ShortCoMINGS of the CurreNt aSSeSSMeNt of SeISMIC …ShortCoMINGS of the CurreNt aSSeSSMeNt of SeISMIC haZarD: exaMPLeS froM SoMe ItaLIaN reGIoNS M. Viti1, D. Babbucci1, M. Baglione2,

ShortCoMINGS of the CurreNt aSSeSSMeNt of SeISMIC haZarD: exaMPLeS froM SoMe ItaLIaN reGIoNSM. Viti1, D. Babbucci1, M. Baglione2, N. Cenni3, V. D’Intinosante2, e. Mantovani1, L. Martelli4,C. tamburelli11 Dipartimento di Scienze Fisiche, della Terra e dell’Ambiente, Università di Siena, Italy2 Coordinamento Regionale Prevenzione Sismica, Regione Toscana, Italy3 Dipartimento di Scienze Biologiche, Geologiche e Ambientali, Università di Bologna, Italy4 Servizio Geologico, Sismico e dei Suoli, Regione Emilia-Romagna, Italy

1. Introduction. It is well known that, both in Italy and in the rest of the world, theassessment of seismic hazard is currently carried out by probabilistic approaches, indicated as Probabilistic Seismic Hazard Assessment (PSHA). Such procedure, implemented over the last four decades, estimates the parameters of ground motion, as the Peak Ground Acceleration (PGA). This methodology takes into account the frequency of seismic events occurring at the seismogenic sources recognized in the study area, as deduced from seismic history. Moreover, the effects caused by each shock are estimated by empirical relationships, describing the attenuation of ground motion with the distance from the hypocenter to the site of interest.

The consequences of the destructive earthquakes occurred worldwide in the last decade have cast doubts about the reliability of the above methodology, also inducing some authors to suggest a substantial revision of the probabilistic approach, in order to achieve a more realistic estimate of seismic hazard (e.g., Stein et al., 2012 and references therein). Taking into account the above suggestion, this study is aimed at focusing the main shortcomings of the PSHA, with particular attention to the assessment of seismic hazard in four administrative regions of central-northern Italy (Emilia-Romagna, Toscana, Umbria and Marche).

Description of the PSHA. The PSHA develops by four distinct, consecutive phases (e.g., Cornell, 1968; McGuire, 1978; Algermissen e Perkins, 1976; Reiter, 1990; Frankel et al., 2000). In the first phase, it is assumed that the seismicity of the study area is generated by some seismic sources, usually schematized by polygons including recognized (or presumed) active fault systems. For the Italian region, the previous ZS4 seismic zoning (Meletti et al., 2000) has been replaced by the ZS9 model, formed by 36 major seismogenic zones (Meletti et al., 2008).

In the second phase, a set of parameters characterising the seismic activity, as e.g. the maximum expected magnitude Mmax, is assigned to each of the above zones. Moreover, for each value of magnitude, from the damage threshold to Mmax, the number of earthquakes expected in a given time interval is computed. Indeed, the so-called “seismic scenario” is represented by

474

GNGTS 2013 SeSSione 2.3

131218 - OGS.Atti.32_vol.2.sez.3.08.indd 474 07/11/13 09.05

magnitude, epicentral location and average annual rate of occurrence (Gruppo di Lavoro MPS, 2004). In this regard, it is usually assumed that the evolution of seismic activity is a Poisson process, characterized by complete independence among seismic sources (e.g., Stucchi et al., 2011).

In order to estimate the consequences of the various “seismic scenarios” in the sites of interest, the third phase of PSHA concerns the definition of a suitable ground motion model. It is usually represented by a set of empirical relationships between the earthquake magnitude, hypocentral distance and parameters of ground shaking such as the PGA (e.g., Ambraseys et al., 1996; Sabetta e Pugliese, 1996).

