This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

Soil Dynamics and Earthquake Engineering 28 (2008) 866–874

From regional seismic hazard to ‘‘scenario earthquakes’’ for seismicmicrozoning: A new methodological tool for the Celano Project

B. Pace�, P. Boncio, F. Brozzetti, G. Lavecchia, F. Visini

Geodynamics and Seismogenesis Laboratory, Dipartimento di Scienze della Terra, Universita ‘‘G. D’Annunzio’’ Chieti-Pescara,

Campus Universitario Madonna delle Piane, 66013 Chieti Scalo (CH), Italy

Accepted 2 November 2007

Abstract

We present the results of a probabilistic seismic hazard assessment and disaggregation analysis aimed to understand the dominant

magnitudes and source-to-site distances of earthquakes that control the hazard at the Celano site in the Abruzzo region of central Italy.

Firstly, we calculated a peak ground acceleration map for the central Apennines area, by using a model of seismogenic sources defined on

geological-structural basis. The source model definition and the probabilistic seismic hazard evaluation at the regional scale (central

Apennines) were obtained using three different seismicity models (Gutenberg–Richter model; characteristic earthquake model; hybrid

model), consistent with the available seismological information. Moreover, a simplified time-dependent hypothesis has been introduced,

computing the conditional probability of earthquakes occurrence by Brownian passage time distributions.

Subsequently, we carried out the disaggregation analysis, with a modified version of the SEISRISK III code, in order to separate the

contribution of each source to the total hazard.

The results show the percentage contribution to the Celano hazard of the various seismogenic sources, for different expected peak

ground acceleration classes. The analysis was differentiated for close (distance from Celano o20 km) and distant (distance from Celano

420 km) seismogenic sources. We propose three different ‘‘scenario earthquakes’’, useful for the site condition studies and for the seismic

microzoning study: (1) large (M ¼ 6.6) local (Celano-epicentre distance �16 km) earthquake, with mean recurrence time of �590 years;

(2) moderate (M ¼ 5.5) local (Celano-epicentre distance �7.5 km) earthquake, with mean recurrence time of �500 years; and (3) large

(M ¼ 6.6) distant (Celano-epicentre distance �24 km) earthquake, with mean recurrence time of �980 years.

The probabilistic and time-dependent approach to the definition of the ‘‘scenario earthquakes’’ changes clearly the results in

comparison to traditional deterministic analysis, with effects in terms of engineering design and seismic risk reduction.

r 2007 Elsevier Ltd. All rights reserved.

Keywords: Seismic hazard; Seismogenic sources; Time-dependency; Seismic microzoning; Scenario earthquake; Disaggregation analysis

1. Introduction

We present a multidisciplinary approach to obtaindifferent ‘‘scenario earthquakes’’ for the seismic microzon-ing of the Celano town (Abruzzo, Italy), during a project(‘‘Celano Project’’) which involved different researchgroups (e.g., Refs. [1,2]).

It is common practice in seismic microzoning studies touse ‘‘scenario earthquakes’’ which are well-known historical

or instrumentally recorded significant earthquakes, possi-bly occurred near the area of interest or having seismotec-tonic affinity with the region within which the area islocated. In the case of the town of Celano, the most easiestway is to choose the Avezzano earthquake of January 13,1915 (Ms=7.0, according to Ref. [3]; see also Refs. [4,5] forthe source parameters discussion), located only 8 km SSWof Celano. This earthquake was one of the most destructiveevents in the seismic history of Italy (epicentral Intensity ofXI on the MCS scale); it produced damages up to IX MCSin the Celano site. Nevertheless, basic principles of earth-quake geology suggest that the fault responsible for the1915 event needs a period of recharge; therefore, this is not

ARTICLE IN PRESS

www.elsevier.com/locate/soildyn

0267-7261/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.soildyn.2007.11.001

�Corresponding author. Tel.: +390871 3556417;

fax: +390871 3556454.

E-mail address: b.pace@unich.it (B. Pace).

Author's personal copy

the only fault which contributes to the seismic hazard atCelano in the near future, but there are other importantactive faults which might strongly contribute to the Celanohazard.

One of the benefits of probabilistic seismic hazardassessment is that it sums the hazard to a site from allrelevant earthquake sources, integrating over all potentialearthquake occurrences and ground motions to estimatethe mean frequency of exceedance of any given spectralacceleration at the site. However, for design analysis orother seismic risk decisions, a single ‘‘scenario earthquake’’(characterized by a single magnitude, distance, and perhapsother parameters) is often desired. Performing geographichazard disaggregation allow us to determine predominantsources of seismic hazard, identifying plausible scenarioearthquakes. Important contributions to seismic hazarddisaggregation methods and applications are found inRefs. [6–11].

