NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Nat. Hazards Earth Syst. Sci. Discuss., 2, 25972637, 2014www.nat-hazards-earth-syst-sci-discuss.net/2/2597/2014/doi:10.5194/nhessd-2-2597-2014 Author(s) 2014. CC Attribution 3.0 License.Natural Hazards and Earth System SciencesOpen AccessDiscussionsThis discussion paper is/has been under review for the journal Natural Hazards and EarthSystem Sciences (NHESS). Please refer to the corresponding nal paper in NHESS if available.Integration of HVSR measures andstratigraphic constraints for seismicmicrozonation studies: the case of Oliveri(ME)P. Di Stefano, D. Luzio, P. Renda, R. Martorana, P. Capizzi, A. DAlessandro,N. Messina, G. Napoli, S. Todaro, and G. ZarconeDipartimento di Scienze della Terra e del Mare DiSTeM, University of Palermo, ItalyReceived: 6 March 2014 Accepted: 1 April 2014 Published: 10 April 2014Correspondence to: R. Martorana (raaele.martorana@unipa.it)Published by Copernicus Publications on behalf of the European Geosciences Union.2597NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|AbstractBecause of its high seismic hazard the urban area of Oliveri has been subject of rstlevel seismic microzonation. The town develops on a large coastal plain made of mixeduvial/marine sediments, overlapping a complexly deformed substrate. In order to iden-tify points on the area probably suering relevant site eects and dene a preliminary 5Vs subsurface model for the rst level of microzonation, we performed 23 HVSR mea-surements. A clustering technique of continuous signals has been used to optimize thecalculation of the HVSR curves. 42 reliable peaks of the H/Vspectra in the frequencyrange0.610 Hzhavebeenidentied. Asecondclusteringtechniquehasbeenap-plied to the set of 42 vectors, containing Cartesian coordinates, central frequency and 10amplitude of each peak to identify subsets which can be attributed to continuous spa-tial phenomena. The algorithm has identied three main clusters that cover signicantparts of the territory of Oliveri. The HVSR data inversion has been constrained by strati-graphic data of a borehole. To map the trend of the roof of the seismic bedrock, fromthe complete set of model parameters only the depth of the seismic interface that gen- 15erates peaks tting those belonging to two clusters characterized by lower frequencyhas been extracted.The reconstructed trend of the top of the seismic bedrock highlight its deepening be-low the mouth of the Elicona Torrent, thus suggesting the possible presence of a buriedpaleo-valley. 201 IntroductionThe dynamic characteristics of a seismic phase of body or surface waves, generatedby an earthquake and incident on a portion of the ground surface, often show sharpvariations, frequency dependent, sometimes having extremely local character.A very minor heterogeneity of the same parameters, in large part explained by dif- 25ferences between the ray paths from the source to the observation points and by the2598NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|source model, has often been observed in extensive sub-planar outcrops of rocks char-acterizedbyhighbodywavevelocityandhighlayerthickness.Theseconditionsarecommonly approximated by the requirement that the velocity of shear waves is greaterthan 800ms1. A layer with these characteristics, generally not outcropping, is calledseismic bedrock. 5Therecanbemanycausesforthesharpspatial variationof thetransferfunctionbetween the roof of the bedrock and the surface of the soil as e.g.: resonance in soils,multiple reection, constructive interference and focusing geometry for topographic andburied interface irregularities.Signicant increases in amplitude of the phases for which occur peak values of the 10kinematic parameters of ground shaking, which may be the cause of relevant coseis-mic eects (fractures, landslides, liquefaction, . . . ) for buildings and infrastructure, arecalledsiteeects.Ingeneral all thisphenomenaarecontrolledbyanomaliesinthemechanical properties of the shallowest layers of subsoil, when it consists of soft sed-iments, or bytheshapeof surfacesof discontinuityclosetoor coincident withthe 15topographic surface.Ifweknewaverydetailedmechanical model ofthecoverlayer,consistingofsoftsediment, assuming that the layer behaves with respect to the propagation of wavesinanapproximatelylinear way, wecouldpredict thesiteeectsorevendetermineunivocally its transfer function by means of numerical calculation. 20The diculty and costliness of the reconstruction of a suitable model of subsurfaceusinggeophysical techniquesandthenecessityof not neglectingnonlinear eectsas: deamplication of strong motion and increase of the eective resonance period ofsoil depositsasthelevel ofexcitationincreases(Dimitriuetal.,2000)hasledmanyresearchers to develop and adopt more direct techniques that allow to determine ap- 25proximated empirical transfer functions.Everywhere on the Earths surface a seismic station may record ground vibrationsproduced by interference of body and surface waves, generated by sources randomlydistributed in the subsoil and known as seismic noise or microtremors.2599NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Nakamura (1989) proposed a technique based on the calculation of the ratio of thepower spectra of the horizontal and vertical components of the seismic noise recordedasinglethreecomponentsseismicstation.Thepowerspectraofthehorizontal andvertical components of the seismic noise are everywhere aected by characteristics ofthesources,aswell asbythedistributionofthemechanical parametersinthesub- 5soil. Nakamura (1989) showed that in correspondence to the resonance frequencies ofa sequence of layers, the Horizontal to Vertical Spectral Ratio (HVSR) of the seismicnoise presents peaks well correlated with the amplication factor of S waves, generatedby earthquakes, between the bottom and the roof of the stratication. Not necessarily,however, an HVSR peak must be attributed to resonance frequencies of a buried struc- 10ture, it might also depend on characteristics of the sources of noise and in such casewill not be correlated with amplication eects on incident waves trains. However, ap-propriate techniques of data analysis may help to discriminate, between peaks causedby source eects and those due to wave propagation.The main sources of seismic noise are: atmospheric and hydrodynamic phenomena, 15and uid circulation and micro-fracturing processes in the subsoil. Close to residentialareas, anthropogenic sources produce seismic noise, mainly with a relatively high av-erage frequency, compared to the natural noise (typically higher than 10Hz).Assumingasimplecompositionofthenoiseintermsofbodyandsurfacewaves,and having an adequate information on the subsoil, which can be expressed by means 20ofconstraintsforaone-dimensional model ofthesubsoil,itispossibletodeterminethe unknown parameters of the model by inversion of the curves HVSR.The seismic microzonation of a territory aims to recognize the small scale geologi-cal and geomorphological conditions that may signicantly aect the characteristics ofthe seismic motion, generating high stress on structures that could produce permanent 25and critical eects (site eects) (Ben-Menahem and Singh, 1981; Yuncha and Luzon,2000). The rst phase of the seismic micro-zoning is the detailed partition of the ter-ritory in homogeneous areas with respect to the expected ground shaking during anearthquake.2600NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|This paper focus on a recent micro-zoning study of the Oliveri urban area, a villagesettled on a coastal plain in Northeastern Sicily.This area is among those with highest seismic hazard, selected for the rst year ofthe project Studies of Seismic Micro-zoning, established by the Prime Ministers Orderno. 