In the fourth and last phase, the amplitude of ground motion is computed for all possible “seismic scenarios” related to the various sources of the adopted seismogenic model. This elaboration is performed by using specific resources, as the SeisRisk III code (Bender e Perkins, 1987 and further developments). The results obtained are the so-called “hazard functions”, from which the seismic hazard maps are then obtained. Such maps report the ground motion expected within a given time interval in each point of the study area, with a given exceedence probability. A number of maps can be obtained by the procedure, in relation to different probability values (e.g., 81%, 63%, 50%, 39%, 30%, 22%, 10%, 5% e 2%) and different time intervals (e.g., 30 or 50 years, Gruppo di lavoro MPS, 2004). Moreover, the estimated ground motion parameter may be the PGA or the spectral acceleration, which defines the response of the soil-building system at different natural period of oscillation (e.g., 0.1, 0.15, 0.2, 0.3, 0.4, 0.5, 0.75, 1.0, 1.5 e 2.0 seconds, Gruppo di lavoro MPS, 2004). However, it should be noted that current Italian regulations about seismic classification (e.g., Gruppo di Lavoro MPS, 2004; Ordinanza PCM/3519, 2006) require the map of PGA values expected in 50 year for a 10% exceedence probability. Furthermore, the PSHA may provide hazard maps in term of expected macroseismic intensity, usually in the scale Mercalli-Cancani-Sieberg or MCS (e.g., Gomez Capera et al., 2010). This last kind of elaboration will be discussed later, in relation to the probabilistic estimates proposed for central-northern Italy.

Main problems of the PSHA. An outstanding problem of the PSHA is the reliability of the hypotheses underlying this kind of approach. For instance, seismic activity is considered a steady-state process. This implies that the knowledge of past earthquakes may be used to predict some aspects of the future evolution of the seismicity. However, earthquake catalogues are reliable and satisfactorily complete for short time intervals only (the last centuries at most, e.g., Gruppo di Lavoro MPS, 2004). Of course, such intervals are much shorter than the periods in which seismogenic fault systems have developed in the study area (of the order of a million years). Moreover, several investigations (e.g., Castanos e Lomnitz, 2000; Stein et al., 2013 and references therein) point out that concepts often involved in the probabilistic hazard assessments, such as “seismic cycle”, “characteristic earthquake” and “seismic gap”, can not easily be applied to the interpretation of the real seismic activity. Finally, there is a growing awareness that seismogenic faults may have long-term and long-range interactions, in contrast with the hypothesis of complete independence of seismic sources (e.g., Scholz e Gupta, 2000; Lou e Liu, 2010, Mantovani et al., 2012a,b, 2013; Viti et al., 2012, 2013).

Another major problem is testing the predictions provided by PSHA. Indeed, a rigorous verification of hazard maps would require many records of strong motion in a very long time interval, i.e. several consecutive periods of 50 years (Meletti et al., 2012; Frankel, 2013). A preliminary analysis has revealed a considerable discrepancy between the strong motion values, recorded at various Italian accelerometric stations in the period 1975-2005, and the PGA values predicted in the same sites by the PSHA (Albarello and D’Amico, 2008). In particular, the probabilistic assessment has underestimated the level of seismic shacking in 19 out the 68 accelerometric sites considered.

Several authors (e.g., Reyners, 2011; Stein et al., 2012) have shown that the PSHA estimates performed at the end of the last century have not adequately predicted the ground motion

475



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



earthquakes, the probabilistic expected PGA may well be smaller than the recorded ground motion, due to significant site effects (e.g., Celebi et al., 2010; Meletti et al., 2012).

As discussed earlier, the first phase of PSHA is the definition of a set of seismogenic zones where past and future seismicity does occur. For the Italian region, the rather detailed ZS4 model (80 zones) has been replaced by the much simpler ZS9 model (36 zone only). This choice has favoured the statistical analysis, since in average each new, larger zone includes more earthquakes. On the other hand, the new seismic zoning is less realistic with respect to the seismotectonics of Italy. For instance, the Apennine chain is currently subdivided in few, very long belts (as shown in figure 1 for the central and northern Apennines), which heavily affect the resulting hazard estimates. However, the distribution of historical seismicity (Fig. 1) shows that the real seismic sources are not uniformly distributed within the adopted seismogenic zones (Mantovani et al., 2011, 2012a, 2013). Thus, the seismic classification of the communes of the administrative regions located in the northern Apennines (Emilia-Romagna, Toscana, Umbria and Marche) are significantly affected by the above problems, since the classification is based by law on the PGA level predicted by PSHA (Ordinanza PCM/3519, 2006).

