation model similarly conditions the results, but in a more magnitude-dependent fashion. Here, the two relationships selected account for the effects of strong rare seismicity and moderate frequent earthquakes differently: differences going from 20% to 80% of the predicted spectral accelerations are, therefore, model-dependent. The uniform hazard response spectra for some major Italian towns are then compared to the elastic response spectra of the present Italian seismic zonation, indicating that two important towns such as Rome and Naples are not adequately represented by the present code. The effective peak acceleration is considered, finally, a good synthetic parameter for representing seismic hazard in most engineering applications. 1. Introduction A global project for seismic hazard assessment on Italian territory was defined in the framework of the “Gruppo Nazionale per la Difesa dai Terremoti” (GNDT) for updating the national seismic zonation. The project consisted in three main objectives: compilation of an earthquake catalogue and a seismological data base, preparation of the seismogenic zone (SZ) map, and hazard assessment by probabilistic methodologies. The final results were obtained and officially presented in the Corresponding author: A. Rebez; Osservatorio Geofisico Sperimentale, C.P. 2011 Trieste, Italy; tel. +39 040 2140250; fax +39 040 2140266; e-mail: [email protected] © 1999 Osservatorio Geofisico Sperimentale 31 Boll. Geof. Teor. Appl., 40, 31-51 REBEZ et al. summer of 1996 to the Civil Protection Department, which financed the project. The results will be considered by the Ministry of Public Works for use in legislation. The hazard assessment was done according to the Cornell (1968) approach by applying the Seisrisk III (Bender and Perkins, 1987) code; the details of the computation are exhaustively described in Slejko et al. (1998), and the final maps, in terms of peak ground acceleration (PGA) and macroseismic intensity, are reported there. Even if PGA is the most widely used parameter in seismic hazard analysis for its easy and practical meaning nevertheless, it is, a rough shaking indicator because it is generally associated with a short impulse of very high frequency and, therefore, cannot be easily correlated to the damage observed. For this reason the majority of building codes define the elastic response spectrum and adopt the design spectrum to represent seismic actions. Examples of analyses in terms of uniform hazard response spectrum, in Italy, are rare (see, for example, Peruzza et al., 1994, where the extreme theory statistics is applied in northeastern Italy), and only recently have they been performed for the entire country and neighbouring areas (e.g. Romeo and Pugliese, 1997; Peruzza et al., 1998a, 1998b). The aim of this paper is to present spectral seismic hazard maps, calculated for the Italian territory using the Cornell (1968) method, and uniform hazard response spectra for some major towns. The basic PGA map is presented in Fig. 1: it differs from the official one presented to the Civil Protection Department in 1996 (Slejko et al., 1998) as the attenuation relation has been substituted with a more recent version (Ambraseys et al., 1996) which takes into account different soil types and is related to a spectral version as well. The maximum values of PGA in Fig. 1 are reached, starting from the south: in the Calabrian Arc (north of Reggio Calabria), in the southern Apennines near Potenza and less extensively near Campobasso, and in the eastern Alps NE of Venice. In all these areas the PGA values exceed 0.32 g (is a gravity acceleration). Some other areas in the northern and central Apennines, in the zones between Florence and Bologna and between Perugia and L'Aquila, for example, show PGA values that are higher than 0.28 g. The results are generally lower than the previous ones (Slejko et al., 1998) as rocky conditions are mapped here, while the previous relation (Ambraseys, 1995) takes into accounts only average soil conditions: the volcanic areas now show a higher PGA, as no peculiar modifications of the attenuation were considered in this more recent Table 1 - Acceleration values (in g) with a 475-year return period for some main Italian towns. The zonation code indicates the anti-seismic level to which the towns are subjected at present (1 to 3 for decreasing severity, n. c. means not classified). PGA_95 refers to the Ambraseys (1995) attenuation relation, PGA_96 to the Ambraseys et al. (1996) attenuation relation. SA indicates spectral acceleration at different periods. 