numerical modeling of the 13 december 1990 m 5.8 east sicily earthquake at the catania...

241

Bulletin of the Seismological Society of America, Vol. 95, No. 1, pp. 241–251, February 2005, doi: 10.1785/0120030126

Numerical Modeling of the 13 December 1990 M 5.8 East Sicily

Earthquake at the Catania Accelerometric Station

by Giovanna Laurenzano and Enrico Priolo

Abstract The record of Catania ENEA–ENEL accelerometric station, with itsanomalously high amplitude, represents an outlier for the dataset of seismogramsrecorded by the Italian accelerometric network during the 13 December 1990 M 5.8East Sicily earthquake. Using numerical modeling tools, we corroborate the inter-pretation of this anomaly as an effect of the crustal structure or site response, ratherthan as a finite source effect.

In the first part of this study, we compare the recorded seismograms to thosecomputed by two numerical methods—the 2D Spectral Element Method (SPEM), atechnique which solves the 2D full-wave propagation through a complex geologicalstructure; and the Wavenumber Integration Method (WIM), which solves the 3D full-wave propagation in a horizontally-layered structure. The comparison shows that therecorded waveforms are reproduced accurately using a laterally heterogeneous struc-ture, whereas none of the representations in terms of horizontal plane layers, whichsimplify the same structure, provides satisfactory results. Furthermore, we comparethe horizontal-to-vertical spectral ratios (HVSR) obtained from (1) the earthquakerecord, (2) the seismogram simulated by the SPEM, and (3) the seismic noise recordedat the same site. Again, the overall agreement is very good.

This study has several outcomes. Firstly, that the numerical modeling approachbased on the SPEM, and used in a previous study to simulate ground shaking for adestructive scenario earthquake, provides reliable results. Secondly, that the 2Dmodel used to represent the crustal structure beneath this area is realistic. Indeed,simplified 1D models may not be an adequate means to reproduce realistic seismo-grams and predict the ground motion in the frequency band of interest for seismichazard (i.e., 0.5–10 Hz). Finally, the high amplitude displayed by the Catania stationduring the 1990 earthquake is explained as a combined effect of site and structure-path, while finite source models appear unnecessary. In general, this study empha-sizes the importance of methods that accurately model the wavefield propagationthrough realistic geologic structures for predicting ground motion.

Introduction

Catania is one of the Italian cities more exposed to seis-mic risk. Two large earthquakes struck the surrounding areain the last thousand years: the MS 7.3 in 1169 and the MS �7.0 in 1693 (Monachesi and Stucchi, 1997). The latter earth-quake caused as many as 54,000 deaths in East Sicily (Bos-chi et al., 1995). The absence of strong earthquakes for along time period (about three hundred years), the high den-sity of population, industries, and infrastructures, and thefact that earthquake-reinforced buildings are by far the mi-nority, are the key factors that contribute to increasing seis-mic risk.

A number of national projects have recently been car-ried out, or are still ongoing, with the goal of estimating andreducing the seismic risk of the city (Faccioli and Pessina,

2000; Maugeri, 2004) and its surrounding areas (Decaniniand Panza, 2000). Within these projects, several studies areaimed at predicting the ground motion and at building upground shaking scenarios, as well as estimating site responseat selected sites. Most studies have focused on developingdestructive scenarios, assuming the MS � 7.0 event of 11January 1693 as a reference earthquake; but some of themalso tried to estimate damaging scenarios using weaker, me-dium sized reference earthquakes, such as the MS 6.2 of 1818and MS 5.5 of 1848 events (Monachesi and Stucchi, 1997).The approaches used to solve the scenario studies range fromempirical methods (Pessina, 1999, 2000) to methods basedon numerical simulations of the seismic wavefield radiated

242 G. Laurenzano and E. Priolo

from the source (Langer et al., 1999a, 1999b; Priolo, 1999,2000; Romanelli and Vaccari, 1999; Romanelli et al., 2000;Zollo et al., 1999; Suhadolc et al., 2000; Zollo and Emolo,2000). The latter significantly improve the detail of the anal-ysis; however, the results obtained should be validated withrecorded data. Demonstrating the effectiveness of the nu-merical technique used in the scenario study by Priolo (1999,2000) is one of the goals of this work.