Shortcomings of the PSHA for some Italian regions. The problems discussed in the previous section are better focused when specific sectors of the Italian region are considered. Moreover, the discrepancies between the results provided by PSHA and the information coming from seismic history become more evident when the probabilistic assessment is given in terms of maximum expected macroseimic intensity (Imax, Gomez Capera et al., 2010 and references therein).

Fig. 2 – Detail of the seismic hazard map obtained from PSHA, related to the four Italian region here considered (Emilia-Romagna, Toscana, Umbria and Marche). The map shows the maximum value expected for the Mercalli-Cancani-Sieberg (MCS) macroseismic intensity, which has a 10% exceedence probability within the next 50 years (modified after Gomez Capera et al., 2010). Regional and provincial borders are indicated by thick and thin blue lines, respectively.

477



The maximum intensity estimate by PSHA for the Emilia-Romagna, Toscana, Umbria and Marche Regions (Fig. 2) shows that exceeding 8° MCS in the next decades will be unlikely. This is at variance with information about past earthquakes, which suggests Imax has repeatedly reached 10° MCS in several areas of the above regions (e.g., Locati et al., 2011). The discrepancy is due to the fact that the probability of next shocks producing an intensity exceeding the 8°MCS is considered low, because past similar events has been relatively rare. In other words, very damaging shocks would be characterized by a very long seismic cycle. However, this assumption does not take into account the shortness of the seismic catalogues and the fact that we do not known when a seismic cycle started for each major seismic source. Furthermore, the reliability of the seismic cycle approach is challenged by the interaction of seismic sources, which implies a reorganization of stress and strain regimes after major shocks (e.g., Viti et alii, 2012, 2013).

As suggested above, the distribution of expected Imax (Fig. 2) is clearly controlled by the geometry of the adopted seismogenic zoning, in particular by the zones 915 and 919 of ZS9 model (Fig. 1). For Emilia-Romagna, the Imax expected in the Piacentino, Parmense and Reggiano (Fig. 2) are scarcely affected by the strong earthquakes occurred in AD 1117, whose importance has recently been highlighted by Mantovani et al. (2013). As discussed above, the PSHA assumes a very low probability that an event similar to the 1117 shock will occur in the next 50 years. For Tuscany, the zone with the largest expected Imax covers without interruptions the main Apennine watershed (Fig. 2). This not agrees with the spatial distribution of past earthquakes. Indeed, the seismicity of intramontane basins such as Lunigiana-Garfagnana, Mugello ed Upper Tiber Valley is far stronger than that of nearby zones, such as the Pistoiese Apennine and Casentino basins (e.g., Mantovani et al., 2011, 2012a, 2013). Furthermore, the probabilistic map predicts an uniform value (7° MCS) for most of Tuscany, out of the above-mentioned watershed strip and part of the Tyrrhenian coast (6° MCS). This result apparently neglects the role of important seismic sources located in western Tuscany, such as the Florentine Chianti hills, Middle Valdarno, Leighorn-Pisan hills and the Amiata Mt., where strong earthquakes were recorded in the past (Mantovani et al., 2011, 2012a, 2013). Similar shortcomings also affect both Marche and Umbria. The Imax predicted by PSHA for most of such regions is 8°MCS (Fig. 2), whereas the past and recent seismic history suggests that much higher seismic damage may be observed (e.g., Locati et al., 2011). For instance, the recorded macroseismic intensity was 9° MCS for the 1997, Colfiorito event and presumably reached 10° MCS for the 1751, Gualdo Tadino and 1781, Cagli earthquakes.