In order to consider the possible contribution of activefaults to the Celano hazard, we use a multidisciplinaryprobabilistic seismic hazard model which combine geolo-gical data (fault length, slip rate and paleoearthquake data)with historical and instrumental seismicity data to estimatethe future ground motion.

The seismic hazard calculations are disaggregated, inorder to separate the relative contribution of each source to

the total hazard of the Celano site, and formulate thehypothesis of three different ‘‘scenario earthquakes’’: largelocal earthquake (M46.0; distanceo20 km); moderatelocal earthquake (Mo6.0; distanceo20 km) and largedistant earthquake (M46.0; distance420 km).

2. Seismotectonic context

The town of Celano is located in the central part of theApennines of Italy, a Late Miocene–Early Pliocene fold-and-thrust belt post-dated by Plio-Quaternary extensionaltectonics. Extension produced normal faults with dip-slipto normal-oblique kinematics, striking on averageNNW–SSE and determining the formation of intra-montane continental basins (Fig. 1). The Fucino basin,extending for �250 km2 south of Celano, is one of the mainexpression of Plio-Quaternary extension in central Apen-nines (see Ref. [12] for a comprehensive stratigraphic andtectonic description). Normal faults are still active alongthe Apennines, as testified by geologic and morphologicevidence ([13–19] and references therein), paleoseismology(e.g., Refs. [20–23]) as well as by strong historical earth-quakes with associated surface faulting (e.g., Ref. [24]) andrecent earthquake focal mechanisms ([25–30]; Fig. 1). TheAvezzano earthquake of January 1915, (Ms ¼ 7.0) locatedin the Fucino basin, demonstrated the primary role played

ARTICLE IN PRESS

Fig. 1. Shaded relief of central-northern Italy with active normal faults and earthquake focal mechanisms within the Apennine belt; focal mechanisms are

for earthquakes with magnitudeX4.5 from 1915 to 2002 (data selection from Ref. [16], implemented); cited sources: CMT [17]; FA97 [18]; D97 [19]; GA85

[20]; R89 [21]; and WV89 [5].

B. Pace et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 866–874 867

Author's personal copy

by normal faults in the Apennines seismogenesis. Theearthquake produced SE-striking SW-dipping normalfaulting surface ruptures along the fault system borderingthe Fucino basin to the NE, from the southeastern edge ofthe basin to near Celano, for a total length of �29 km[5,31–34].

In recent years, several works were dedicated to theidentification and geometric–kinematic–energetic charac-terization of active seismogenic faults in the Apennines,mainly for seismic hazard purposes (see the compilationsby Refs. [16,17,19] and references therein). Fig. 2a showsthe model of seismogenic sources that we have used for thepresent analysis; it was proposed by Ref. [35], to whom(including references therein) we refer for details onseismotectonic constraints. This model was the firstattempt to constrain in a systematic way the seismogenicfaults in 3D, by integrating surface geological data withsubsurface (deep geological–geophysical prospecting data,detailed earthquake data, thermo-mechanical properties ofcrustal rocks) data. The 3D seismogenic sources can beused as input parameters for geology-based seismic hazardcomputations. Each source corresponds to a ‘‘seismogenicmaster fault’’, that is a major fault structure consideredcontinuous at depth for several kilometers. The seismo-genic master faults are constrained by one or more of thefollowing requirements: (a) geologic evidence of repeateddisplacement episodes during late Quaternary times (e.g.,last 125 ky); (b) paleoseismological evidence of historical orprehistoric earthquakes; and (c) activation during instru-mental seismic sequences and/or clear association tohistorical earthquakes. The seismogenic master faults areseparated from each other by first-order structural orgeometrical complexities, e.g., complexities of kilometricdimensions (likely barriers to the rupture propagation).The 3D geometry of the master faults is represented by the‘‘seismogenic boxes’’, that are the surface projection of theseismogenic master faults. The shape and dimension ofeach box strictly reflect the 3D geometry of the relatedmaster fault. The width (Ws in Fig. 2b) is not constant inthe area, being dependent on the down-dip length of theseismogenic structure (W in Fig. 2b) and, therefore, on theinclination at depth of the fault plane and on the depth ofthe local seismogenic layer (D in Fig. 2b).

The seismogenic normal faults of central Apennines dipto the WSW and are spread along three main alignmentswith average NNW–SSE strike. The eastern alignmentextends from Mt. Vettore to Mt. Gorzano and Gran Sasso;the intermediate alignment extends from Norcia toSulmona to the 5 Miglia basin; the western alignmentextends from Salto Valley to the Fucino basin and Barrea.At the Celano site, located along the western alignment,three master faults converge. One is the fault activated bythe 1915 Avezzano earthquake (no. 15 in Fig. 2a). Theother two are the M. Velino-Magnola (no. 13 in Fig. 2a)and Campo Felice-Ovindoli (no. 14 in Fig. 2a) faults. Alarge historical earthquake, probably able to activate theentire fault, can be found for the Campo Felice-Ovindoli

structure (1349 earthquake, Intensity IX–X on the MCSscale, M ¼ 6.5; [36]). The M. Velino-Magnola fault has alikely associated event of only moderate magnitude,therefore unable to rupture the entire fault (1904 earth-quake, Intensity VIII–IX, M ¼ 5.5). Besides the earth-quakes cited above, a number of destructive events