3907. 5Taking into account that a geological feature of this area is a thick blanket of coastaldeposits that cover the substrate seismic, in this paper we focused only on the recon-struction of the thickness of the sedimentary cover and its eect on the site response.The experimental data that have been collected are derived from: a geological survey,an inventory of existing wells and a geophysical survey, carried out using the technique 10HVSR.2 Seismotectonic settingThe area of Oliveri is part of the Peloritani Mountains, a segment of the Maghrebianchain that is mainly constituted by a pile of tectonic units consisting of Hercynian meta-morphic rocks and their Meso-Cenozoic sedimentary covers. 15AccordingtoGiuntaet al. (1998), thetectonicpileof thePeloritani Mountainsisformed by the following units:Longi-Taormina Unit, made of an epimetamorphic basement covered by a Meso-Cenozoic sedimentary sequence;Fondachelli Unit, consistingof phyllitesandraremetarenites, metabasitesand 20meta-limestones;Mandanici Unit, consisting of phyllites, quartzites, metabasites, marbles and por-roids;Mela Unit, characterized by a polimetamorphic basement, represented by parag-neiss and micaschists, with intercalations of metabasites and marbles; 252601NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|AspromonteUnit,consistingofamedium-highdegreemetamorphicbasement,with paragneiss and micaschists, gneiss, metagranits, amphibolites, and marbles.TheseunitsareoverlappedbyOligo-MiocenesiliciclasticturbiditespertainingtotheStilo-Capo dOrlando Formation.Upwardfollow, throughatectoniccontact, theAntisicilideUnits(Ogniben, 1970), 5which are represented by the Argille Variegate, that are in turn unconformably overlainby the Miocene Floresta Calcarenites (Ogniben, 1960, 1969; Guerrera e Wezel, 1974).Plio-Pleistocene calcarenites and clays are preserved in tectonic depression.In the Peloritani Mountains dierent deformation steps that are related to the Alpineorogeny, are superimposed to the Hercynian ductile deformations. The Oligo-Miocene 10contractionhasbeencharacterizedbyseveral phasesinwhichtherehasbeentheformationoffoldsassociatedwiththrustsystemsthathavefragmentedandstackedthe crystalline rocks and their sedimentary covers in dierent tectonic units.Reversefault systems(breaching)aectedtheareasincetheUpperMiocenere-sultinginmoderateshortening(GiuntaandNigro,1998).AttheendoftheMiocene 15the Tyrrhenian rifting produced a series of extensional low angle faults that resulted ina thinning of the chain. During Plio-Pleistocene times the area occupied by the Pelori-tani Mountains was aected by a strike-slip tectonic phase that generated two dierentsystems: the rst one synthetic with right cinematic and oriented NWSE and EW; thesecond one antithetical, mainly sinistral and oriented NS and NESW (Ghisetti and 20Vezzani, 1977, 1984; Ghisetti, 1979; Boccaletti et al., 1986; Finetti and Del Ben, 1986;MalinvernoandRyan,1986;Giunta,1991;MauzandRenda,1995;NigroandSulli,1995; Abate et al., 1998; Nigro, 1998; Nigro and Renda, 1999, 2000, 2002, 2005).The deformations are still actives and result in thrusts, within the chain, and crustalthinning in the Tyrrhenian area (Finetti, 1982; Finetti et al., 1996; Giunta et al., 2000). 25The geometry of thrusts are ramp-at-ramp like (Nigro, 1994; Giunta and Nigro, 1998)or in some cases are the result of duplex structures (Nigro,1994; Giunta and Somma,1996).2602NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Around the Oliveris plain there is evidence of active tectonics due to a right-lateralshearzonenamedAeolianTindariLetojanni fault(Ghisetti,1979;Ghisetti andVez-zani, 1979). This fault is generated by two dierent geodynamic domains: compressiveto west, related to a roughly NS trusting produced by continental collision, and exten-sional to east, controlled by NWSE expansion of the Calabro-Peloritan Arc (De Guidi 5et al., 2013). The analysis of seismic and geodetic data, acquired in the last 15years,and structural and morphological surveys show evidence of tectonic activity in this areaas recent as the middle-upper Pleistocene (Catalano and Di Stefano, 1997; Ghisetti,1979; De Guidi et al., 2013). Recent studies, not far from the Oliveris plain, have char-acterized some geothermal springs and gas vents and discussed their possible role in 10the generation of earthquakes (Giammanco et al., 2008) and a very high rate of up-lift (up to 5.5 mmyear1) has been estimated on the base of paleontological data (DiStefano et al., 2012).Themorphological featuresof Oliveri townarerepresentedmainlybyanalluvialplain, where lies the town itself, and by a hilly landscape that surrounds the southern 15and western part of the plain. Both the hills and the plain are crossed by the TorrenteCastelloandTorrenteEliconariversvalleys, that arepartlylledbysilty-sandyandpebbly sediments.The lithostratigraphic framework of the substrate of the urban area (Fig. 1) is made upby the metamorphic basement of the Aspromonte Unit that is covered unconformably 20by the arenaceus member of the Capo DOrlando Flysch. Upwards the yschoid sed-imentsaretectonicallyoverlainbytheArgilleScagliosebelongingtotheAntisicilideUnit. The previous deposits are capped by sands and calcarenites of Plio-Pleistoceneage. The most recent sediments consist of alluvial and beach deposits.The Oliveri territory is located in the area 932 of the ZS9 seismogenic zonation of 25Italy(Meletti andValensise,2004).Itincludesseismogenicstructureslargelyknownthrough geophysical exploration. The faults in this area partly allow the retreat of theCalabrian Arc and some are synthetic faults that segment the Gulf of Patti. The area932isoneof theareaswiththehighest seismicpotential of thewholeSicily. Init2603NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|occurredtheearthquakeof1908,forwhichdierentsourcemechanismshavebeenproposed, some of which are, probably, associated with the activation of complex faultsystems or blind faults (Ghisetti, 1992; Valensise and Pantosti, 1992; Monaco and Tor-torici, 1995).The high rate of seismicity in the Gulf of Patti and in the area between Alicudi and Vul- 5cano is associated to the two right-slip fault systems, PattiVolcanoSalina and Sisy-phus, respectively oriented NWSE and WNWESE. The most energetic instrumentalearthquakeshaveoccurredinthisarea15April 1978and28May1980withalocalmagnitude of 5.5 and 5.7 respectively.The earthquake of 1978 (MW= 6.6), related to movements along these faults (Neri 10et al., 1996), caused considerable damage in many countries of the Messina provinceandwasfelt uptotheCosenzaprovinceinthenorth, toDubrovnikinthesouth, toTrapani to the west and to the Ionian coast of Calabria to the east. The estimated max-imum macroseismic intensity has been VIII-IX MCS. Its mean macroseismic intensityin the Oliveriarea was VII MCS, where it caused serious damage in some buildings 15and signicant soil fractures (Barbano et al., 1979).The earthquakes of Novara di Sicilia, with lower magnitude and shallower hypocen-ters, seem to be linked to structures other than the fault system PattiVolcanoSalina.The earthquakes of Naso might instead be associated with NESW normal faults re-sponsiblefor theliftingof thechain. Amongthestructurespresent inthesouthern 20Tyrrhenian sea, those oriented around EO, would be responsible for the events of thewestern sector of the Aeolian Islands, and may have generated earthquakes like theone in 1823.Neri et al. (2003) have determined the focal mechanisms of earthquakes with depthsless than 50 km, occurred in Calabria and Sicily between 1978 and 2001. These indi- 25cate a high heterogeneity of the deformation eld in the area 932. Here the deformationstyle ranges fromextensional with southwest exposure to compressional with northeastexposure, proceeding from SE to NW.2604NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|The rst earthquake whit a destructive eect in the Oliveri area, reported in the cat-alogsof historical Italianseismicity, occurredin10March1786. Thisseismiceventwas characterized by MW= 6.15, epicenter at Oliveri and MCS intensity equal to IX inthe urban area. This earthquake severely damaged all the cities of the Gulf of Patti innorthern Sicily and almost destroyed the town of Oliveri (Guidoboni et al., 2007). 