32 Town zonation PGA_95 PGA_95 PGA_96 PGA_96 SA(0.2) mean+σ mean code mean mean+σ mean SA(0.2) SA(1.0) mean+σ mean SA(1.0) mean+σ Milan Venice Trieste Florence Rome Naples Messina Catania n.c. n.c. n.c. 2 n.c. 3 1 2 0.178 0.213 0.231 0.404 0.353 0.461 0.715 0.591 0.045 0.064 0.058 0.106 0.092 0.146 0.281 0.160 0.039 0.050 0.057 0.131 0.113 0.123 0.227 0.135 0.080 0.099 0.112 0.203 0.168 0.181 0.365 0.240 0.044 0.049 0.052 0.108 0.092 0.135 0.189 0.139 0.074 0.087 0.097 0.169 0.149 0.195 0.313 0.256 0.098 0.112 0.112 0.235 0.192 0.289 0.398 0.292 0.019 0.032 0.024 0.055 0.043 0.081 0.204 0.096 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 8 10 12 14 16 18 A CH AOSTA A S P L 46 PGA PGA (g)(g) 0.5 - -0.6 0.20 0.24 0.6 - -0.7 0.24 0.28 SLO TRENTO 0.7 - -0.8 0.28 0.32 ≥ 1.0 0.32 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r FLORENCE 44 i A F ANCONA a P t E PERUGIA i c N N L'AQUILA 42 I 42 S N ROME e E CAMPOBASSO a S BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 1 - PGA with a 475-year return period for rock. The Ambraseys et al. (1996) attenuation relation with σ was used. analysis. For example, the high spot in eastern Sicily (mount Etna area), near Catania, is largely caused by the attenuation relation which does not hold in volcanic environments. Table 1 shows the PGA differences for a 475-year return period in the eight major Italian towns selected for this study. The differences are limited, considering the different soil conditions as well, less than 15% with the exception of Florence, where the increase is of 20%. Only for Naples and Catania are the present hazard estimates higher than the previous ones and this is because of the influence of volcanic SZ's, 33 Boll. Geof. Teor. Appl., 40, 31-51 REBEZ et al. where the attenuation was modelled differently, as already pointed out. 2. Spectral seismic hazard maps for Italy Maps showing the expected spectral acceleration at different periods can be properly used as the basis for seismic zonation (e. g.; Basham et al., 1997). Spectral attenuation relations useful for Italy are available in literature (e. g.; Pugliese and Sabetta, 1989; Tento et al., 1992; Sabetta and Pugliese, 1996). A new relation (Ambraseys et al., 1996) was recently defined for absolute spectral acceleration (SA) based on the European strong motion data bank (Ambraseys and Bommer, 1991) and calibrated on 422 triaxial records generated by 157 earthquakes in Europe and adjacent regions in the magnitude Ms range 4.0 to 7.9. It takes into account four soil typologies: rock, stiff, soft, and very soft soil. It is universally accepted that the two shaking periods of 0.2 s and 1 s can give a good representation of the most important part of the ground motion. They are suitable for considering high and low frequency seismic shakings, respectively. The methodology used for defining the PGA seismic hazard map of Italy (Slejko et al., 1998 and Fig. 1) was applied, in terms of SA, considering spectral attenuation relations (Ambraseys et al., 1996). The SA values reported in all the following maps were calculated taking into account the proper standard deviation (σ ) of the spectral attenuation relation which changes slightly (from 0.27 to 0.32 in the frequency range of 0.5 - 10 Hz) from one period to another; mean values are also reported in Table 1 for the selected towns, to show the influence of the scattering in the model. This is rather remarkable as, for both the periods represented, σ produces values almost double the mean ones. The largest differences are encountered in Trieste, where the predicted values are, in any case, small. In Fig. 2 the SA(0.2) map with a 475-year return period referred to rock is shown. This figure shows that the largest