ARTICLE IN PRESS

Fig. 2. (a) Map of the seismogenic master faults and related seismogenic

boxes for the central Apennines, with epicentres of historical earthquakes

larger than VIII on the MCS scale (MX5.7) from CPTI catalogue [30];

dates refer to the year of earthquake occurrence; (b) representation of a

seismogenic box: Ws, box width; L, along-strike length of the seismogenic

master fault; W, down-dip length of the seismogenic master fault; and D,

thickness of the local seismogenic layer.

B. Pace et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 866–874868

Author's personal copy

characterized the seismic activity of the active faults duringhistorical times (Fig. 2a; see Ref. [35] and references thereinfor a discussion on earthquake-structure association), whilesome active faults have been still silent since historicaltimes (e.g., faults no. 1, 3, 10, 12, and 16 in Fig. 2a).

3. Probabilistic seismic hazard analysis

Probabilistic seismic hazard analysis (PSHA) integratesoverall potential earthquake occurrence and groundmotions to estimate the mean frequency of exceedance ofany given spectral acceleration at the site. In this paper,starting from the above described seismogenic boxes, wedid studies finalized to the seismic hazard evaluation(expected peak ground acceleration maps) and so to thechoice and definition of the ‘‘scenario earthquakes’’ for theCelano site.

Integrating the above-discussed seismogenic sourcemodel, based on the geological and geophysical availabledata, with all the seismological information regarding theoccurrences of earthquakes, we obtained probabilisticseismic hazard maps of the area surrounding Celano town.The approach for the seismic hazard evaluations is the onealready used in central Italy [37–39], that permits toquantify the expected seismicity of the studied area startingfrom the geometrical and energetic characteristics of thegeologically defined sources.

The first step was the computation of the maximumexpected magnitude (Mmax), on the basis of somegeometrical data, such as the maximum length of eachseismogenic box and the maximum rupture area of thesource. In particular, we calculated the Mmax values via the

use of empirical relationships, calibrated on normalfaulting [40], and relationships between fault dimensionsand seismic moment (considering constant strain drop of3� 10�5 from Ref. [41]; see Ref. [39] for details). The finalchoice was a weighted average value of Mmax (20% Mmax

obtained from the maximum length; 40% Mmax from therupture area; 40% Mmax from the seismic moment). Thefinal Mmax values for each seismogenic box are reported inTable 1.The second step was the evaluation of the mean

recurrence time (TMmax) of the expected maximummagnitude on each seismogenic box. For the studied areamultiples observations of earthquakes inter-times on asingle structure are not available, so the mean recurrencetimes were obtained from the sources parameterization,using all the available paleoseismological information, suchas slip rate and slip per event values. In particular, theTMmax values were obtained using two different techniques:the first is the ratio of slip per event over the slip rate; thesecond approach obtains the TMmax values by using thecriteria of the ‘‘segment seismic moment conservation’’,proposed by Ref. [42] (see Ref. [39] for details). The finalvalues, used in seismic hazard computations, are theaverage values between the two and are reported inTable 1.The Mmax and TMmax values alone are not sufficient to

quantify the whole seismic activity of a source, so the nextstep in the seismic hazard assessment was the definition of acomplete recurrence model. Comparing the predictedoccurrences with the seismological information, we chosedifferent recurrence models for each seismogenic box.The choice was between two extreme models for the

ARTICLE IN PRESS

Table 1

Energetic parameters of the seismogenic boxes in central Apennines (Italy)

Seismogenic box Mmax TMmax Recurrence model Tdep Last event

1. Bove-Vettore 6.8 1190 CH Yes 4000 years BPa

2. Gorzano 6.7 1785 HY Yes 8000 years BPa

3. Gran Sasso 6.7 673 CH Yes 3400 years BPa

4. Nottoria-Preci 6.6 1149 HY Yes Jan. 14, 1703

5. Cascia-Cittareale 6.6 770 HY Yes Jan. 14, 1703

6. Montereale 6.4 699 CH Yes Jan. 16, 1703

7. Pizzoli-Pettino 6.6 824 CH Yes Feb. 02, 1703

8. Paganica 6.3 643 CH Yes Nov. 26, 1461

9. Media V. Aterno 6.6 1649 CH No –

10. Sulmona 6.6 980 CH Yes II cent. A.D.

11. Pizzalto-Cinquemiglia 6.5 1092 CH Yes 930 years BPa

12. Valle del Salto 6.5 827 CH No –

13. Velino-Magnola 6.6 918 HY No –

14. Campo Felice-Ovindoli 6.6 588 CH Yes Sep. 09, 1349

15. Fucino 6.6 654 CH Yes Jan. 13, 1915

16. M. Marsicano 6.5 703 CH No –

17. Barrea 6.4 1140 GR No –

Columns stand for: Mmax, maximum expected magnitudes (Mw); TMmax, mean recurrence times of the related maximum expected earthquakes (years),