5Theseismicactivityrecordedinthelast 20yearstheareasurroundingtheurbancenter of Oliveri represented in Fig. 2 is characterized by about 3500 events with localmagnitude greater than 2. Of these about 89 % have hypocenters tightly concentratedaround an average depth of about 10 km. The remaining 11 % consists of events whosehypocenters are thickened around a plane that dips toward the northwest at an angle 10of about 60and are distributed rather uniformly with respect to depths up to 400km.ThesourcesoftheseeventsarelocatedinsidetheIonianlithosphericslabthatdipsunder the Calabrian Arc (Giunta et al., 2004).Ananalysisof completenessof thesedatasetshasindicatedalocal magnitudethreshold equal to about 2.6. 15Theshallowseismicactivityhasamarkedtendencytooccurthroughsequencesof aftershockssometimesprecededbyforeshocks. Thistendencywasassessedina quantitative way by comparing the estimates of the correlation dimension in the do-mains of the epicenter coordinates and time, both for the component of the seismicityconstituted by events not followed by aftershocks and the main shock of the sequences 20thatforthetotal setofevents(Adeloetal.,2006).Theestimatesofthespaceandtime correlation dimension relating to independent events are close to those expectedfor uniform distributions, as opposed to those for the total set that are much smaller.The geometric characteristics of the clusters related to the shallower seismicity iden-tifyorientationsthat areconsistent withthoseof themaintectonicstructuresof the 25area.The seismicity of the last 20 years has also been analyzed in the magnitude domain.For the events of the area of Fig. 2 have been estimated the value of b of the law ofGutenbergRichter and the 95 % condence interval, using the estimator of Tinti and2605NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Mulargia(1987)not distortedbythedatagrouping. Inparticular, fordeeperevents,associated with the Ionian slab, the value of b is equal to 0.6 0.2 while for supercialevents has been estimated the value 1.150.13 which is comparable with the value ob-tained for the independent component of the seismicity of the whole south-Tyrrhenianarea (see Adelo et al., 2006). 53 HVSR measurements and processingThe HVSR technique, is today one of the most commonly applied methods of expeditemicrozonation studies of urban areas.Some open questions and doubts remain, however, on the reliability of HVSR tech-nique,mainlyrelatedtothenumerousassumptionsabout:spatial distributionofthe 10microtremor sources, composition of the seismic noise in terms of body, Rayleigh andLove waves and characterization of the propagation medium, often hard to verify.Infact,iftheshapeoftheHVSRcurvesismainlycontrolledbythe Swavetrans-fer function within the shallowest sedimentary layers (Nakamura, 1989, 2000; Parolaiet al., 2000; Mucciarelli et al., 2003), bothH/V peakfrequencyandamplitudemay 15be, respectively, related to subsoilresonance frequency and site amplication factor.Ontheotherhand,iftheshapeofthe H/V curvesismainlycontrolledbythepolar-ization of the fundamental mode of the Rayleigh waves (ellipticity) (Lachet and Bard,1994; Kudo, 1995; Bard, 1999; Konno and Ohmachi, 1998; Fh et al., 2001), only anindirect correlation between the H/Vpeak amplitude and the site amplication can be 20expected.Because of the above conceptual diculties in the assessing of these techniques,many experimental works have played an important role. In these were crosswise an-alyzedcorrelationsbetween: peaksof thespectral ratioH/V , transferfunctionsde-rivedbythemethodSSR(Borcherdt, 1970) andsiteeectsexpressedintermsof 25increasesof macroseismicintensityorpeakvaluesof theusual groundmotionpa-2606NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|rameters (Dimitriu et al., 2000; Rodriguez et al., 2000; Mucciarelli and Gallipoli, 2004;Panou et al., 2005; De Rubeis et al., 2011).The results of these analyzes indicate that there is a high spatial correlation betweenareas in which there are site eects and areas in which the spectral ratios are charac-terized by peaks of signicant amplitude, also highlight the correlation between some 5of the frequencies of the peaks of the HVSR curves and the main eigenfrequencies ofthe subsurface structures. A very minor correlation is observed, instead, between theamplitudes of the peaks relating to the dierent survey techniques and between theseand the local amplication of the macroseismic intensity or peak values of the shakingparameters. 10Inmost of thestudiesreportedintheliteraturethistechniquehasbeenusedtostudy site eects in sedimentary basins, near faults or cavities or to estimate the res-onance frequency of building (e.g. Field and Jacob, 1993; Lermo and Chvez-Garca,1994;Mucciarelli,1998;Bard,1999;Fhetal.,2001).Insomerecentstudiessomeauthors (Fh et al., 2003; Scherbaum et al., 2003; Arai and Tokimatsu, 2004; Parolai 15et al., 2005, 2006; Picozzi et al., 2005) perform joint and/or constrained inversion of theHVSR curves with information relating to neighboring perforations and other types ofgeophysical survey, to obtain S waves velocity models.The foregoing highlights the importance of improving the processing techniques ofthe recorded signals to be able to link, more reliably, the results of HVSR investigations 20with site eects.For this reason an issue to be addressed is related to the processing procedure forthedeterminationofthespectral ratio.Inparticular,inordertoobtainreliableHVSRcurves, several criteriamust berespectedfor: acquisitionof 3-componentssignals,selectionof thetimewindowstobeusedintheSRcalculation, computationof the 25amplitude spectrumfor each time window, composition of the amplitude spectra relativeto the horizontal components, computation of the HVSR for each window and, nally,computation of the average HVSR (SESAME Project 2004).2607NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Oneof themost controversial aspectsintheapplicationof theHVSRtechniqueconcernsthecriteriafor theselection, inthemicrotremor signals, of timewindowsappropriate for the calculation of the HVSR curves. Several authors assume that spikesand transients, common in all the records, bring information highly dependent from thesourceandcannotbeusedtoestimatetheresonancefrequencyofthesite(Horike 5et al., 2001). For this reason many authors exclude the non-stationary portion of therecorded noise for the computation of the average HVSR curve, thus considering onlythe low-amplitude part of signal. Some