The identification of the major seismogenic structuresof East Sicily is still subject of debate. Despite few alter-native hypotheses (Sirovich and Pettenati, 1999; Barbanoand Rigano, 2001), most authors agree on identifying theonly structure able to generate earthquakes with magnitudeas large as 7 in the Ibleo–Maltese fault system, which islocated some tens of kilometers offshore from the easternSicilian coast (Bianca et al., 1999; Azzaro and Barbano,2000a, 2000b). Unfortunately, for seismological studies, theseismicity of this structure is rather anomalous, since it fea-tures a very low number of weak to medium sized earth-quakes. The 13 December 1990, M 5.8 earthquake is actuallythe only medium-sized, instrumentally recorded event thatoccurred along the Ibleo–Maltese system which can be usedto validate numerical simulations.

The 13 December 1990, M 5.8 earthquake is associatedto the rupture of a transcurrent segment of the Ibleo–Maltesefault. It has been recorded by eight stations of the ItalianENEA–ENEL accelerometric network (Di Bona et al., 1995)(Fig. 1), as well as by three stations of the MedNet broad-band network. Waveforms recorded by the MedNet stationshave been used for moment-tensor inversion (Giardini et al.,1995), while the strong motion records constrained thesource model (Di Bona et al., 1995). The main features ofthe waveforms were correctly reproduced using a point-source model for all records, with the exception of the ac-celerogram recorded by the Catania station. None of thequoted works points out the need of adopting finite sourcemodels.

The seismograms recorded by the Catania ENEA–ENELstation (CAT in Fig. 1) exhibit relatively large ground ac-celerations. Amplitudes are 2–3 times larger than those re-corded at Sortino, although both stations are at similar epi-central distance (25–30 km). This anomaly has beenattributed (Di Bona et al., 1995) to local site effects and tothe presence of strong crustal heterogeneities, but no argu-ments were provided to support this statement. The presentstudy aims at validating this hypothesis by numericalsimulations. To this purpose, the 2D Chebyshev spectral-element method, which solves the seismic full-wave prop-agation through a complex geological structure, representsa suitable tool of investigation.

The outline of this article is as follows. First, we de-scribe the 2D spectral-element simulation of the 13 Decem-ber 1990, M 5.8 East Sicily earthquake, and compare theresults of the simulations to the recorded data. Then, we tryto evaluate the importance of the 2D structure, (i.e., whetherthe recorded data can be reproduced adequately using a

plane layer structure). In this part, we model the earthquakeusing the Wavenumber Integration Method (Herrmann,1996a, 1996b). In the third and last part, we evaluate the thesite response at the Catania station by looking at the hori-zontal-to-vertical spectral ratios (HVSRs) computed from therecorded and synthetic seismograms, as well as those esti-mated from records of seismic noise.

The 13 December 1990 M 5.8 East Sicily Earthquake

The M 5.8 earthquake that struck East Sicily on 13 De-cember 1990, caused widespread damage (I � VII–VIIIMercalli maximum intensity) and 19 casualties. This eventhas been the only instrumentally-recorded earthquake asso-ciated with the Ibleo–Maltese fault system. This motivatedseveral geological, seismological, and geodetical studies, anumber of which are reported in a special issue of the Annalidi Geofisica (Basili et al., 1995).

Table 1 summarizes the seismological parameters esti-mated for the main shock by different studies as well as thevalues assumed in this study. The earthquake occurred about8–10 km offshore of the promontory of Augusta (Siracusa),at a hypocentral depth of about 15–20 km. In this study,among the several available fault-mechanism solutions, weadopt the one estimated by Amato et al. (1995), using themoment tensor solution obtained by Giardini et al. (1995),which corresponds to a nearly pure strike slip, with east-west orientation and left lateral motion. This mechanismsuggests that this event can be associated to a dislocation oftwo large segments of the Ibleo–Maltese fault system, whichis nearly north-south oriented and undergoes extensional de-formation. As concerns seismic moment and corner fre-quency of the main event, Di Bona et al. (1995) found thatthe earthquake cannot be modeled over the whole frequencyband by assuming a simple x�2 spectral model. Rather, twodifferent sets of values provide the same flat level of accel-eration spectrum: fc � 0.6 Hz, Mo � 3.7 � 1017 N m, andfc � 1.3 Hz, Mo � 0.8 � 1017 N m, respectively, wherethe lower corner frequency provides a better fit of the low-frequency spectral amplitudes (Di Bona et al., 1995).

2D Modeling of the Earthquake

The numerical simulations are performed by the 2DChebyshev spectral element method (SPEM). The SPEM is ahigh-order finite element technique, which is particularlysuitable to compute numerically accurate solutions of full-wave equations in heterogeneous media. One of the attrac-tive features of this method is that it uses irregular meshes,and therefore, can accurately reproduce complex geometries.More details about SPEM and its application to the simula-tion of earthquakes, and in particular, to the development ofthe ground motion scenario in the Catania area, can be foundin (Priolo, 1999, 2000, 2001, 2003).