recurrence model, type of model chosen for the seismogenic boxes as follows: GR, Gutenberg–Richter exponential distribution function; CH, bell-shaped

approximation of a characteristic earthquake model; HY, hybrid model with a characteristic peak and a GR tail for smaller magnitude; Tdep, with yes are

highlighted the sources for which we used a time-dependent approach; last event, dates of the major last event for each box.aElapsed times from the last paleoseismological earthquake (for details about the earthquakes/sources associations see Refs. [29,33]).

B. Pace et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 866–874 869

Author's personal copy

seismogenic behaviour of faults: the ‘‘characteristic earth-

quake model’’ [43], and the ‘‘Gutenberg–Richter relation-

ship’’ (G–R) of Ref. [44]. As in Ref. [39], we decided one ofthese two well-known earthquake recurrence models,chosen on the basis of the seismological observations oneach seismogenic box. In some cases, we modelled thequeue of low magnitude by a small tail of G–R exponentialdistribution anchored to the rate of occurrence of small-to-moderate events (we name this approach ‘‘Hybrid’’according to the definition given by Ref. [45]). The typeof model used for all the seismogenic boxes is indicated inTable 1.

The last step in our approach to the seismic hazardevaluation was to introduce time-dependence where thetime elapsed since the last maximum event is known. Weadopted the simplest time-dependent process, namely therenewal one, using the formulation of Brownian passagetime (BPT) distributions, one of the most physicallymotivated models that has appeared in recent literature[46]. In Table 1, we point out the sources where the date ofthe last maximum event is known and so where we used atime-dependent approach (see Ref. [39] for details). Inparticular, the time elapsed since the last event is used todetermine the conditional probability of having an event inthe next 50 years. Then, using the simplification proposed

by Ref. [45], equivalent fictitious seismicity rates have beenmerged into the seismic hazard code. The differencesbetween a poissonian approach and a time-dependent oneare quite important (see Ref. [39] for details), and thecontribution of the recently active source, like the Fucinoone (Box 15), vanishes in the overall seismic hazard,whereas some other sources, for the high BPT conditionalprobabilities, became the most hazardous sites (e.g., Box 11Sulmona and Box 14 Campo Felice-Ovindoli).The final seismic hazard computation was done using the

well-known SEISRISK III code [47]. The results (Fig. 3)are expressed in peak ground acceleration (PGA) expectedto be exceeded with a probability of 10% in 50 years, in‘‘bedrock’’ conditions. Attenuation functions are thoseproposed by Ref. [48], based on empirical regressions of theEuropean strong-motion data. The Celano site givesexpected shaking values included between 0.3 and 0.4 g,without considering the standard deviation in attenuation,and the area around the site gives maximum values higherthen 0.4 g (Fig. 3).Near field effects are not considered in these calculations,

inasmuch as this possible effects have been counted by theother Researchers involved in the ‘‘Celano Project’’, whoused the defined ‘‘scenario earthquakes’’ for seismicmicrozoning purposes (e.g., Refs. [1,2]).

ARTICLE IN PRESS

Fig. 3. Map of peak ground acceleration (PGA) expected not to be exceeded in 50 years at 90% probability level in the central Apennines (Italy) region,

without standard deviation in attenuation. The legend reports a correlation between PGA values and macroseismic intensities (MCS), derived from Italian

literature [50,51].

B. Pace et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 866–874870

Author's personal copy

4. Disaggregation of the PSHA for the Celano site

The above-described regional seismic hazard calculationwas disaggregated in order to determine the predominantsources of seismic hazard, useful for the subsequentmicrozoning studies [2]. The disaggregation analysis ofthe regional seismic hazard is a relative recent techniqueuseful to generate ‘‘scenario earthquakes’’ [8,11]. Two mainmethods of disaggregation of hazard should be distin-guished. The first method separates the contributions into alimited number of ranges of values of distance, magnitudeand ground-motion uncertainty [8]. The second method,named ‘‘geographic disaggregation’’ of hazard, separatesthe contributions into ranges of location (latitude andlongitude), magnitude, and ground-motion uncertainty([10]; see also Ref. [11] for an application in the UnitedStates). This latter method allows us to identify specificfaults or single seismogenic sources as most probablesources contributing to the hazard.