values (exceeding 0.7 g) are found in the Calabrian Arc, southern and central Apennines, and in the eastern Alps. When considering Italian territory, values larger than 0.6 g are expected along almost all the Apenninic chain, in the Gargano (east of Campobasso), in the eastern Alps and east of Genova. Fig. 3 shows the SA(1.0) map with a 475-year return period referred to rock. In this case the values of acceleration are remarkably lower than in the equivalent map and the maxima are reached in Calabria and the southern Apennines, with values larger than 0.3 g. Values larger than 0.2 g are found extensively along the Calabrian Arc and in the southern Apennines, as well as in a small portion of the central Apennines. Some of the seismic areas identified in the previous SA(0.2) map are repeated here too; it is noteworthy that a few, such as in the northern Apennines between Florence and Bologna, and especially in the eastern Alps, do not appear in this latter map. Fig. 4 shows the SA(0.2) map with a 475-year return period referred to soft soil. The general features are the equivalent as in the similar map for rock: the largest values, which in this case exceed 1.0 g, are found in small spots in Calabria, in the southern, central and northern Apennines, and widely in the eastern Alps. We can generally say that the 0.6 g isoline in the map for rock (Fig. 2) refers to 0.8 g in the map for soft soil (Fig. 4), but in the latter the hazardous areas of the Gargano and along the Calabrian Arc are wider. The equivalent map for stiff soil does not show remarkable differences as against the soft soil map: the same values are reached but 34 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 8 10 12 14 16 18 A PGA SA (g) 0.5 - 0.7 0.6 0.6 CH 0.6 - 0.8 0.7 0.7 46 SLO TRENTO AOSTA 0.7 - 1.0 0.8 0.8 ≥ 1.0 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F i FLORENCE 44 ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 2 - SA(0.2) with a 475-year return period for rock. The Ambraseys et al. (1996) attenuation relation with σ was used. the pertinent areas are slightly smaller. According to the Cornell (1968) approach, the hazard estimates are determined by three main items: the seismogenic zonation, the earthquake catalogue, and the attenuation relation. There are no choices at present for the first two, as they were expressly designed by GNDT for the probabilistic hazard map of Italy; on the contrary, an alternative spectral attenuation relation suitable for the present investigation does exist. The Sabetta and Pugliese (1996) relation was calibrated on 95 35 Boll. Geof. Teor. Appl., 40, 31-51 8 10 REBEZ et al. 12 14 16 18 A PGA SA (g) 0.5 - 0.2 0.1 0.6 CH 0.6 - 0.3 0.2 0.7 46 SLO TRENTO AOSTA 0.7 - 0.4 0.3 0.8 ≥ 0.4 1.0 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F i FLORENCE 44 ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 3 - SA(1.0) with a 475-year return period for rock. The Ambraseys et al. (1996) attenuation relation with σ was used. Italian strong motion data from 17 earthquakes in the magnitude range of 4.6 to 6.8; ground shakings are given in terms of pseudo-velocity (PSV) and its σ ranges from 0.19 to 0.32 for frequencies from 0.25 to 25 Hz. This Italian relation is considered here for evaluating the differences produced and, then, quantifying the consequent uncertainty range in the hazard results. As the Sabetta and Pugliese (1996) attenuation relation gives the values of PSV expected at the site, the derivative of these values represents, then, the pseudo-acceleration (PSA) values. For earthquake- 36 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 8 10 12 14 16 18 A PGA SA (g) 0.5 - 0.7 0.6 0.6 CH 0.6 - 0.8 0.7 0.7 46 SLO TRENTO AOSTA 0.7 - 1.0 0.8 0.8 ≥ 1.0 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F i FLORENCE 44 ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 4 - SA(0.2) with a 475-year return period for soft soil. The Ambraseys et al. (1996) attenuation relation with σ was used. like excitations PSA and SA are almost identical over most of the usual frequency range (Hudson, 1979). The results of PSA(0.2) for a 475-year return period on rock are