The computational model is defined along transect t06(Fig. 2). The structure along this transect, and the values of

Numerical Modeling of the 13 December 1990 M 5.8 East Sicily Earthquake at the Catania Accelerometric Station 243

Figure 1. Base map of the study area,showing geography, transect position (blackline), and the accelerometric stations (trian-gles) which recorded the 13 December 1990,earthquake. The Catania station location is em-phasized by the circle. The acceleration seis-mograms (north–south component) recordedby each station are shown. The dashed lineshows the trace of the Ibleo–Maltese faultadopted in this study. The earthquake is asso-ciated to the transcurrent fault segment.

Table 1Summary of the Seismological Parameters Estimated for the 13 December 1990, East Sicily Earthquake Main Shock by Previous

Studies and in This StudyThe following parameters are indicated: focal mechanism (�, d, k) of the two nodal planes P1 and P2, respectively; source depth, zS;

fault size, L � W; moment magnitude, MW (or ML in some cases); seismic moment M0; stress drop Dr; corner frequency fC;and average dislocation on fault D.

Reference(�, d, k)

(degrees) (Lat., Lon.)zS

(km)L � W

(km) MW

M0

(Nm)Dr

(bar)fC

(Hz)D

(m)

CMT (274, 64, 174) P1(37.25, 14.90) 15 — 5.6 3.27 � 1017 — — —

(7, 85, 26) P2Amato et al. (1995) (190 � 15, 80 � 25,

�10 � 20) P2(37.34, 15.26) 22 � 2 5 � 5* 5.4 (ML) — — — —

Di Bona et al.(1995)

— — 22 � 2 —5.7 3.7 � 1017 210 0.6

—5.2 0.8 � 1017 500 1.3

Giardini et al.(1995)

(95, 86, 180) P1†

— 13–17 — 5.8 3.7 � 1017 — — —(4, 82, �5) P2†

Boschi et al. (1997) — — — — 5.3 (ML) 1.1 � 1017 — — —

This study (95, 86, 180) P1 (37.34, 15.26) 20 5 � 8 5.83.7 � 1017

—0.6

0.70.8 � 1017 1.3

*Estimated from the aftershocks.†Fault-plane solution determined by Amato et al. (1995) using the moment tensor inversion obtained by Giardini et al. (1995).


Figure 2. Computational model along transect t06. The panels show the modelstructure with different levels of enlargement. The location of the point source is in-dicated by a star and label “s.IBM.” Units in km. South–east and north–west correspondto positive and negative abscissas, respectively. The vertical scale is exaggerated, anddifferent horizontal-to-vertical ratios are used for each panel. Material parameters aredescribed in Table 2.

the physical parameters used to define it, are shown in Figure2 and Table 2, respectively. The model structure is consistentwith that of the other transects modelled in past studies(Priolo, 1999). A point-source model is used, with faultmechanism defined by (�, d, k) � (95�, 86�, 180�), andsource depth at zS � 20 km.

The source-time function combines a low-frequency de-terministic part with a high-frequency stochastic contribu-tion, with a transition band between 2 Hz and 3 Hz. Thestochastic part is characterized by a random phase spectrumwhile the amplitude spectrum is kept constant at the upperfrequency level of the deterministic part.

To accommodate the presence of two corner frequen-cies, as suggested by Di Bona et al. (1995), we build up thedeterministic part as a combination of two pulses. The twopulses are characterized by distinct corner frequencies fc andfasp, which represent the inverse of the average rupture du-ration of the whole source area (RA) and the largest asperityarea (RAasp), respectively. Estimating the average slip du-ration as (Heaton, 1990), where S and Vr areT � 2 S / 3V� rs

the rupture area and velocity, respectively, and the ratio be-tween the total rupture area and that of the largest asperityis RAasp � 17.5% RA (Somerville et al., 1999), it comes outthat the ratio between the two corner frequencies can be


Figure 3. (a) Normalized source-time function and (b) normalized amplitude spec-trum used for the 2D spectral element simulations. See text for more details.

Table 2Description of the Soil and Rock Formations Used to Define the Transect Structure

The following parameters are reported: density (q), compressional and shear-wave velocities(VP, VS), and attenuation (Q).