authors instead (Mucciarelli and Gallipoli, 2004)observed by correlation studies, that the non-stationary, high-amplitude noise windowsimproves the capability of the HVSR curve to mimic the HVESR curves, obtained with 10recordings of vibrations induced by weak earthquakes and, therefore, suggest that theyshould not be removed.The windows selection can be made scanning in time domain the three componentsnoise signal, or observing the whole set of HVSR curves, related to single time win-dows, as function of time. 15In both cases, while there are in the literature numerous suggestions and prescrip-tions about this stage of the processing (SESAME Project 2004), the selection of thewindows to be used in the calculation of the HVSR curve is highly subjective.The lack of objective selection criteria leads to several drawbacks. In particular, in-creasing of the processing duration of the; producing results whose reliability depends 20on the skill of the operator. The permanence of abnormal windows can cause bias ofthe estimates of the expected value and the dispersion of the set of selected curvescan lead to overestimates of the true uncertainty of the spectral ratios.Another problematic aspect must be addressed when the HVSR method is appliedfor the seismic microzonation of an area, that is, for its subdivision into areas relatively 25uniform with respect to their response to a seismic input. The experimental informationwithwhichwearefacingthisproblemveryoftenconsistsof aset of HVSRcurvesrelatedtoafairlydensedistributionof measurement points, andsometimesalittle2608NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|informationaboutthesubsurfacefrompastgeophysical andgeological surveysandfrom some drilling.Each HVSR curve may contain more signicant peaks attributable to dierent causesnamely dierent characteristics of the subsurface. This raises the problem of deningtheareasinwhichitisacceptabletoassumethepresenceofthesiteeectdueto 5aparticular cause. Ineachinterior point tothat area, dierent fromthemeasuringpoints,itwill bereasonabletoassumeasiteeectandcharacterizeitinfrequencyand amplitude with a technique of 2-D interpolation of parameters determined at themeasurement points.Inthispaper, boththeaforementionedproblems, havebeenaddressedbyclus- 10ter analysistechniquesdescribedandassessedthroughsomeexperimental tests(DAlessandro et al., 2013).To try to identify areas of the town of Oliveri probably interested in site eects and todene a preliminary subsurface model 23 HVSR measurements have been performed,quite uniformly distributed with a mean spacing of about 250m. 15Forthemeasurementshasbeenusedthe3Cseismicdigital stationTROMINO(Micromed), equipped with three velocity transducers.For each measurement point the recording length was 46min and the sampling fre-quency 256 Hz. Each record was divided into 247 windows 10 s long. This choice al-lowed us to have a minimum numerosity of the sets of sampling windows selected for 20the analysis, with the clustering technique, not less than 150 (Sesame, 2004). Followingthe SESAME criteria, the signal relative to each time window was de-trended, baselinecorrected, tapered and band pass ltered between 0.1 and 25 Hz. Than we performedthe FFT of each signal components and determined the HVSR curve following KonnoandOhmachi (1998). Theanalysiswerelimitedtothefrequencyrange0.120Hz, 25which is the frequency range of interest for seismic microzonation and earthquake en-gineering. The data in the frequency domain have been ltered with a triangular windowto obtain a smoothing of 10 %.2609NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|To determine the HVSR curves of each measurement point, we implemented a pro-cedure based on an Agglomerative Hierarchical Clustering (AHC) algorithm. This hasbeenselectedbytestingdierent hierarchical andnon-hierarchical clusteringalgo-rithms.WeusedasproximitymeasuretheStandardCorrelation(SCxy)andtheav-erage linkage (AL) criteria (DAlessandro et al., 2014a). 5This procedure allowed us to split, almost automatically, peaks probably linked to siteeects from other perhaps related to source eects.Theseparationofthetwoeectsallowedustodetermineforeachclusteranav-erage HVSR curve better related to the local site eects and to identify all the signif-icant peaksof theH/V spectrainthefrequencyrange0.620 Hz. Theirattribution 10to resonance phenomena of buried structures has been validated, in agreement withSESAME criteria, by analyzing the standard deviation of the spectral amplitudes andthe independence from the azimuth of the mean spectral ratio (Fig. 3). Assuming anapproximately one-dimensional ground around each measurement point, a peak of themean curve of a cluster have been attributed to characteristics of the subsoil only it is 15azimuthally stable.IneachHVSRcurveall thesignicantpeakswereidentiedandcharacterizedbytheir center frequency and amplitude. These parameters are reported in Table 1.The HVSR data (Fig. 4) have revealed the likely presence of amplication of groundmotioninthefrequencyrange0.72.6 Hz, overalargepart of theurbanarea. The 20thickness of the soft coverage in few places where it is known and its lithological com-position allow us to hypothesize that the cause of amplication is the resonance of thesame coverage.4 Frequency mapsAclusteringproceduremaybealsousedtogrouppeaksof HVSRaveragecurves 25relatedtodierent sites, soastoidentifyareascharacterizedbysiteeects, whichbecause of the similarity in frequency and amplitude of the corresponding HVSR peaks2610NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|and the not excessive distance between the measuring points, are probably caused bythe same buried structure. Considering the parameters of such peaks as sampling ofspatial trendsthatarecontinuousontheareacontainingthemeasuringpoints,itispossible to estimate the expected peak amplitude and frequency at each point of thearea by two-dimensional interpolation techniques. 5To this end we implemented a clustering procedure to be applied to the parametervectors that characterize the individual peaks. Each vector contains: period, amplitude,plane Cartesian coordinate and a parameter indicative of the outcropping lithology. Thenumber of vectors is equal to the number of peaks considered signicant. The proce-dure is based on an Agglomerative Hierarchical Clustering (AHC) algorithm, using the 10averagelinkage(AL)criteria.Theelementsofthesimilaritymatrixareweightedav-erages of Euclidean relative distances between the quantitative parameters of a pairof peaks, while for the lithological parameter the distance has been set equal to 0 forcoincident lithologies and 1 otherwise. Each dierence between pairs of parameters isnormalized by the maximum dierence in the sample. To disadvantage the inclusion in 15the same cluster of more peaks relative to the same measurement point, the relativedistance of coincident points was placed equal to 1. The weight to be attributed to anytypeofparametermustbetheresultofanoptimizationthattakesintoaccountbothparameters intrinsic to the clustering process and geological constraint (DAlessandroet al., 2014b). For this application we set wper= 0.4, wampl= 0.2, wdist= 0.4, wlith= 0. 