Conclusions. The probabilistic procedure currently adopted to estimate seismic hazard is affected by several shortcomings, concerning both the main assumptions underlying the procedure and its application to specific zones. The effects of the destructive earthquakes occurred in the last decade in Italy and in the rest of the world have pointed out many of the flaws of PSHA. The analysis of the estimate of macroseismic intensity, provided by the probabilistic procedure for four Italian regions, suggests that the maximum intensity values expected from the next strong shocks may be considerably underestimated. Moreover, several local features of the distribution of maximum intensity values (as deduced from macroseismic databases) are neglected in the probabilistic seismic hazard map. This fact casts doubt on the possibility that Italian public bodies (such as Regioni, Province and Comuni) can reliably use PSHA results in order to manage seismic risk. referencesAlbarello D., D’Amico V.; 2008: Testing probabilistic seismic hazard estimates by comparison with observations: an

example in Italy. Geophys. J. Int., 175, 1088-1094.Algermissen S.T., Perkins D.M.; 1976: A probabilistic estimate of maximum acceleration in rock in the contiguous United

States. U.S. Geological Survey Open-File Report 76-416.Ambraseys, N.N., Simpson K.A., Bommer J.J., 1996; Prediction of horizontal response spectra in Europe, Earthq. Eng.

Struct. D., 25, 371-400.

478



Bender B., Perkins D.M.; 1987: SEISRISK III: a computer program for seismic hazard estimation. U.S. Geological Survey Bulletin, 1772, pagg. 48.

Castanos H., Lomnitz C.; 2002: PSHA: is it science? Engineering Geology, 66, 315-317.Celebi M., Bazzurro P., Chiaraluce L., Clemente P., Decanini L., De Sortis A., Ellsworth W., Gorini A., Kalkan E.,

Marcucci S., Milana G., Mollaioli F., Oliveri M., Paolucci R., Rinaldis D., Rovelli A., Sabetta F., Stephens C.; 2010: Recorded motion of the 6 April 2009 MW 6.3 L’Aquila, Italy, earthquake and implications for building structural damage: overview. Earthquake Spectra, 26, 651-684.

Cornell C.A.; 1968: Engineering seismic risk analysis. Bull. Seism. Soc. America, 58, 1583-1606. Frankel A. et al.; 2000: USGS national seismic hazard maps. Earthquake Spectra, 16, 1-20.Frankel, A.; 2013: Comment to “Why earthquake hazard maps often fail and what to do about it”, by S. Stein, R.J. Geller,

and M. Liu. Tectonophysics, 592, 200-206. Gomez Capera A.A., D’Amico V., Meletti C., Rovida A., Albarello D.; 2010: Seismic hazard assessment in terms of

macroseismic intensity in Italy: a critical analysis from the comparison of different computational procedures. Bull. Seism. Soc. America, 100, 1614-1631.

Gruppo di Lavoro MPS; 2004: Redazione della mappa di pericolosità sismica prevista dall’Ordinanza PCM 3274 del 20 marzo 2003. Rapporto conclusivo per il Dip.to della Protezione Civile, INGV, Milano-Roma, 65 pp. + 5 appendici (http://zonesismiche.mi.ingv.it/).

Locati M., Camassi R., Stucchi M. (a cura di); 2011: DBMI11, la versione 2011 del Database Macrosismico Italiano. Milano, Bologna (http://emidius.mi.ingv.it/DBMI11).

Luo, G., Liu, M.; 2010: Stress evolution and fault interactions in eastern Tibet: implications for the 2008 Mw 7.9 Wenchuan earthquake. Tectonophysics, doi:10.1016/j.tecto.2009.12.019.

Mantovani E., Viti M., Babbucci D., Cenni N., Tamburelli C., Vannucchi A., Falciani F., Fianchisti G., Baglione M., D’Intinosante V., Fabbroni P.; 2011: Sismotettonica dell’Appennino Settentrionale. Implicazioni per la pericolosità sismica della Toscana. Regione Toscana, Centro stampa Giunta Regione Toscana, Firenze, pagg. 88 (www.regione.toscana.it).

Mantovani E., Viti M., Babbucci D., Cenni N., Tamburelli C., Vannucchi A., Falciani F., Fianchisti G., Baglione M., D’Intinosante V., Fabbroni P.; 2012a: Potenzialità sismica della Toscana e definizione di criteri per interventi di prevenzione. Regione Toscana, Centro stampa Giunta Regione Toscana, Firenze, pagg. 140 (www.regione.toscana.it).