For the ‘‘Celano Project’’, we decided to use thegeographic disaggregation method, using an implementa-tion by LaForge [49] of the SEISRISK III code. Theimplemented code allowed us to calculate the percentagecontribution of each seismogenic source to the total hazardin terms of both mean values and values increased by astandard deviation of the attenuation relationships. In thiswork, we used the mean PGA values (without standarddeviation of the attenuation relationships) obtained in theregional probabilistic seismic hazard calculation. Wedecided to not consider the standard deviation of theattenuation relationships because the reference shaking isexpected to be further amplified/deamplified by computingthe local seismic response later in the ‘‘Celano Project’’.Fig. 4 shows the results of the application of thismethodology. In particular, the plot in Fig. 4a show thepercentage contribution to the Celano hazard of eachseismogenic source for different expected PGA classes.We differentiated ‘‘local sources’’ (Fig. 4b, Celano-sourcedistance o20 km) from ‘‘distant sources’’ (Fig. 4c, Celano-source distance 420 km).

On the basis of the disaggregation analysis, we chose theseismogenic sources liable for three different ‘‘scenarioearthquakes’’ useful for site condition studies (Fig. 5 andTable 2):

(1) large local earthquake;(2) moderate local earthquake; and(3) large distant earthquake.

Large earthquake suggests a magnitude 46.0 and localsuggests an epicentral distance from Celano site o20 km.Fig. 5 and Table 2 show the locations and the hypo- andepi-central parameters of the three chosen ‘‘scenarioearthquakes’’. In Table 2, we reported the expectedmagnitude, estimated from the expected rupture area, theepicentral coordinates, the most likely hypocentral depth,the Celano-epicentre distance and the Celano-epicentre

ARTICLE IN PRESS

Fig. 4. Disaggregation plots showing the contribution to hazard for each

seismogenic box from four different peak ground acceleration classes and

for 475 years return period (corresponding to values expected not to be

exceeded in 50 years at 90% probability level). The legend reports a

correlation between PGA values and macroseismic intensities (MCS),

derived from Italian literature [50,51]. (a) All sources; (b) local sources,

(c) distant sources.

B. Pace et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 866–874 871

Author's personal copy

azimuth. We have also evaluated the mean recurrence timesof the three earthquakes, considering the different slip ratevalues associated to the referred seismogenic sources ([39]and references therein), and calculated as described inparagraph 3. For the two large earthquakes, the recurrencetime values are related to the maximum expected earth-quake (see Table 1), and for the moderate earthquake wehave used the same approach, considering the expectedmagnitude as a fictitious maximum magnitude.

In details, the ‘‘scenario earthquakes’’ are the following:

(1) Large local earthquake: for local seismogenic sources(distance from Celano o20 km) and PGA valuesexpected in the Celano site between 0.2 and 0.3 g(corresponding to macroseismic intensity between VIIand IX on the MCS scale; [50,51]), the most probablyactivable structure is the Campo Felice-Ovindoliseismogenic source (no. 14 in Figs. 2 and 4b). Theexpected earthquake, hypothesizing a complete ruptureof the seismogenic structure, has a magnitude of 6.6(Mw) and a mean recurrence time of �590 years. The

expected hypocentre was arbitrarily fixed in the centraldeepest part of the structure, supposing a rupturepropagating upwards and bilaterally, with a hypocen-tral depth of �12 km. The epicentral distance fromCelano is �16 km (Fig. 5 and Table 2).

(2) Moderate local earthquake: for local seismogenicsources (distance from Celano o20 km) and PGAvalues between 0.02 and 0.2 g (macroseismic intensitybetween III and VII–VIII MCS), the most probablyactivable structure is the Velino-Magnola seismogenicsource (no. 13 in Figs. 2 and 4b). In this case, we cannotconsider the maximum expected magnitude for thissource, because is not matching with the PGA classeswe are exploring for a moderate expected event. Forthese reasons, the magnitude of the expected earth-quake was evaluated to be 5.5 (Mw), for analogy withthe 1904 earthquake that ruptured a portion of theVelino-Magnola structure, with a related mean recur-rence time of �500 years. The equivalent rupture areaof �35 km2 was placed in the deepest part of thestructure, closest to Celano (for precautionary reasons),supposing a circular rupture with hypocentral depth of�10 km. The epicentral distance from Celano is�7.5 km (Fig. 5 and Table 2).

(3) Large distant earthquake: for distant seismogenicsources (distance from Celano 420 km) and PGAvalues between 0.1 and 0.2 g (macroseismic intensitybetween VI and VII–VIII MCS), the most probablyactivable structure is the Sulmona seismogenic source(no. 10 in Figs. 2 and 4c). The expected earthquake,hypothesizing a complete rupture of the seismogenicstructure, has a magnitude of 6.6 (Mw) and a meanrecurrence time of �980 years. Also in this case, theexpected hypocentre was arbitrarily fixed in the centraldeepest part of the structure, with a hypocentral depthof �14 km. The epicentral distance from Celano is�24 km (Fig. 5 and Table 2).