20Theprocedureofclusteringmadeitpossibletosplitthetotal setofpeaksinfourclusters referred to as: B (Blue), R (Red), G (Green), V (Violet), respectively, consistingof 21, 12, 7, and 2 peaks. The cluster made up of only two peaks has been neglectedin the overall description of the area due to poor reliability of both peaks, highlightedby the parameters listed in Table 1. Figure 5 shows the dendrogram of the clustering 25process. Fromthis it is clear that, adopting a cut threshold slightly lower than the optimalone adopted by the calculation code, the three remaining clusters would be gatheredin a single cluster.2611NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Figure 6 shows the characteristics of the clusters in the diagram H/Vvs. F0 and theirspatial distribution.As the two clusters B and G are very near in the scatter plot and their areas have anempty intersection, it was considered reasonable to assume that they represent the ef-fects, continuously variable on the union of their areas, of a single source phenomenon. 5In particular it is assumed that B and G are related to the resonance eect of a cov-ering layer delimited at its base by the same discontinuity surface with variable depth.Conversely it is assumed that the cluster R, whose extent represent a signicant part ofthat of BG and is characterized by higher frequencies respect to the correspondingones of the peaks of BG, show the resonance eects of the sedimentary cover down 10to a shallower interface.Therefore, two dierent maps can be reconstructed, each of them grouping and sep-aratelyrepresentingthecontinuoustrendofthefrequencyfortheclustersABGand R. The amplication estimated at the measurement points is reported numericallyin the maps, being not reliably mappable with smoothed trends over large areas. 15For the processing of the maps, the kriging algorithm has been applied to interpolatethe experimental peak frequencies at the nodes of a regular lattice with spacing equalto 10m. The frequency maps are plotted using a contour equidistance equal to 0.05(Fig. 7).An area south of the town, interesting for the spatial planning, seems to be charac- 20terized by higher amplication at frequencies greater than 1.2Hz (Fig. 7, right).5 Seismic bedrock mapping by HVSR inversionFor mediaconsistingof homogeneouselasticor viscoelastichorizontal layers, it ispossibletoexpressthebehaviourofH/V vs.frequencyfordierentmechanical pa-rameters, in particular shear wave velocity, and thicknesses of the layers, if you make 25simplifyingassumptionsabout thecompositionof themicroseismiceldintermsofbody and surface waves. For example that the microseismic eld is totally constituted2612NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|bybodywavesemergingalmostverticallyortotallybythefundamental modeoftheReyleigh waves.Evenwiththesesimplisticassumptions, thedeterminationof theparametersthatcharacterizethesubsoil fromtheH/V (f )curvesrequiresthesolutionofanonlinearinverse problem, poorly constrained by the experimental data and therefore highly non- 5univocal and unstable, especially in complex cases as velocity inversions. The resultswill certainly be unreliable if you are not using an initial model already representativeof the real geological stratication, because it is well constrained by drilling data, ge-ological observationsorresultsofothergeophysical surveysmostly,in-holeseismicsurveys. 10It should be, also, pointed out that, due to lateral variations of geologic parametersas: porosity, water content, fracturing degree, etc. the mechanical parameters of eachlayer will not be uniformly distributed with obvious negative consequences on the ac-curacy of the estimate of the depth of geological interfaces and in particular the roof ofthe seismic bedrock. 15Taking into account the available geological information, the H/V curves have beeninverted to estimate the depth of the seismic bedrock, also reported in Table 1.The inversion has been carried out using the Dinver inversion code developed in theframe of the project Geopsy (Wathelet et al., 2004; Wathelet, 2008). In this techniquethecomputationofthedispersioncurveforastackofhorizontal andhomogeneous 20layersisdetermined(Haskell,1953).OnlytheRayleighphasevelocitiesareconsid-ered as the experimental dispersion curve is generally obtained from processing thevertical components of noise. As ambient vibrations may contain waves travelling in alldirections, as body waves and Rayleigh and Love waves, a limit of this method lies inthe uncertain composition of seismic noise and in the impossibility to include among 25theunknownparametersthoserelatingtothemodel of themicroseismiceld(Pe-terson, 1993). For the inversion the neighbourhood algorithm (Sambridge, 1999) wasused. This is a stochastic direct-search method for nding models of acceptable datat inside a multidimensional parameter space.2613NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|The a priori density of probability is set as uniform over the whole parameter space,the limits of which are dened by the a priori ranges of all chosen parameters. Oncethe data mist function is determined, the neighbourhood algorithm provides a simpleway of interpolating an irregular distribution of points, making use of Voronoi geometryto nd and investigate the most promising parts of the parameter space. Hence, the ro- 5bustness of the nal results is generally checked by running the same inversion severaltimes with dierent random seeds, an integer value that initializes the pseudo-randomgenerator.Unfortunately,ithasbeenpossibletoconstrainthemodel usingthedataofasin-gleborehole,locatedattheeasternboundaryofthemunicipality(Fig.8),toimpose 10constraints on the thicknesses of the layers in the inversion of the HVSR curve relatedto the nearest recording point. From this model estimates of the Swave velocity wereobtained, which were subsequently used to dene the starting model for the inversionof the other HVSR curves.The depth of transition of the shear wave velocity from less to greater than 800 ms115is considered the depth of the seismic bedrock. When evaluating the reliability of theestimate of this depth, we must consider that the trends represented are heavily inu-encedbytheinterpolationprocessbetweentheHVSRmeasuringpoints.Thedepthvalues obtained from every measuring points are in fact only possible estimates, as-suming minimal lateral variations. To avoid interpolation between depths of interfaces 20due to dierent geological structures, it was decided to group and correlate frequen-cies related to the same cluster. However it is not possible to exclude that frequenciesbelonging to the same cluster are due to dierent structures or vice versa.In almost all the HVSR measurements carried out on the alluvial material a maximumpeak is evident in the range 0.71.4 Hz. This, by the inversions results, seems to be 25related to a variation of the shear-wave velocities that, under a depth of 4060m reachvalues attributable to a seismic bedrock (about 800 ms1).Analyzing the 1-D inverse models (Fig. 9) derived by inverting HVSR measurements,estimates of the depth of the seismic bedrock have been obtained. These, together with2614NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|other reliable data like drilling and other geophysical surveys, were used to constructthemapofthethicknessofthesedimentarycover(Fig.10,left)andthemapofthealtitude of the top of seismic bedrock a.s.l. (Fig. 10, right). The estimates of the depth ofthe bedrock were interpolated by imposing a constraint of a depth equal to zero in areaswheregeological stratawithelasticcharacteristicsof seismicbedrockoutcrop. The 5kriging interpolation algorithm was used, obtaining a lattice with equidistance betweennodes equal to 10 m.The map of the bedrock depth contours was obtained by extracting from the