Description Id q (kg/m3) VP (m/s) VS (m/s) Q (s�1)

Fine alluvium deposits with sand Alf/M 1870 360 200 12Alf 1900 340 190 15

Clay and silt interbedded with sand Asg 1950 450 250 18Sand, coarse gravel, and conglomerate SG 2000 810 450 18Clay interbedded with sand Aa 2000 900 500 30Pliocenic sediments and alloctonus Spa1 2000 1400 775 45

Spa2 2050 1655 915 50Spa3 2120 2000 1105 80Spa4 2150 2300 1310 100Spa5 2200 2800 1570 100

Vulcanits V2 2630 4100 2335 120Limestone CC1 2580 4700 2680 120

CC2 2600 5000 2835 150CC3 2630 5200 2950 150CC4 2740 6100 3460 200CC5 2780 6400 3630 200CC6 2835 7000 3970 300CC7 2800 6700 3750 300

approximated by fasp � 2.4 fc. Therefore, if we assume fc �0.6 Hz, it results in fasp � 1.4 Hz. The deterministic com-ponent of the source time function has the following form:

2y(t; f , f ) � M (2 p f ) t exp(�2 p f t)c asp o c casp 2� M (2 p f ) t exp(�2 p f t) ,o asp asp

where Mo � 3.7 � 1017 N m and Moasp � 0.8 � 1017 N m

according to Di Bona et al. (1995), The resulting source-time function and its spectrum are shown in Figure 3. Otherdetails are as in Priolo (1999).

Figure 4 displays seismograms computed at and re-corded by the ENEA–ENEL Catania station. The maximumcomputed frequency is 6 Hz and the total simulation time is40 sec. They liken each other quite well in terms of the

overall shape, polarity of the main arrivals, S-wave ampli-tude, and seismogram duration.

How Important is the 2D Structure?

To understand the importance of the effect of the 2Dstructure on the propagating wavefield and the ground shak-ing at the surface, we perform a set of simulations assuminga layered Earth structure. The simulations are performed us-ing the wavenumber integration method (Herrmann, 1996a,1996b). The method solves the seismic full-wave equationthrough a 3D layered medium.

We use three different layered structures that progres-sively approximate the 2D structure (Fig. 5). The first modelis the mean regional model of Sicily, adopted by the partic-ipants of the modeling activity of the first Catania Project


Figure 4. Three-component velocity and acceleration seismograms recorded by theCatania ENEA–ENEL accelerometric station (thick lines), and computed by the 2Dspectral element method (thin lines). The ENEA–ENEL seismograms are band-passfiltered at 0.25–6 Hz. The ENEA–ENEL velocity is obtained by time integration of theacceleration records. The origin time of predicted seismograms has been aligned to thatof the recorded seismograms. The peak value is explicitly written at the end of thetraces.

(F. Vaccari, personal comm. 1997). This model was built upfrom the inversion of surface waves of teleseisms. The sec-ond structure uses the deep structure (z � 1.5 km) of themean regional model and approximates the shallow structure(z � 1.5 km) of the 2D structural model specifically builtup for this study. The third model approximates the wholestructure of the 2D model of this study. As the layered struc-ture approximates the 2D structure, the model becomes pro-gressively more complex, featuring more and thinner layers.The mean regional model and the model which approximatesthe 2D model, feature three main differences: (1) the depthof the Moho, which is at about 32 km and 36 km in the twomodels, respectively; (2) the velocity contrast at about 6 kmand 14 km depth in the two models, respectively; and (3) thevelocities of the shallowest layers (z � 1.5 km), which de-crease much more rapidly toward the surface in the modelthat approximates the 2D structure.

The mean regional structure provides very poor seis-mograms (Fig. 6). As the plane layer models approximatethe 2D structure more accurately, the synthetic seismogramsfit the recorded ones progressively better. See for instancethe direct S wave (indicated by S in Fig. 6), the first reflectionS� of the direct S wave at the top of the limestone unit (i.e.,the interface between Spa5 and Cc in Fig. 2) and the fun-damental Rayleigh mode R0. The use of the mean regionalmodel “WIM: Mean” leads to over-simplified seismogramswith very low amplitude. The amplitude of the direct S waveis much better approximated only by inserting the low-velocity surface structure of the Priolo 99 model above themean model at 6 km of depth (model WIM: Mean � Priolo99). Note that the Mean regional velocity model (Fig. 5)usually features higher velocities than model Priolo 99. The

1D approximation of the 2D model (model WIM: Priolo99)improves the fit further, but the best results remain thoseobtained using the 2D structure of the SPEM model. Withthis model, the main phases can be clearly identified, andthe overall amplitude of the converted waves in the intervalsS–S� and S�–R0 is reproduced accurately. These convertedwaves are generated by the wavefield propagation throughthe 1–1.5 km thick portion of sediments (units Alf, Aa, Spain Fig. 2).