5. Conclusions

The present work, performed during a multidisciplinaryproject directed to the seismic risk assessment in the urbanarea of Celano (‘‘Celano Project’’), was aimed to thechoice and the characterization of some ‘‘scenario earth-quakes’’, useful for site effects evaluation. We proposed

ARTICLE IN PRESS

Fig. 5. Structural sketch of the area surrounding the town of Celano with

location of the three scenario earthquakes: A, large local earthquake; B,

moderate local earthquake; and C, large distant earthquake. Hypocentral

and energetic parameters of the three earthquakes are listed in Table 2.

Table 2

Hypocentral parameters of the three scenario earthquakes

Epicentral coordinates Hypocentre

depth (km)

Celano-epicentre

distance (km)

Celano-epicentre

azimuth (1)

Expected

magnitude (Mw)

Mean recurrence

time (year)Latitude (1) Longitude (1)

A 13.38 42.15 12 16 N300 6.6 588

B 13.48 42.04 10 7.5 N230 5.5 500

C 13.83 42.03 14 24 N105 6.6 980

A, large local earthquake; B, moderate local earthquake; C, large distant earthquake. Epicentral location in Fig. 5.

B. Pace et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 866–874872

Author's personal copy

three different ‘‘scenario earthquakes’’: large local earth-quake (Mw ¼ 6.6 on the Campo Felice-Ovindoli seismo-genic box), moderate local earthquake (Mw ¼ 5.5 onthe Velino-Magnola seismogenic box) and large distantearthquake (Mw ¼ 6.6 on the Sulmona seismogenic box),used directly in the subsequent numerical site effectscalculations [2].

In order to define the three ‘‘scenario earthquakes’’, weutilized the technique of the ‘‘geographic disaggregation’’[10,49] of a regional probabilistic seismic hazard assess-ment, for the determination of the location of the mostprobable sources contributing to the hazard of the site.Normally, in microzoning studies the choice of the‘‘scenario earthquakes’’ is essentially driven by the knowninstrumental and historical seismicity localized around thesite of interest. Moreover, using this classical approach, theseismic hazard evaluation is based on poissonian estimates,without taking into account the geometric, kinematic,energetic characteristics and time-dependency of theseismogenic sources. In this paper, we present a multi-disciplinary analysis and so a new methodological tool toobtain probabilistic seismic hazard evaluations for theinvestigated site, combining structural-geological data withthe information obtained from the instrumental andhistorical seismicity. The regional ‘‘geological’’ seismichazard evaluation permitted to ascribe to the definedseismogenic sources a ‘‘memory’’ of their seismic history,so that the probability of activation of the seismogenicstructures depends on the elapsed time from the lastearthquake having the maximum expected magnitude,using a renewal model.

This kind of approach permits to obtain more detailedseismic hazard assessments in areas where we have at ourdisposal a time window of seismicity information which isnot long enough to have detected the complete seismiccycle of the active structures. Moreover, the subsequentdisaggregation analysis permits to determine distance,average magnitude and site-to-source azimuth of theearthquakes that control the hazard of the interested site.We believe that this probabilistic methodology is useful forunderstanding the earthquakes that are likely to dominatethe hazard at the site over time, and so useful for thepurposes of engineering design and the seismic riskreduction.

Acknowledgments

This study was carried on in the frame of the agreement‘‘Analisi sismotettonica di dettaglio con caratterizzazione del

terremoto di progetto, finalizzata alle indagini di micro-

zonazione sismica del comune di Celano’’, between RegioneAbruzzo-Dir. O.O.P.P. e Protezione Civile (Resp. Dott. E.Iannarelli), Comune di Celano (Resp. Ing. V. Antenucci)and Dipartimento di Scienze della Terra-GeoSisLab Chieti(Resp. Prof. G. Lavecchia). We are very grateful to twoanonymous reviewers and the Guest Editor George E.Mylonakis for their valuable comments on the manuscript.

References

[1] Cardarelli E, Cercato M, de Nardis R, Di Filippo G, Milana G. A

workflow for geophysical survey planning and execution oriented to

seismic risk assessments. Soil Dyn Earthquake Eng, submitted for

publication.

[2] Pergalani F, Compagnoni M, Petrini V. Evaluation of site effects

using numerical analyses in Celano (Italy) finalized to seismic risk

assessment. Soil Dyn Earthquake Eng, submitted for publication.

[3] Margottini C, Ambraseys NN, Screpanti A. La magnitudo dei

terremoti italiani del XX secolo. Rome: ENEA; 1993. 57p.

[4] Amoruso A, Crescentini L, Scarpa R. Inversion of source parameters

from near- and far-field observations: an application to the 1915

Fucino earthquake, central Apennines, Italy. J Geophys Res 1998;

103:29989–99.