Mantovani E., Viti M., Babbucci D., Cenni N., Vannucchi A.; 2012b: Middle-term prediction of earthquakes in Italy: some remarks on empirical and deterministic approaches. Boll. Geof. Teor. Appl., 53, 89-111.

Mantovani E., Viti M., Babbucci D., Cenni N.), Tamburelli C., Vannucchi A., Falciani F., Fianchisti G., Baglione M., D’Intinosante V., Fabbroni P., L.Martelli, P.Baldi, M.Bacchetti, 2013. Assetto tettonico e potenzialità sismica dell’Appennino Tosco-Emiliano-Romagnolo e Val Padana. Centro Stampa - Regione Emilia-Romagna, pagg. 168 (www.regione.toscana.it).

McGuire R.K.; 1978: FRISK: Computer program for seismic risk analysis using faults as earthquake sources. U. S. Geol. Surv. Open File Report, 78-1007, pagg. 71.

Meletti C., Patacca E., Scandone P.; 2000: Construction of a seismotectonic model: the case of Italy. Pure Appl. Geophys., 157, 11-35.

Meletti C., Galadini F., Valensise G., Stucchi M., Basili R., Barba S., Vannucci G., Boschi E.; 2008: A seismic source zone model for the seismic hazard assessment of the Italian territory.Tectonophysics, 450, 85-108.

Meletti C., D’Amico V., Ameri G., Rovida A., Stucchi M.; 2012: Seismic hazard in the Po Plain and the 2012 Emilia earthquakes. Ann. Geophysics, 55, 623-629.

Ordinanza del Presidente del Consiglio dei Ministri n. 3519 del 28/4/2006. Criteri generali per l’individuazione delle zone sismiche e per la formazione e l’aggiornamento degli elenchi delle medesime zone. Gazzetta Ufficiale della Repubblica Italiana, Serie Generale, Anno 147°, N. 108, pagg. 11-14.

Reiter L.; 1990: Earthquake hazard analysis-issues and insights. New York, Columbia University Press, pagg. 254.Reyners M.; 2011: Lessons from the Destructive MW 6.3 Christchurch, New Zealand, Earthquake. Seism. Res. Letters,

82, 1-4.Sabetta F., Pugliese A.; 1996: Estimation of response spectra and simulation of nonstationary ground motion. Bull. Seism.

Soc. Am., 86, 337-352. Scholz C.H., Gupta A.; 2000: Fault interactions and seismic hazard. J. Geodynamics, 29, 459-467.Stein S., Geller R.J., Liu M.; 2012: Why earthquake hazard maps often fail and what to do about it. Tectonophysics, 562-

563, 1-25. Stein S., Geller R.J., Liu M.; 2013: Reply to comment by Arthur Frankel on “Why Earthquake Hazard Maps Often Fail and

What to do About It”. Tectonophysics, 592, 207-209.Stucchi M., Meletti C., Montaldo V., Crowley H., Calvi G.M., Boschi E.; 2011; Seismic Hazard Assessment (2003-2009)

for the Italian Building Code. Bull. Seismol. Soc. Am., 101, 1885-1911.Viti M., Mantovani E., Cenni N.,Vannucchi A..; 2012: Post-seismic relaxation: An example of earthquake triggering in the

Apennine belt (1915-1920). J. Geodynamics, 61, 57-67, http://dx.doi.org/10.1016/j.jog.2012.07.002. Viti M., Mantovani E., Cenni N., Vannucchi A.; 2013: Interaction of seismic sources in the Apennine belt. J. Phys. Chem.

Earth, 63, 25.35, http://dx.doi.org/10.1016/j.pce.2013.03.005.x

479



Documents

ShortCoMINGS of the CurreNt aSSeSSMeNt of SeISMIC …ShortCoMINGS of the CurreNt aSSeSSMeNt of SeISMIC haZarD: exaMPLeS froM SoMe ItaLIaN reGIoNS M. Viti1, D. Babbucci1, M. Baglione2,