reported in Fig. 5. The maxima are located in Calabria and the southern Apennines with values which are larger than 1.0 g. Values larger than 0.8 g relate to the Messina Straits, Calabria, some segments of the southern, central, and northern Apennines, and the eastern Alps. Roughly speaking we can say that for most 37 Boll. Geof. Teor. Appl., 40, 31-51 8 10 REBEZ et al. 12 14 16 18 A PGA PSA (g) (g) 0.5 - 0.7 0.6 0.6 CH 0.6 - 0.8 0.7 0.7 46 SLO TRENTO AOSTA 0.7 - 1.0 0.8 0.8 ≥ 1.0 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F i FLORENCE 44 ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 5 - PSA(0.2) with a 475-year return period for rock. The Sabetta and Pugliese (1996) attenuation relation with σ was used. parts of the territory the 0.8 g isoline of this map (Fig. 5) corresponds to the 0.7 isoline of the SA(0.2) map (Fig. 2); but, for quantifying the difference caused by the use of the two different attenuation relations, the ratio between the PSA(0.2) values according to Sabetta and Pugliese (1996) and SA(0.2) according to Ambraseys et al. (1996) is mapped in Fig. 6. The results according to Sabetta and Pugliese (1996) are always higher than those of Ambraseys et al. (1996): the most 38 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 8 10 12 14 16 18 A RATIO Ratio 0.5 - -0.6 1.00 1.19 CH 0.6 - -0.7 1.20 1.39 46 SLO TRENTO AOSTA 0.7 - -0.8 1.40 1.59 ≥ 1.0 1.60 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F i FLORENCE 44 ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 6 - Ratio between PSA(0.2) computed with Sabetta and Pugliese (1996) and SA(0.2) with Ambraseys et al. (1996) attenuation relations for rock with σ. Cells where at least one of the two quantities is lower than 0.4 g are not mapped. striking differences are encountered in Calabria (values larger by more than 40%) and, to a lesser extent, in all of southern Italy. This is explained by the characteristics of the two relations: the Sabetta and Pugliese (1996) relation predicts greater accelerations where the contribution of high magnitudes is relevant. Fig. 7 highlights the different behaviour of the two relations, when σ are 39 Boll. Geof. Teor. Appl., 40, 31-51 REBEZ et al. 10 Acceleration (g) Acceleration (g) 10 T=0.2s 1 T=1.0s 1 Ms=7.0 0.1 Ms=7.0 0.1 Ms=5.5 0.01 Ms=5.5 0.01 Ms=4.0 a b 0.001 1 10 Distance (km) 100 1 10 Distance (km) 100 10 Acceleration (g) 10 Acceleration (g) Ms=4.0 0.001 T=0.2s SD 1 T=1.0s SD 1 Ms=7.0 0.1 Ms=7.0 0.1 Ms=5.5 Ms=5.5 0.01 0.01 Ms=4.0 Ms=4.0 c d 0.001 0.001 1 10 Distance (km) 100 1 10 Distance (km) 100 Fig. 7 - Comparison between the Ambraseys et al. (1996) (solid lines show SA) and the Sabetta and Pugliese (1996) (dashed lines show PSA) spectral attenuation relations for rock: a) 0.2 s; b) 1.0 s; c) 0.2 s considering s; d) 1.0 s considering σ. ML values were converted (Camassi and Stucchi, 1996) into Ms values when necessary in the Sabetta and Pugliese (1996) relation. considered and when they are not, for three magnitude values (4.0, 5.5, and 7.0) and for the two periods considered in the present work (i. e. 0.2 and 1.0 seconds). At 0.2 s (Figs. 7a and 7c) the relations are always rather similar and σ seems to reduce their differences slightly. At 1.0 s (Figs. 7b and 7d) the Sabetta and Pugliese (1996) relation is definitely higher, especially in the near field (a less than 50 km distance). As a general comment, it can be said that differences among attenuation relations are always significant (see Abrahamson and Shedlock, 1997); those shown here, although notable, especially for 1.0 s, are not surprising if we consider that the relations were cali- 40 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 8 10 12 14 16 18 A PGA PSA (g) (g) 0.5 - 0.2 0.1 0.6 CH 0.6 - 0.3 0.2 0.7 46 SLO TRENTO AOSTA 0.7 - 0.4 0.3 0.8 ≥ 0.4 1.0 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F i FLORENCE 44 ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 8 - PSA(1.0) with a 475-year return period for rock. The Sabetta and Pugliese (1996) attenuation relation with σ was used. brated on different data sets (the Italian one has only two earthquakes with a magnitude larger than 6), and regressions were made for different parameters (SA and PSV). The hazard results depend, therefore, on these aspects of the attenuation considered here. It is, then, interesting to analyze the differences encountered when considering the values at a 1 s period. Fig. 8 shows PSA(1.0) computed according to the Sabetta and Pugliese (1996) relation. 