DigitalElevationModel oftheareaagridofpointsequidistant50 m.Fromthesethecorre-sponding lattice of the depth of the bedrock were subtracted. This choice was made 10in order to obtain a trend of the isobaths that was not excessively tied to topographicdetail but which, however, did not present unrealistic areas with higher elevations of thetopographic surface. It should however emphasize that the real detail of the maps cannot be higher than the sampling density of the HVSR measurements, which generallyis not less than 300 m. 15Considering the geological conguration of the subsoil and the values of shear-wavevelocity (deduced from direct bore-hole measurements) a preliminary identication wasmade of the geological structures that can cause seismic resonance.For the above mentioned reasons the inversion of HVSR curves and evaluation of thethickness of the cover did not take into account the available seismic down-hole data 20and instead, the mean values of shear-wave velocity of the cover, available in literature,were considered. In particular, we considered about 200 to 250 ms1for alluvium andabout 350 to 400 ms1in correspondence of silty sands.However the seismic velocity values used results close to those characteristic of therock in the area. 25Onthebasisof the3-Dtrendof seismicbedrockandof geological informationfrom eld surveys and from published geological maps, three geological sections weredrawn (Fig. 11).2615NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|In the section A, oriented fromwest to east, it possible to observe, in the westernmostpart, the stratigraphic overlay of the Flysch of Capo DOrlando on the metamorphic sub-strate, which is lowered eastward by normal faults. The alluvial deposits of coastal plainare onlapping on the substrate whose lithology is not known in most of the area (sec-tions B and C) due to the lack of information deriving from deep boreholes. Therefore 5the only information is related to geophysical analysis. Along the Oliveris plain somecatalogs (e.g. ITHACA) and structuralstudies (Ghisettiand Vezzani,1977) testify thepresence of an active fault that in geological sections is approximately shown.6 Geological-technical modelThe Oliveris territory consists of a oodplain surrounded by mountains. These moun- 10tains are constituted by a metamorphic basement (Aspromonte Unit) covered by yschand calcarenites sediments. The dominant lithology of this oodplain is sand with vari-able percentage of silt and clay and frequent alternations of silty-sandy gravels. The de-position of these sediments is related to the presence of two rivers, Torrente Eliconaand Torrente Castello, which cross the oodplain. The only outcrop, inner to the ood- 15plain, is a hill, named Il Castello, characterized by Gneiss of the Aspromonte Unit. Thishill is the result of the morphological evolution of the Torrente Castello, in the northernside, while the southern side is bounded by buried faults. According to ITHACA cata-logue the oodplain is crossed by a NNWSSE normal fault named TindariLetojanni.However,inthestudiedarea,therearenotmorphological evidencesthattestifythe 20existence of this fault (Ghisetti, 1979).The lack of good data of wells has hampered the denition of the lithological distribu-tion, in particular under the town. It is been possible to use only the data from two wells;the rst is located close to Torrente Elicona, the second is located close to the Castellohill. The analysis of the core of the rst well shows the presence of anthropic deposits 25in the rst 1.6 m, followed by 3.4 m of well graded sands with gravel elements of meta-morphic origin, 5.5m of silty sands, with rare rounded gravel elements, and 5.4m of2616NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|poorly graded sand. The altered substrate, according to its mechanical characteriza-tion by HVSR data, seems to be made by incoherent sand and grey silt. The analysisof core samples has permitted to recognize a bivalve mollusk fauna. The groundwaterlevel is placed at a depth of 7.5 m.Thecorerelativetothesecondwell showsevidenceofatectonicdislocationthat 5put in contact the Argille Scagliose unit with the metamorphic basement. This frame-workisalsosupportedbygeophysical data. FromHVSRresultsit waspossibletoreconstructthethicknessofthecoverage.InthenorthernsideofTorrenteCastello,the yschoid coverage has a thickness of about 20 m while, in the southern part, thethickness of clay and calcarenites coverages exceeds 45m (Fig. 1). The nature of the 10coverages is related to the sedimentary contribution, of the rivers and of sea level uc-tuations. Thanks to the interpretation of geophysical data it has been possible recon-struct the variation of the cover thickness in the studied area. The reconstructed modelsuggest thepresenceof twoburiedriverbedsseparatedbyarelief. Thesedimentthickness in these two areas is about 45 m. Using a DEM support was produced the 15seismic bedrock chart in which is possible to observe as the river beds have suereda migration toward the present day position. In particular the Torrente Elicona was alittle bit shifted while the other one, the Torrente Castello, was much moved north thantoday. According to the bedrock map the relief that separates the two depressions hasa depth of 40 m while the oor of basins show a depth of 6070 m. In the southern part 20of the town there is another basin in which the bedrock reaches a deepness of 40m.This could be related to another minor river joined with the Torrente Elicona. Based onthe hydrogeological map the presence of groundwater level in the oodplain is attestedat a depth less than 15m. This condition renders the oodplain exposed to liquefactionrisk. 252617NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|7 ConclusionsThe main purpose of this work is not, of course, to present the complete results of therst level microzonation study carried out in an urban center of an area of high seismichazard,buttoshowhowasurveyperformedbytheseismicnoiseHVSRtechnique,thankstosomesignicant improvementsintheprocessingprocedure, canproduce 5results that go far beyond the simple identication, in some recording points, of likelysite eects at frequencies of engineering interest. The most innovative aspects of theprocessing of the recorded signals, underlying the signicant improvement in the nalresult, are based on the use of clustering techniques in two phases of the study.The application of a clustering procedure to the set of signals related to the elemen- 10tary recording intervals, allowed a more objective identication and grouping in a clus-ter of the noise windows in which the spectral eects of the subsurface structures aredominant, compared to those dominated by trends attributable to spectral characteris-tics of the noise sources or to accidental interference eects between waves trains ofa dierent nature. 