The structure of the Catania area south of the Etna vol-cano differs from the mean regional structure in the follow-ing two aspects. Firstly, a sharper velocity contrast is presentat about 12–14 km, in what is indicated as the carbonaticbasement (Makris et al., 1986; Della Vedova et al., 1989).This interface dips northward with an angle of about 8�–10�(Lentini, 1982; Petronio, 1997). Secondly, the sediment pro-file, which features an overall thickness of 2 km, a strongvelocity contrast at its bottom, and a decreasing gradienttoward the surface, is a good average representation of thereal structure. Our simulations are not sensitive to the depthof the Moho interface, which we set at about 4–5 km shal-lower than the mean regional model, following the resultsof the active seismic surveys performed in the area (Hirn etal., 1997). In fact, the signature of this interface does notshow up at the distance simulated here. The fact that the 1Dseismograms feature a lower coda amplitude than the 2Dseismograms, suggests that the high value of attenuationshould not be used for the whole distance of the shallowlayers, but only locally.

Though this study is limited to the analysis based on therecord of one event at one station, it shows that the use ofrealistic models can significantly improve modeling effec-


Figure 5. Plane layer models used in this study. Top panels, full depth; bottompanels, enlargment at depth 0–3.5 km. Thick, gray line, mean regional structure; Black,solid line, plane layer model derived from the 2D model specifically built up for transectt06 in this study; black, dashed line, plane layer model which accounts for the deepstructure (z � 1.5 km) of the mean regional model from, and the shallow structure (z� 1.5 km) of the 2D model.

tiveness. Besides that, even if we adopt a simple pointsource, the seismograms calculated using the 2D model areextremely realistic. This supports the following conclusions:(1) the 13 December 1990 East Sicily earthquake does notfeature significant rupture directivity effects, and (2) thewave propagation along the transect considered in this studydisplays significant 2D/3D subsurface structure effects.

Influence of the Catania ENEA–ENEL Site onthe Seismic Response

In addition to the earthquake data, seismic noise wasrecorded at the same site as the Catania station (Priolo et al.,

2001) with the aim of improving the estimation of local seis-mic response. We used Nakamura’s technique (Nakamura,1989), which provides the main features of the dynamicground response (i.e., the fundamental frequency) throughthe calculation of the spectral ratio between the horizontaland vertical components of seismic noise. In Figure 7, wecompare the HVSRs obtained from (i) seismic noise, (ii) theseismograms recorded by the accelerometric station duringthe 13 December 1990 M 5.8 East Sicily earthquake, and(iii) the seismograms computed by the 2D spectral elementmethod for the same earthquake.

The seismic noise signals have duration of about 30minutes of continuous recording. The HVSR are estimated


Figure 6. Three-component velocity and acceleration seismograms recorded by theCatania ENEA–ENEL accelerometric station (thick lines) and computed numerically(thin lines) for different plane layer models. Mean regional structure, WIM: Mean; deepstructure of the mean regional model and shallow structure of the 2D model, WIM:Mean � Priolo 99; the best plane layer approximation of the 2D structure, WIM: Priolo99; 2D model (SPEM). With the dashed, gray lines, an intuitive identification of thesame phases is proposed. Other details as in Fig. 4.

Figure 7. HVSRs computed for the ENEA–ENEL Catania station. Grey curves inboth panels: ratios determined from the recordings of the 13 December 1990 East Sicilyearthquake (full record and coda from 20 to 45 sec, left and right panels, respectively).Left panel, black curve: ratio simulated for the same event and using the SPEM 2Dspectral element code. Right panel, black curve: ratio obtained from the seismic noisemeasurements (Priolo et al., 2001).


Figure 8. Detail of the shallow structure beneath the ENEA–ENEL station (see alsoFig. 2 for the complete structure). This site corresponds to site A15 of the seismic noisesurvey (Priolo et al., 2001). The fundamental frequencies of the first three layers areexplicitly indicated.

as the average of the spectral ratios computed on a set ofabout ten 3-minutes-long running time windows. For the ge-neric i-th time window, the signal is processed as follows:(1) DC removal and linear detrending; (2) 5% cosine taper-ing; (3) computation of the horizontal component of themotion as a vector sum of the two horizontal components;(4) computation of the square root of the power spectraldensity (PSD) of the horizontal and vertical components(Hi and Vi, respectively), with FFT length of 1024 samples;(5) computation of the {H/V}i � Hi/Vi ratio; and (6) esti-mation of the median H/V and associated errors.

The earthquake records and synthetics seismogramshave a duration of about one minute. The procedure used forestimating the HVSR is similar to that used for seismic noise,except for the fact that only one time window is considered.This results in less smooth HVSR curves.