41 Boll. Geof. Teor. Appl., 40, 31-51 8 10 REBEZ et al. 12 14 16 18 A RATIO Ratio 0.5 - -0.6 1.00 1.19 CH 0.6 - -0.7 1.20 1.39 46 SLO TRENTO AOSTA 0.7 - -0.8 1.40 1.59 ≥ 1.0 1.60 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F 44 i FLORENCE ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 9 - Ratio between PSA(1.0) computed with Sabetta and Pugliese (1996) and SA(1.0) with Ambraseys et al. (1996) attenuation relations for rock with σ. Cells where at least one of the two quantities is lower than 0.1 g are not mapped. Values larger than 0.4 g are found extensively in Calabria and the southern Apennines; values larger than 0.2 g cover most of the Apennines and eastern Alps. The results according to the Sabetta and Pugliese (1996) relation are, again, definitely larger than those of the Ambraseys et al. (1996) relation. This aspect clearly stands out in Fig. 9, where the ratio between PSA(1.0) and SA(1.0) is 42 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 MILAN VENICE TRIESTE FLORENCE ROME NAPLES MESSINA CATANIA 1st cat. 2nd cat. 3rd cat. spectral acceleration (g) 0.8 0.6 0.4 0.2 0 0.1 1 period (s) Fig. 10 - Uniform hazard response spectra with a 475-year return period on rock for Catania, Messina, Naples, Rome, Florence, Milan, Venice, and Trieste. The Ambraseys et al. (1996) attenuation relation with σ was used. Solid thick lines represent the elastic response spectra for the three categories of the Italian seismic code. mapped. The general features of this map are similar to those of the PSA/SA ratio at 0.2 s (Fig. 6) but the values are now remarkably higher, since the Sabetta and Pugliese (1996) PSA(1.0) values in the southern Apennines and eastern Sicily are now from 60% to 90% larger than the Ambraseys et al. (1996) SA(1.0) ones. 3. Uniform hazard response spectra The previous spectral maps give enough information on the regional seismic hazard but certainly do not give the complete information which could be useful in urban planning. This kind of information is given by the uniform hazard response spectra that are calculated here for rocky soil at 14 periods in the range of 0.1 - 2.0 s and refer to SA with a 475-year return period. Some apparently anomalous fluctuations in the shape of the spectra are related to the processing of the accelerograms performed by Ambraseys et al. (1996): the authors, in fact, computed raw regression coefficients, admitting some statistical fluctuation of the coefficients that can be caused by the spec- 43 Boll. Geof. Teor. Appl., 40, 31-51 REBEZ et al. trum sampling too. This effect is particularly evident when σ is taken into account, as the different values of σ for each frequency can modify the shape of the spectrum, independently from the characteristics of the seismicity affecting the site (see Peruzza et al., 1998b). The complete spectrum was computed for all the 8101 Italian municipalities. We show those which refer to eight main Italian towns, from south to north: Catania, Messina, Naples, Rome, Florence, Milan, Venice, and Trieste (see their location in Fig. 1). The results, referred to rocky sites, are shown in Fig. 10; rock conditions only partly represent the municipalities, but are chosen in this case for comparison. It can be seen that for all the towns the maximum values are reached in the period range of 0.1 to 0.3 s, when the decay becomes rapid. The maximum values obtained are remarkably different, varying from less than 0.2 g (Milan) to more than 0.7 g (Messina). The results for the eight towns can be roughly grouped into three categories: high (Messina with values exceeding 0.7 g), medium (Catania, Naples, Florence, and Rome with values between 0.4 g and 0.55 g), and low accelerations (Trieste, Venice, and Milan with values lower than 0.3 g). In the same figure the elastic response