15The average spectral ratio of the cluster dominated by structural eects, often hasbeen characterized by smaller variance of the frequencies and amplitudes of the sig-nicant peaks, compared to those of the peaks determined with standard methods.The good quality of the estimates of frequency and amplitude of the peaks, allowedto use, with good results, a second clustering procedure to identify areas in the inside 20of which could be considered reasonable to interpolate the values of the parameters ofpeaks attributable to the eects of the same buried structures.Assumingthat thethreeidentiedclusterscontainpeaksproducedbyresonanceeects of layers with varying thickness, it was possible to reconstruct the possible trendof theroof of theseismicbedrockbyinversionof theHVSRcurvesconstrainedby 25all the geological and lithological available information and by a criterion of minimumlateral variability of the physical and geometrical parameters. Although the reliability ofthe3-Dmodel obtainedislimitedbythelackofgeophysical anddrillingdata,which2618NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|wouldhaveallowedustobetterconstrainthevaluesoftheunknownparametersofthe model, however the general characteristics of the reconstructed subsoil portion arefully compatible with the geological data. The investigations necessary to increase theresolution and reliability of the model will be carried out in-depth studies provided bythe micro-zoning of the second and third level. 5ReferencesAbate, D., De Pippo, T., Ilardi, M., and Pennetta, M.: Studio delle caratteristiche morfoevolutivequaternarie della piana del Garigliano, Il Quaternario, 11, 149158, 1998.Adelo, G., Chiodi, M., De Luca, L., Luzio, D., and Vitale, M.: Southern-Tyrrhenian seismicity inspace-time-magnitude domain, Ann. Geophys.-Italy, 49, 12451257, 2006. 10Arai, H. and Tokimatsu, K.: S wave velocity proling by inversion of microtremor H/Vspectrum,B. Seismol. Soc. Am., 94, 5363, 2004.Barbano, M. S., Bottari, A., Carveni, P., Cosentino, M., Federico, B., Fonte, G., Lo Giudice, E.,Lombardo, G., andPatan, G.: Macroseismcstudyof thegulf of Patti earthquakeinthegeostructural frame of the North-Eastern Sicily, Bollettino della Societ Geologica Italiana, 1598, 155174, 1979.Bard, P. Y.: Microtremormeasurements: atool forsiteeect estimation?, in: TheEectsofSurface Geology on SeismicMotion.Balkema, edited by: Irikura, K., Kudo.K., Okada, H.,and Sasatami, T., Rotterdam, 3, 12511279, 1999.Ben-Menahem, A. and Singh, S. J.: Seismic Waves and Sources, Springer-Verlag, New York, 201981.Boccaletti, M., Tortorici, L., and Ferrini, G. L.: The Calabrian Arc in the frame of the evolutionof the Tyrrhenian Basin, in: Paleogeography and Geodynamics of the Perityrrhenian Area,edited by: Boccaletti, M., Gelati, R. Ricci Lucchi, F., and Pepe, F., Giorn. Geol., 48, 113120,1986. 25Borcherdt, R. D.: Eects of local geology on ground motion near San Francisco Bay, B. Seismol.Soc. Am., 60, 2961, 1970.Catalano, S. and Di Stefano, A.: Sollevamenti e tettogenesi pleistocenica lungo il margine tir-renico dei Monti Peloritani: integrazione dei dati geomorfologici, strutturali e biostratigraci, IlQuaternario, 10, 337342, 1997. 302619NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|DAlessandro, A., Capizzi, P., Luzio, D., Martorana, R., and Messina, N.: Improvement of HVSRtechnique by cluster analysis, Geoitalia 2013, IX FIST, Pisa, 1618 Settembre 2013, 2013.DAlessandro, A., Luzio, D., Capizzi, P., and Martorana, R.: Improvement of seismic microzona-tion by means of cluster analysis, in preparation, 2014a.DAlessandro, A., Luzio, D., Martorana, and R., Capizzi, P.: Improvement of the noise Horizontal 5to Vertical Spectral Ratio by cluster analysis, in preparation, 2014b.DeGuidi,G.,Lanzafame,G.,Palano,M.,Puglisi,G.,Scaltrito,A.,andScarf,L.:Multidisci-plinary study of the TindariFault (Sicily, Italy) separating ongoing contractionaland exten-sional compartments along the active Africa-Eurasia convergent boundary, Tectonophysics,588, 117, 2013. 10De Rubeis, V., Cultrera, G., Cadet, H., Bard, P. Y., and Theodoulidis, N.: Statistical estimation ofearthquake site response from noise recordings, 4th IASPEI/IAEE International Symposium:Eects of Surface Geology on Seismic Motion, ESG4, 2011.Di Stefano, E., Agate, M., Incarbona, A., Russo, F., Sprovieri, R., and Bonomo, S.: Late Qua-ternaryhighupliftratesinnortheasternSicily:evidencefromcalcareousnannofossilsand 15benthic and planktonic foraminifera, Facies, 58, 115, 2012.Dimitriu, P., Theodulidis, N., and Bard, P. Y.: Evidence of non linear site in HVSR from SMART1(Taiwan) data, Soil Dyn. Earthq. Eng., 20, 155165, 2000.Fh, D., Kind, F., and Giardini, D.: A theoretical investigation of average HIV ratios, Geophys. J.Int., 145, 535549, 2001. 20Fh, D., Kind, F., and Giardini, D.: Inversion of local Swave velocity structures from averageH/Vratios, and their use for the estimation of site-eects, J. Seismol., 7, 449467, 2003.Field, E. H. and Jacob, K. H.: The theoretical response of sedimentary layers to ambient seismicnoise, Geophys. Res. Lett., 20, 29252928, 1993.Field, E. H. and Jacob, K. H.: A comparison and test of various site response estimation tech- 25niques, including three that are not reference site dependent, B. Seismol. Soc. Am., 85, 4,11271143, 1995.Finetti, I.: Structural, stratigraphy and evolution of central Mediterranean, Boll. Geof. Teor. Appl.,24, 247312, 1982.Finetti, I. and Del Ben, A.: Geophysical study of the Tyrrhenian opening, Boll. Geof. Teor. Appl., 3028, 75155, 1986.2620NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Finetti, I. R., Lentini, F., Carbone, S., Catalano, S., andDel Ben, A.: Il SistemaAppenninoMeridionale-Arco Calabro-Sicilia nel Mediterraneo centrale: studio geologico-geosico, Boll.Soc. Geol. It., 115, 529559, 1996.Ghisetti, F.: Relazioni trastruttureefasi trascorrenti edistensivelungoi sistemi Messina-Fiumefreddo, TindariLetojanni e Alia-Malvagna (Sicilia nord-orientale): uno studio microtet- 5tonico, Geol. Rom., 18, 2358, 1979.Ghisetti, F.: Fault parameters in the Messina Strait (southern Italy) and relations with the seis-mogenic source, Tectonophysics, 210, 117133, 1992.Ghisetti, F. and Vezzani, L.: Evidenze di linee di dislocazione sul versante meridionale dei MontiNebrodi e Madonie e loro signicato neotettonico, Boll. Geodesia e Sc. ani, 36, 411437, 101977.Ghisetti, F. andVezzani, L.: Thin-skinneddeformationsof thewesternSicilythrust belt andrelationships with crustal shortening: mesostructural data on the Mt. Kumeta-Alcantara FaultZone and related structures, Boll. Soc. Geol. It., 103, 129157, 1984.Giammanco, S., Palano, M., Scaltrito, A., Scarf, L., and Sortino, F.: Possible role of 15uid overpressure in the generation of earthquake swarms in active tectonic areas:thecaseof thePeloritani Mts. (Sicily, Italy), J. Volcanol. Geoth. Res., 178, 795806,doi:10.1016/j.jvolgeores.2008.09.005, 2008.Giunta, G.: Elementi per un modello cinematico delle maghrebidi siciliane, Mem. Soc. Geol. It.,47, 297311, 1991. 20Giunta, G. and Nigro, F.: Some tectono-sedimentary constraints to Oligo-Miocene evolution ofthe Peloritani Thrust Belt, Tectonophysics, 315, 287299, 1998.Giunta, G. and Somma, R.: Nuove osservazioni sulla struttura dellUnit di Al (M.ti Peloritani,Sicilia), Boll. Soc. Geol. It., 115, 489500, 1996.Giunta,G.,Messina,A.,Bonardi,G.,Nigro,F.,Somma,R.,Cutrupia,D.,Giorgianni,A.,and 25Sparacino, V.: Geologia dei Monti Peloritani (Sicilia NE), Guida allescursione, 77 Riunioneestiva, Palermo, 1998.Giunta, G., Nigro, F., Renda, P., and Giorgianni, A.: The Sicilian-Maghrebides Tyrrhenian Mar-gin: a neotectonic evolutionary model, Mem. Soc. Geol. It., 119, 553565, 2000.Giunta, G., Luzio, D., Tondi, E., De Luca, L., Giorgianni, A., DAnna, G., Renda, P., Cello, G., 30Nigro, F., andVitale, M.: ThePalermo(Sicily)seismicclusterof September2002, intheseismotectonic framework of the Tyrrhenian Sea-Sicily border area, Ann. Geophys.