The ratios obtained from the seismic noise measure-ments detect the fundamental mode of vibration at about1.5 Hz. As Figure 8 shows, this frequency corresponds tothe resonant frequency of either the thin fine alluvium de-posit (Alf), or its combination with the underlying layer oflight blue clays (Aa). We remind the reader here that theshallow structure of the 2D model has been built up on thebase of the existing, rich geotechnical dataset (Faccioli,1997), and therefore, simulates the real structure veryclosely. The HVSRs, estimated from seismic noise, miss thepeak at the higher frequency of 4–6 Hz. This fact may eitherconfirm that Nakamura’s method can be used for identifyingonly the fundamental mode of vibration, or that the peak at4–6 Hz in the earthquake records is a feature of the earth-quake source. The HVSRs determined from the syntheticseismograms are generally noisier, however they well repro-duce the behavior of the spectral ratios determined from thefull ENEA–ENEL recordings for frequencies larger than

1 Hz. The low-frequency peaks at 0.5 Hz and 0.7 Hz of thesynthetic HVSR correspond to the natural frequencies ofthicker portions of sediments, corresponding to the bottomof Spa2 (z � 450 m) and Spa1 (z � 200 m) formations,respectively (see Figs. 2 and 8). The measured HVSRs donot display such peaks, while they show a clear rise at f �1 Hz. We point out that several sites in the Catania area havedisplayed amplification in the low-frequency band 0.3–1 Hz(Priolo et al., 2001). Although not definitively confirmed,this low-frequency peak is likely the effect of the geologicalstructure, namely of some hundreds of meters of the sedi-mentary formations (Giampiccolo et al., 2001; Priolo et al.,2001). The presence of such peaks in the simulated HVSRssuggests that our 2D model features too-high impedancecontrasts at those depths.

Conclusions

In this article we reproduce the waveforms and spectrarecorded by the Catania station during the M 5.8, 13 Decem-ber 1990 earthquake. This has been done using a pointsource representation and modeling the wave propagationalong a 2D structure, defined in detail. Our study demon-strates that the large amplitude recorded by the CataniaENEA–ENEL station during the 1990 earthquake can be ex-plained as a combined effect of crustal structure and sitecharacteristics.

Moreover, this paper validates the whole 2D approachused in previous studies to estimate the ground motion forthe Catania area. In general, it emphasises the importance ofusing methods that accurately model the wavefield propa-gation through laterally heterogeneous geologic structuresfor predicting ground motion. In particular, we show that theuse of a 2D model that represents both deep and shallow


structure in detail provides much more accurate waveformsthan simple (plane layer) models.

Acknowledgments

This work was partially funded by the National Group for the DefenseAgainst Earthquakes (GNDT), under The Catania Project (contract n.98.03227.PF54) of the National Council of Research (CNR), and the projectDetailed Scenarios and Actions for Seismic Prevention of Damage in theUrban Area of Catania of the National Institute of Geophysics and Volca-nology (INGV). We would like to thank the two anonymous reviewers andthe Associate Editor, I. Beresnev, for their valuable comments and sugges-tions. Figure 1 was made by GMT software (Wessel and Smith, 1995).

References

Amato, A., R. Azzara, A. Basili, C. Chiarabba, M. Cocco, M. Di Bona, andG. Selvaggi (1995). Main shock and aftershocks of the December 13,1990, Eastern Sicily earthquake, Ann. Geofis. 38, 255–266.

Azzaro, R., and M. S. Barbano (2000a). Seismogenic features of SE Sicilyand scenario earthquakes for Catania, in The Catania Project: Earth-quake Damage Scenarios For a High-Risk Area in the Mediterranean,E. Faccioli and V. Pessina (Editors), CNR-GNDT, Roma, 225 pp.

Azzaro, R., and M. S. Barbano (2000b). Analysis of the seismicity of south-eastern Sicily: a proposed tectonic interpretation, Annali di Geofisica43, 171–188.

Basili, A., M. Cocco, and E. Boschi (Editors) (1995). Collection of paperson the 1990 Eastern Sicily earthquake, Annali di Geofisica 38.

Barbano, M. S., and R. Rigano (2001). Earthquake sources and seismichazard in Southeastern Sicily, Annali di Geofisica 44, 723–738.

Bianca, M., C. Monaco, L. Tortorici, and L. Cernobori (1999). Quaternarynormal faulting in Southeastern Sicily (Italy): A seismic source forthe 1693 large earthquake, Geophys. J. Int. 139, 370–394.