spectra for the three categories of the present Italian seismic code are shown. It can be noted that the computed spectra for most of the studied towns are significantly lower than those of the proper seismic categories, especially in the low frequency band. Exceptions are in the spectrum of Naples, which exceeds the official 3rd category one where the town is placed, and in that of Rome, because the town is not classified (see Table 1). The spectra show limited differences in their shape due to the dominant seismicity; this is an obvious consequence of the Cornell (1968) methodology, used here in considering homogeneous seismicity over seismogenic area sources. In addition, the σ values modify the original shape of the spectrum, and, therefore, the effects of the source influences. Nevertheless, some towns present quite a flat spectrum, in the range of 0.1-0.3 s (Messina, Naples, Venice) while other localities exhibit a clearly defined maximum at 0.15 s (Catania, Florence, Rome, Trieste). It is interesting to note how similar the Naples and Catania spectra are, differing only in the high frequency domain (periods smaller than 0.3 s). Spectra referring to two soil types (rock and soft soil) are presented for Trieste, and Catania; both cities have a great terrain variability, from hard limestone (Trieste) and lava banks (Catania) to soft Holocene clays and incohesive sea deposits. Fig. 11 shows that the soil type has a notable influence in determining the expected values, which are about 30% larger for soft soils, and also in defining the shape of the spectra, as the maximum values move to higher periods when taking soft soils into account. 4. The effective peak acceleration map The complete uniform hazard response spectra, presented in the previous section, give exhaustive information about seismic hazard at a site. On the other hand, comparing hazard for different sites cannot be easily appreciated. It is quite interesting to compare seismic hazard given in a more complete way than PGA, at regional or national scale, on a single map. The average SA value in a frequency range of engineering interest, which is called effective peak acceleration (EPA), can satisfy this need. Although EPA is a philosophically sound parameter for sei- 44 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 0.7 soft spectral acceleration (g) 0.6 0.5 CATANIA rock 0.4 soft 0.3 TRIESTE rock 0.2 0.1 0 0.1 1 period (s) Fig. 11 - Uniform hazard response spectra with a 475-year return period on rock and soft soil for Trieste and Catania. The Ambraseys et al. (1996) attenuation relations with their σ’s were used. smic hazard analysis, there is no standardised definition of this parameter at present (Uang and Bertero, 1988). ATC-13 (1985) defines EPA as follows: EPA = PSA'/2.5 where PSA' is the 5% dumped mean PSA value in the period range of 0.1 to 0.5 seconds. EPA maps can, then, represent the characteristics of the dominant part of the response spectrum well and have been prepared according to the previously cited criteria (Ambraseys et al. (1996) attenuation relation for rock). Fig. 12 shows EPA with a 475-year return period for Italy. Many areas with values larger than 0.24 g can be identified, in southern Italy (Messina Straits and Calabria), along the southern, central, and, less widely, the northern Apennines, as well as in the eastern Alps: among these, one near Potenza and Calabria also exceeds 0.28 g. Comparing this EPA map to the PGA one (Fig. 1), a general similarity can be noted but the EPA map emphasises only some of the areas were PGA reaches the highest values. For better pinpointing the seismic hazard with a short return period and, consequently, the regions frequently damaged by earthquakes, Fig. 13 shows EPA with a 100-year return period. Almost all the previously cited hazardous areas again appear with high values (larger than 0.12 g). It is interesting to notice that the two high spots in Calabria do not correspond to the seismic 45 Boll. Geof. Teor. Appl., 40, 31-51 8 10 REBEZ et al. 12 14 16 18 A PGA EPA (g)(g) 0.5 - -0.6 0.20 0.24 CH 0.6 - -0.7 0.24 0.28 46 SLO TRENTO AOSTA 0.7 - -0.8 0.28 0.32 ≥ 1.0 0.32 