[5] Ward SN, Valensise GR. Fault parameters and slip distribution of

the 1915 Avezzano, Italy, earthquake derived from geodetic

observations. BSSA 1989;79:690–710.

[6] Stepp JC, Silva WJ, McGuire RK, Sewell RW. Determination of

earthquakes design loads for a high level nuclear waste repository

facility. In: Proceedings of the natural phenomena hazards mitiga-

tion conference, Atlanta, Gergia, 19–22 October 1993, vol. II.

P. 651–57.

[7] Chapman MC. A probabilistic approach for ground motion selection

for engineering design. BSSA 1995;85:937–42.

[8] McGuire RK. Probabilistic seismic hazard analysis and design

earthquakes: closing the loop. BSSA 1995;85:1275–84.

[9] Cramer CH, Petersen MD. Predominant seismic source distance and

magnitude maps for Los Angeles, Orange, and Ventura Counties,

California. BSSA 1996;86:1645–9.

[10] Bazzurro P, Cornell CA. Disaggregation of seismic hazard. BSSA

1999;89:501–20.

[11] Harmsen S, Frankel A. Geographic deaggregation of seismic hazard

in the United States. BSSA 2001;91:13–26.

[12] Cavinato GP, Carusi C, Dall’Asta M, Miccadei E, Piacentini T.

Sedimentary and tectonic evolution of Plio-Pleistocene alluvial and

lacustrine deposits of Fucino basin (central Italy). Sedimentary Geol

2002;148:29–59.

[13] Calamita F, Pizzi A. Recent and active extensional tectonics in the

southern umbro-marchean Apennines (central Italy). Memorie della

Societa Geololgica Italiana 1994;48:541–8.

[14] Blumetti AM, Dramis F, Michetti AM. Fault-generated mountain

fronts in the central Apennines (central Italy): geomorphological

features and seismotectonic implications. Earth Surf Processes

Landforms 1993;18:203–29.

[15] Boncio P, Brozzetti F, Lavecchia G. Architecture and seismotectonics

of a regional low-angle normal fault zone in central Italy. Tectonics

2000;19:1038–55.

[16] Galadini F, Galli P. Active tectonics in the Central Apennines

(Italy)—input data for seismic hazard assessment. Nat Hazards

2000;22:225–70.

[17] Galadini F, Meletti C, Vittori E. Stato delle conoscenze sulle faglie

attive in Italia: elementi geologici di superficie. In: Galadini F, Meletti

C, Rebez A, editors. Le ricerche del GNDT nel campo della

pericolosita sismica (1996–1999). Roma: CNR-Gruppo Nazionale

per la Difesa dai Terremoti; 2000. p. 107–36.

[18] Pizzi A, Scisciani V. Methods for determining the Pleistocene-

Holocene component of displacement on active faults reactivating

pre-Quaternary structures: examples from the central Apennines

(Italy). J Geodyn 2000;29:445–57.

[19] Valensise G, Pantosti D. Database of potential sources for earth-

quakes larger than M 5.5 in Italy. Annali di Geofisica 2001;

44(Suppl.):180 with CD-ROM.

[20] Pantosti D, D’Addezio G, Cinti FR. Paleoseismicity of the Ovindoli-

Pezza fault, central Apennines, Italy: a history including a large,

previously unrecorded earthquake in the Middle ages (860–1300

A.D.). J Geophys Res 1996;101:5937–59.

ARTICLE IN PRESSB. Pace et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 866–874 873

Author's personal copy

[21] D’Addezio G, Masana E, Pantosti D. The Holocene paleoseismicity

of the Aremogna-Cinque Miglia Fault (Central Italy). J Seismol

2001;5:181–205.

[22] Galadini F, Galli P, Moro M. Paleoseismology of silent faults in the

Central Apennines (Italy): the Campo Imperatore Fault (Gran Sasso

Range Fault System). Ann Geophys 2003;46(5):793–813.

[23] Galadini F, Galli P. Paleoseismology of silent faults in the Central

Apennines (Italy): the Mt. Vettore and Laga Mts. Faults. Ann

Geophys 2003;46(5):815–36.

[24] Blumetti AM. Neotectonic investigations and evidence of paleoseis-

micity in the epicentral area of the January–February 1703, central

Italy, earthquakes. In: Serva L, editor. Perspectives in paleoseismol-

ogy, vol. 6, 1995. p. 83–100.

[25] Montone P, Mariucci MT, Pondrelli S, Amato A. An improved stress

map for Italy and surrounding regions (Central Mediterranean).

J Geophys Res 2004;109:B10410.

[26] Harvard CMT catalogue available from /www.seismology.harvar-

d.eduS or MedNet regional CMT available from /www.mednet.

ingv.itS.

[27] Frepoli A, Amato A. Contemporaneous extension and compression

in the Northern Apennines from earthquake fault-plane solutions.