Boschi, E., E. Guidoboni, G. Ferrari, G. Valensise, and P. Gasperini (Edi-tors) (1995). Catalogo dei forti terremoti in Italia dal 461 a.C. al1990, ING, Roma—SGA, Bologna, 1997, 644 pp. (in Italian).

Decanini, L., and G. F. Panza (Editors) (2000). Scenari di pericolositasismica ad Augusta, Siracusa e Noto, CNR-GNDT, Roma, 200 pp.(in Italian).

Della Vedova, B. D., G. Pellis, and E. Pinna (1989). Studio geofisicodell’area di transizione tra il Mar Pelagico e la Piana Abissale delloJonio, in Atti del VIII Convegno Nazionale GNGTS (CNR), Roma.Esagrafica, 543–558 (in Italian).

Di Bona, M., M. Cocco, A. Rovelli, R. Berardi, and E. Boschi (1995).Analysis of strong-motion data of the 1990 Eastern Sicily earthquake,Annali di Geofisica 38, 283–300.

Faccioli, E. (Coordinator) (1997). Geotechnical earthquake engineeringcharacterisation of the Catania municipal area. Technical Report andCD-ROM prepared by Ingegneria Geotecnica for CNR-Gruppo Na-zionale Difesa Terremoti, Milan, Italy (in Italian), 33 pp. � encl.

Faccioli, E., and V. Pessina (Editors) (2000). The Catania Project: Earth-quake Damage Scenarios For a High-Risk Area in the Mediterranean,CNR-GNDT, Roma, 225 pp.

Giampiccolo, E., S. Gresta, M. Mucciarelli, G. De Guidi, and M. R. Gal-lipoli (2001). Information on subsoil geological structure in the cityof Catania (eastern Sicily) from microtremor measurements, Annalidi Geofisica 44, 1–11.

Giardini, D., B. Palombo, and N.A. Pino (1995). Long-period modeling ofMedNet waveforms for the December 13, 1990 Eastern Sicily earth-quake, Annali di Geofisica 38, 267–282.

Heaton, T. H. (1990). Evidence for and implications of self-healing pulsesof slip in earthquake rupture. Phys. Earth Planet. Interiors 64, 1–20.

Herrmann, R.B. (1996a). Computer Programs in Seismology: An Overviewof Synthetic Seismogram Computation, Dept. of Earth and Atmo-spheric Sciences, St. Louis University, version 3.0 edition.

Herrmann, R.B. (1996b). Computer Programs in Seismology. Volume VI:wavenumber integration. Dept. of Earth and Atmospheric Sciences,St. Louis University, version 3.0 edition.

Hirn, A., R. Nicolich, J. Gallart, M. Laigle, and L. Cernobori, and ET-NASEIS Scientific Group (1997). Roots of Etna volcano in faults ofgreat earthquakes, Earth Planet. Sci. Lett. 148, 171–191.

Langer, H., M. Cristaldi, S. Gresta, and L. Tortorici (1999a). Ground-mo-tion amplification on vertically and laterally heterogeneous media:some typical cases from the urban area of Catania, Seism. Soc. Am.Annual Meeting, May 1999, Seattle (U.S.A.).

Langer, H., M. Cristaldi, S. Gresta, and L. Tortorici (1999b). Simulazionistrong ground motion di risposte del sito nell’area di Catania, inWorkshop on Caratterizzazione dell’input sismico mediante sismo-grammi sintetici: confronti ed applicazioni per la definizione sul ter-ritorio nazionale di scenari di dettagliati di danno, National ResearchCouncil—National Group for the Defense against Earthquakes (CNR-GNDT), 18 February 1999 Roma (in Italian).

Lentini, F. (1982). The geology of the Mt. Etna basement. Mem. Soc. Geol.It. 23, 7–25.

Makris, J., R. Nicolich, and W. Weigel (1986). A seismic study in theWestern Ionian Sea, Ann. Geophys. 4, 665–678.

Maugeri, M. (Coordinator) (2004). Detailed Scenarios and Actions for Seis-mic Prevention of Damage in the Urban Area of Catania, CoordinatedProject, V Relazione Sintetica finale, GNDT-INGV (in Italian).

Monachesi, G., and M. Stucchi (Editors) (1997). DOM4.1, an intensitydatabase of damaging earthquakes in the Italian area, National Re-search Council—National Group for the Defense against Earthquakes(CNR-GNDT), Milano, 2 vols., 1052 pp. Web site: http://emidius.mi.ingv.it/DOM (last accessed, May 2003).

Nakamura, Y. (1989). A method for dynamic characteristics estimation ofsubsurface using microtremor on the ground surface, QR RailwayTech. Res. Inst. 30, 25–33.