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F 44 i FLORENCE ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 12 - EPA with a 475-year return period for rock. The Ambraseys et al. (1996) attenuation relation with σ was used. area evidenced by Fig. 12. The differences between the two maps, in the higher EPA areas, are caused by the type of seismicity experienced: in the southern Apennines strong earthquakes have occurred repeatedly, and have influenced the long return period hazard; in central and northern Italy high magnitude events are rare but medium magnitude quakes occur frequently, and they contribute to the short return period hazard. In addition, the medium magnitude earthquakes show higher acceleration 46 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 8 10 12 14 16 18 A PGA EPA (g)(g) 0.5 - -0.6 0.08 0.12 CH 0.6 - -0.7 0.12 0.16 46 SLO TRENTO AOSTA 0.7 - -0.8 0.16 0.20 ≥ 1.0 0.20 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F 44 i FLORENCE ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 13 - EPA with a 100-year return period for rock. The Ambraseys et al. (1996) attenuation relation with σ was used. in the high frequency range, where EPA is defined. EPA with a 475-year return period for soft soil has been computed as well (Fig. 14): the attenuation relation for this type of soil increases by about 0.08 g of the values computed on rock. In fact, the isolines of 0.24 g in Fig. 12 are very similar to the isolines of 0.32 g in Fig. 14. In addition, EPA according to the Sabetta and Pugliese (1996) relation has been computed for a general comparison (Fig. 15). The same seismic areas as in Fig. 12 are pointed out, and, as 47 Boll. Geof. Teor. Appl., 40, 31-51 8 10 REBEZ et al. 12 14 16 18 A PGA EPA (g)(g) 0.5 - -0.6 0.20 0.24 CH 0.6 - -0.7 0.24 0.28 46 SLO TRENTO AOSTA 0.7 - -0.8 0.28 0.32 ≥ 1.0 0.32 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F 44 i FLORENCE ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA 16 18 Fig. 14 - EPA with a 475-year return period for soft soil. The Ambraseys et al. (1996) attenuation relation with σ was used. expected, the values are larger, especially in southern Italy. As a general comment on the EPA maps, it can be said that they show similar differences between each other in the same way as the PGA maps (Slejko et al., 1998), such as: similar shape but increasing value passing from rock to soft soil, and greater importance of the medium magnitude frequent seismicity moving from the 475-year to the 100-year return period. 48 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 8 10 12 14 16 18 A PGA EPA (g)(g) 0.5 - -0.6 0.20 0.24 CH 0.6 - -0.7 0.24 0.28 46 SLO TRENTO AOSTA 0.7 - -0.8 0.28 0.32 ≥ 1.0 0.32 TRIESTE MILAN 46 VENICE HR TORINO A BOLOGNA GENOVA d BH 44 r F i FLORENCE 44 ANCONA a t PERUGIA i c L'AQUILA 42 S 42 ROME e CAMPOBASSO a BARI NAPLES dini a POTENZA Sar 40 40 CAGLIARI T y r r h e n i a n S e a MESSINA 38 PALERMO Sic 8 10 12 ily 14 38 REGGIO CALABRIA CATANIA EPA_Sr_475 16 18 Fig. 15 - EPA with a 475-year return period for rock. The Sabetta and Pugliese (1996) attenuation relation with σ was used. 5. Conclusions The PGA 475-year return period is the choice for the standard hazard map used for seismic zonation. The information in this map, although relevant, is not exhaustive as representing the hazard at specific sites. For this reason three further types of hazard representations were compu- 49 Boll. Geof. Teor. Appl., 40, 31-51 REBEZ et al. ted and presented. Namely: the spectral hazard maps referred to two periods, the uniform hazard response spectra for the main Italian towns, and the EPA maps. A more precise view of the Italian seismic hazard is given by considering them globally. The PGA and the SA(0.2) maps are quite similar and point out many sectors of the southern and central Apennines, and the eastern Alps with almost the same accelerations. The SA(1.0) map focuses on Calabria and the southern Apennines as the areas with the highest hazard. The EPA map again enhances Calabria and the southern Apennines, giving the other areas of the PGA map slightly lower accelerations. As already mentioned, different