Geophys J Int 1997;129:368–88.

[28] Di Lucio F, Amato A, Azzara RM, Basili A, Delladio A, Di Bona M,

et al. La sequenza sismica del maggio 1997 a Massa Martana (PG).

Riassunti delle comunicazioni, 161 Convegno GNGTS, Roma, 1997.

p. 65.

[29] Gasparini C, Iannaccone G, Scarpa R. Fault-plane solutions and

seismicity of the Italian peninsula. Tectonophysics 1985;117:59–78.

[30] Riguzzi F, Tertulliani A, Gasparini C. Study of the seismic sequenze

of Porto San Giorgio (Marche), 3 July 1987. Il Nuovo Cimento

1989;12:453–67.

[31] Oddone E. Gli elementi fisici del grande terremoto marsicano fucense

del 13 Gennaio 1915. Bollettino della Societa Sismologica Italiana

1915;19:71–291.

[32] Serva L, Blumetti AM, Michetti AM. Gli effetti sul terreno del

terremoto del Fucino (13 gennaio 1915); tentativo di interpretazione

della evoluzione tettonica recente di alcune strutture. Memorie della

Societa Geologica Italiana 1986;35:893–907.

[33] Michetti AM, Brunamonte F, Serva L, Vittori E. Trench investiga-

tions of the 1915 Fucino earthquake fault scarps (Abruzzo, central

Italy): geological evidence of large historical events. J Geophys Res

1996;101:5921–36.

[34] Galadini F, Galli P. The Holocene paleoearthquakes on the

1915 Avezzano earthquake faults (central Italy); implications for

active tectonics in the central Apennines. Tectonophysics 1999;308:

143–70.

[35] Boncio P, Lavecchia G, Pace B. Defining a model of 3D seismogenic

sources for Seismic Hazard Assessment applications: the case of

central Apennines (Italy). J Seismol 2004;8:407–25.

[36] Working Group CPTI. Catalogo Parametrico dei Terremoti Italiani,

version 2004 (CPTI04). INGV, Bologna /http://emidius.mi.ingv.it/

CPTI/S.

[37] Peruzza L, Pace B. Sensitivity analysis for seismic source character-

istics to probabilistic seismic hazard assessment in central Apennines

(Abruzzo area). Bollettino di Geofisica Teorica ed Applicata 2002;

43:79–100.

[38] Pace B, Peruzza L, Boncio P, Lavecchia G. Seismogenic sources in

Central Italy: from causes to effects. Memorie della Societa

Geologica Italiana 2002;57:419–29.

[39] Pace B, Peruzza L, Lavecchia G, Boncio P. Layered seismogenic

source model and probabilistic seismic hazard analyses in Central

Italy. BSSA 2006;96:107–32.

[40] Wells DL, Coppersmith KJ. New empirical relationships among

magnitude, rupture length, rupture width, rupture area, and surface

displacement. BSSA 1994;84:974–1002.

[41] Selvaggi G. Spatial distribution of horizontal seismic strain in the

Apennines from historical earthquakes. Annali di Geofisica 1998;41:

241–51.

[42] Field EH, Johnson DD, Dolan JF. A mutually consistent seismic-

hazard source model for Southern California. BSSA 1999;89:559–78.

[43] Schwartz DP, Coppersmith KJ. Fault behavior and characteristic

earthquakes: examples from the Wasatch and San Andreas Falt

Zones. J Geophys Res 1984;89:5681–98.

[44] Gutenberg B, Richter CF. Frequency of earthquakes in California.

BSSA 1944;34:185–8.

[45] Wu SC, Cornell CA, Winterstein SR. A hybrid recurrence model and

its implication on seismic hazard results. BSSA 1995;85:1–16.

[46] Matthews MV, Ellsworth WL, Reasenberg PA. A Brownian model

for recurrent earthquakes. BSSA 2002;92:2250–93.

[47] Bender B, Perkins DM. SEISRISK III; a computer program for

seismic hazard estimation. Reston, VA, USA: US Geological Survey;

1987. p. 48.

[48] Ambraseys NN, Simpson KA, Bommer JJ. Prediction of horizontal

response spectra in Europe. Earthquake Eng Struct Dyn 1996;25(4):

371–400.

[49] LaForge R. Implementation of SEISRISK III, USGS, Colorado,

Boulder, Open File, 1996.

[50] Decanini L, Gavarini C, Mollaioli F. Proposta di definizione delle

relazioni tra intensita macrosismica e parametri del moto del suolo.

Siena: L’ingegneria Sismica in Italia; 1995. p. 63–72.

[51] Margottini C, Molin D, Serva L. Intensity versus ground motion: a

new approach using Italian data. Eng Geol 1992;33:45–58.

ARTICLE IN PRESSB. Pace et al. / Soil Dynamics and Earthquake Engineering 28 (2008) 866–874874