Pessina, V. (1999). Empirical prediction of the ground shaking scenario forthe Catania area, J. Seism. 3, 265–277.

Pessina, V. (2000). Empirical prediction of ground shaking for level I andII scenario earthquakes, in The Catania Project: Earthquake DamageScenarios For a High-Risk Area in the Mediterranean, E. Faccioliand V. Pessina (Editors), CNR-GNDT, Roma, 225 pp.

Petronio, L. (1997). Prospezioni sismiche per la valutazione del rischiosismico e vulcanico dell’area etnea e dell’offshore ionico della Sicilia,Ph.D. Thesis, University of Trieste, 196 pp. (in Italian).

Priolo, E. (1999). 2D spectral element simulations of destructive groundshaking in Catania (Italy), J. Seism. 3, 289–309.

Priolo, E. (2000). 2D spectral element simulations of the ground motionfor a catastrophic earthquake, in The Catania Project: EarthquakeDamage Scenarios For a High-Risk Area in the Mediterranean, E.Faccioli and V. Pessina (Editors), CNR-GNDT, Roma, 225 pp.

Priolo, E. (2001). Earthquake ground motion simulation through the 2Dspectral element method, J. Comp. Acoustics 9, 1561–1582.

Priolo, E., A. Michelini, G. Laurenzano, R. Addıa, and A. Puglia (2001).Seismic response from microtremors in Catania (Sicily, Italy), Boll.Geofis. Teor. Appl. 42, 335–359.

Priolo, E. (2003). Ground-Motion Modeling Using the 2D Chebyshev Spec-tral Element, in Method Numerical Analysis and Modeling in Geo-mechanics, J. W. Bull (Editor), Spon Press, Taylor and Francis GroupLtd., London, pp. 250–273.

Romanelli, F., and F. Vaccari (1999). Site response estimation and ground-motion spectrum scenario in the Catania area, J. Seism. 3, 253–264.

Romanelli, F., F. Vaccari, and G.F. Panza (2000). Site response estimationand tsunami synthesis in the Catania area, in The Catania Project:Earthquake Damage Scenarios For a High-Risk Area in the Mediter-ranean, E. Faccioli and V. Pessina (Editors), CNR-GNDT, Roma,225 pp.

Sirovich, L., and F. Pettenati (1999). Seismotectonic outline of South-Eastern Sicily: an evaluation of available options for the earthquakefault scenario, J. Seism. 3, 213–233.

Somerville, P., K. Irikura, R. Graves, S. Sawada, D. Wald, N. Abrahamson,


Y. Iwasaki, T. Kagawa, N. Smith, and A. Kowada (1999). Charac-terizing crustal earthquake slip models for the prediction of strongground motion, Seism. Res. Lett. 70, 59–80.

Suhadolc, P., F. Vaccari, F. Romanelli, A. Sarao, G.F. Panza, F. Mollaioli,and L. Decanini (2000). Modellazione del moto del suolo atteso pervari scenari sismici, in Scenari di pericolosita sismica ad Augusta,Siracusa e Noto, L. Decanini and G.F. Panza (Editors), CNR-GNDT,Roma, 200 pp. (in Italian).

Wessel, P., and W. H. F. Smith (1995). New version of the Generic Map-ping Tools, EOS Trans. Am. Geophys. Union 76, 329 pp.

Zollo, A., A. Emolo, A. Herrero, and L. Improta (1999). High-frequencystrong ground motion modeling in the Catania area associated withthe Ibleo–Maltese fault system, J. Seism. 3, 279–288.

Zollo, A., and A. Emolo (2000). Modeling of ground motion accelerationfield in the Catania area associated with the Ibleo–Maltese fault sys-tem, in The Catania Project: Earthquake Damage Scenarios For aHigh-Risk Area in the Mediterranean, E. Faccioli and V. Pessina (Ed-itors), CNR-GNDT, Roma, 225 pp.

Istituto Nazionale di Oceanografia e di Geofisica Sperimentale—OGSSeismological Research Center—CRSBorgo Grotta Gigante 42/cI-34010 Sgonico (Trieste), [email protected]

andUniversity of Trieste, Dept. of Civil Engineeringpiazzale Europa 1I-34127 Trieste, Italy

(G.L.)

Istituto Nazionale di Oceanografia e di Geofisica Sperimentale—OGSSeismological Research Center—CRSBorgo Grotta Gigante 42/cI-34010 Sgonico (Trieste), [email protected]

(E.P.)

Manuscript received 20 June 2003.

numerical modeling of the 13 december 1990 m 5.8 east sicily earthquake at the catania...

Documents