aspects of hazard are displayed by the maps: the influence of very strong earthquakes (SA(1.0) map), the influence of local seismicity in addition to strong seismicity (PGA and SA(0.2) map), as well as the average contribution of major and local events with the exception of low frequency ones (EPA map). In conclusion, it is the authors' opinion that the EPA maps offer more balanced information on the varying contribution of the seismicity affecting a site, and, is, consequently, more suitable for zonation purposes. This knowledge can be improved by analysing the uniform hazard response spectra, where the spectral shaking contents are detailed and the influence of the reference soil can be taken into account. Acknowledgements. This research was conducted in the framework of the activities of the “Gruppo Nazionale per la Difesa dai Terremoti”, contract n. 97.00537.PF54. Many thanks are due to Dieter Mayer-Rosa, for his very helpful review of the paper, and to Luis Decanini, for some information on the elastic response spectra of the italian seismic code. References Abrahamson N. A. and Shedlock K. M.; 1997: Overview. Seismol. Res. Letters, 68, 9 - 23. Ambraseys N. N.; 1995: Reappraisal of the prediction of ground accelerations in Europe - EAEE Working Group report. In: Duma (ed), 10th European Conference on Earthquake Engineering, Balkema, Rotterdam, pp. 3041 - 3048. Ambraseys N. N. and Bommer J. J.; 1991: Database of European strong ground-motion records. European Earthquake Engineering, 5/2, 18 - 37. Ambraseys N. N., Simpson K. A. and Bommer J. J.; 1996: Prediction of horizontal response spectra in Europe. Earth. Eng. Struct. Dyn., 25, 371 - 400. ATC-13; 1985: Earthquake Damage Evaluation Data for California. ATC-13 Report. Applied Technology Council. Redwood City, California. Basham P., Halchuk S., Weichert D. and Adams J.; 1997: New seismic hazard assessment for Canada. Seismol. Res. Lett., 68, 722 - 726. Bender B. and Perkins D. M.; 1987: Seisrisk III: a computer program for seismic hazard estimation. U. S. Geological Survey Bulletin 1772, 48 pp. Camassi R. and Stucchi M.; 1996: NT4.1 un catalogo parametrico di terremoti di area italiana al di sopra della soglia del danno. Rapporto interno GNDT, Milan, 86 pp. Cornell C. A.; 1968. Engineering seismic risk analysis. Bull. Seism. Soc. Am., 58, 1583-1606. Hudson D. E.; 1979: Reading and interpreting strong motion accelerograms. Earthquake Engineering Research Institute, Berkeley, 112 pp. Peruzza L., Rebez A. and Slejko D.; 1998a: Damage oriented spectral seismic hazard. In: 26th General Assembly of the 50 Spectral PSHA for Italy Boll. Geof. Teor. Appl., 40, 31-51 European Seismological Commission (ESC). Papers, ElAl, Tel Aviv, pp. 165 - 169. Peruzza L., Rebez A and Slejko D.; 1998b: Probabilistic seismic hazard spectra for the Adriatic region. In: Bisch P., Labbè P. and Pecker A. (eds), Proceedings of the Eleventh European Conference on Earthquake Engineering, Abstract volume and CD-ROM, pp. 48 and CD-ROM. Peruzza L., Sirovich L. and Slejko D.; 1994: Spectral characteristics of the seismic hazard between the Alps and the Dinarides. Soil Dynamics and Earthquake Engineering, 13, 213-217. Pugliese A. and Sabetta F.; 1989: Stima di spettri di risposta da registrazioni di forti terremoti italiani. Ingegneria Sismica, 6, 3 - 14. Romeo R. and Pugliese A.; 1997: Analisi probabilistica della scuotibilità del territorio italiano. Ingegneria Sismica, 14, 68 - 77. Sabetta F. and Pugliese A.; 1996: Estimation of response spectra and simulation of nonstationarity earthquake ground motion. Bull, Seism. Soc. Am., 86, 337 - 352. Slejko D., Peruzza L. and Rebez A.; 1998: The seismic hazard maps of Italy. Annali di Geofisica, 41, 183 - 214. Tento A., Franceschina L. and Marcellini A.; 1992: Expected ground motion evaluation for Italian sites. In: Proc. 10th World Conf. Earth. Eng., Madrid, vol. 1, pp. 489 - 494. Uang C. M. and Bertero V. V.; 1988: Implications of recorded earthquake ground motions on seismic design of building structures. Earthquake Engineering Research Center, University of California, Report UCB/EERC-88/13, Berkeley. 51 View publication stats