-Italy, 47,17551770, 2004.2621NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Guerrera, F. and Wezel, F. C.: Nuovi dati stratigraci sui ysch oligomiocenici siciliani e consid-erazioni tettoniche relative, Riv. Min. Sic., 145147, 2751, 1974.Guidoboni, E., Ferrari, G., Mariotti, D., Comastri, A., Tarabusi, G., and Valensise, G.: Catalogueof Strong Earthquakes in Italy (CFTI), 461BC1997 and Mediterranean Area 760BC1500,available at: http://storing.ingv.it/cfti4med/, 2007. 5Horike, M., Zhao, B., and Kawase, H.: Comparison of site response characteristics inferred frommicrotremors and earthquake shear waves, B. Seismol. Soc. Am., 91, 15261536, 2001.Konno,K.andOhmachi,T.:Ground-motioncharacteristicsestimatedfromspectral ratiobe-tween horizontal and vertical components of microtremor, B. Seismol. Soc. Am., 88, 1, 228241, 1998. 10Kudo, K.: Practical estimates of site response, State-of-the-Art report, In: Proceedings of theFifthInternational ConferenceonSeismicZonation, 1719October, Nice, France, OuestEditions Nantes, 3, 18781907, 1995.Lachet, C. andBard, P. Y.: Numerical andtheoretical investigationsonthepossibilitiesandlimitations of Nakamuras technique, J. Phys. Earth, 42, 377397, 1994. 15Lermo, J., Francisco, S., andChavez-Garcia, J.: SiteEect Evaluationusingmicrotremors:a review (abstract), EOS, 73, 352, 1992.Malinverno, A. andRyan, W. B. F.: ExtensionintheTyrrhenianSeaandshorteningintheApennines as results of arc migration driven by sinking of the lithosphere, Tectonics, 5, 227245, 1986. 20Mauz, B. and Renda, P.: Tectonic features at the NW-coast of Sicily (Gulf of Castellammare),Implications for the Plio-Pleistocene structural evolution of the southern Tyrrhenian continen-tal margin, Studi Geol. Cam., 2, 343349, 1995.Meletti, C. and Valensise, G.: Zonazione sismo genetica ZS9 A 2 pp. al Rapporto Conclusivo,available at: http://zonesismiche.mi.ingv.it/documenti/A2pp.pdf, 2004. 25Monaco, C. and Tortorici, L.: Tectonic role of ophiolite-bearing terranes in the development ofthe southern Apennines orogenic belt, Terra Nova, 7, 153160, 1995.Mucciarelli, M.: Reliabilityandapplicabilityof Nakamurastechniqueusingmicrotremors: anexperimental approach, J. Earthq. Eng., 2, 625638, 1998.Mucciarelli, M. and Gallipoli, M. R.: The HVSR technique from microtremor to strong motion: 30empirical and statistical considerations, 13th World Conference on Earthquake Engineering,Vancouver, BC, Canada, 16 August 2004, 45, 2004.2622NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Mucciarelli, M., Gallipoli, M. R., and Arcieri, M.: The stability of the horizontal-to-vertical spectralratio of triggered noise and earthquake recordings, B. Seismol. Soc. Am., 93, 14071413,2003.Nakamura, Y.: AMethodfor DynamicCharacteristicsEstimationof SubsurfaceusingMi-crotremor on the Ground Surface, Quarterly Report of Railway Technical Research Institute 5(RTRI), 30, 1, 1989.Nakamura, Y.: Clear Identication of Fundamental Idea of Nakamuras Technique and its Appli-cations, 12th World Conference on Earthquake Engineering, 2656, 2000.Neri, G., Caccamo, D., Cocina, O., and Montalto, A.: Geodynamic implications of earthquakedata in the southern Tyrrhenian sea, Tectonophysics, 258, 233249, 1996. 10Neri, G., Barberi, G., Orecchio, B., and Mostaccio, A.: Seismic strain and seismogenic stressregimes in the crust of the southern Tyrrhenian region, Earth Planet. Sc. Lett., 213, 97112,2003.Nigro, F.: LUnit Longi-Taormina. Stratigraa e Tettonica Delle Coperture Mesozoico-TerziarieDellelemento Peloritano Occidentale, Tesi di Dottorato, Palermo, 276 pp., 1994. 15Nigro, F.: Neotectonic events and kinematic of rhegmatic-like basins in Sicily and adjacent ar-eas, Implications for a structural model of the Tyrrhenian opening, Boll. Soc. Geol. Pol., 69,118, 1998.Nigro, F. andRenda, P.: EvoluzionegeologicaedassettostrutturaledellaSiciliacentro-settentrionale, Boll. Soc. Geol. It., 118, 375388, 1999. 20Nigro, F. and Renda, P.: Un modello di evoluzione tettono-sedimentaria dellavanfossaneogenica siciliana, Boll. Soc. Geol. It., 119, 667686, 2000.Nigro, F. and Renda, P.: Forced mode dictated by foreland fault-indenter shape during obliqueconvergence: the Western Sicily mainland, Boll. Soc. Geol. It., 121, 151162, 2002.Nigro,F.andRenda,P.:Pilo-Pleistocenestrike-slipdeformationinNESicily:theexampleof 25the area between Capo Calav, Bollettino della Societ Geologica Italiana, 124, 377394,2005.Nigro, F. and Sulli, A.: Plio-Pleistocene extensional tectonics in the Western Peloritani area andits oshore, Tectonophysics, 252, 295305, 1995.Ogniben, L.: Nota illustrativa dello Schema geologico della Sicilia nord-orientale, Riv. Min. Sic., 306465, 183212, 1960.Ogniben, L.: Schema introduttivo alla geologia del conne calabro-lucano, Mem. Soc. Geol. It.,8, 453463, 1969.2623NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Ogniben, L.: Schemi paleotettonici anzich paleogeograci in regioni di corrugamento:lesempio della Sicilia, Mem. Soc. Geol. It., 9, 793816, 1970.Panou, A. A., Theodulidis, N., Hatzidimitriou, P., Stylianidis, K., and Papazachos, C. B.: Ambientnoise horizontal-to-vertical spectral ratio in site eects estimation and correlation with seismicdamagedistributioninurbanenvironment: thecaseof thecityof Thessaloniki (Northern 5Greece), Soil Dyn. Earthq. Eng., 25, 261274, 2005.Parolai, S., Bindi, D., and Augliera, P.: Application of the Generalized Inversion Technique (GIT)to a microzonation study: numerical simulations and comparison with dierent site-estimationtechniques, B. Seismol. Soc. Am., 90, 286297, 2000.Parolai, S., Picozzi, M., Richwalski, S. M., and Milkereit, C.: Joint inversion of phase velocity 10dispersionand H/V ratiocurvesfromseismicnoiserecordingsusingageneticalgorithm,considering higher modes, Geophys. Res. Lett., 32, 14, 2005.Parolai, S., Richwalski, S. M., Milkereit, C., and Fh, D.: S wave velocity proles for earthquakeengineering purposes for the Cologne area (Germany), Bulletin of Earthquake Engineering,4, 6594, 2006. 15Peterson, J.: Observations and modeling of seismic background noise, Open-File Report, USGeological Survey, Albuquerque, N M., 93322, 1993.Picozzi, M., Parolai, S., andRichwalski, S. M.: Joint inversionof H/V ratiosanddispersioncurves from seismic noise: estimating the Swave velocity of bedrock, Geophys. Res. Lett.,32, 14, 2005. 20Rodrguez, M., Chvez-Garca, F. J., andStephenson, W. R.: Siteeectsinanalluvial val-ley: a comparison of estimates from earthquake and microtremor records, Proc. 12th WorldConference on Earthquake Engineering, Auckland, New Zealand, New Zealand Society forEarthquake Engineering, Article Id. 1441, 2000.Sambridge, M.: Geophysical inversion with a neighbourhood algorithm I. Searching a param- 25eter space, Geophys. J. Int., 103, 48394878, 1999.Scherbaum, F., Hinzen, K.-G., and Ohrnberger, M.: Determination of shallow shear wave ve-locity proles in the cologne, Germany area using ambient vibrations, Geophys. J. Int., 152,597612, 2003.SESAMEProject: Guidelinesfortheimplementationof theH/V spectral ratiotechniqueon 30ambient vibrations, Measurements, processingandinterpretation, WP12, deliverableno.D23.12, available at: http://sesame-fp5.obs.ujf-grenoble.fr/Papers/HV_User_Guidelines.pdf,(2004).2624NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Tinti, S. and Mulargia, F.: Condence intervals of bvalues for grouped magnitudes, B. Seismol.Soc. Am., 77, 21252134, 1987.Valensise, G. and Pantosti, D.: A 125 kyr-long geological record of seismic source repeatability:theMessinaStraits(southernItaly)andthe1908earthquake(Ms71/2), TerraNova, 4,472483, 1992. 5Wathelet, M.: An improved neighborhood algorithm: parameter conditions and dynamic scaling,Geophys. Res. Lett., 35, L09301, doi:10.1029/2008GL033256, 2008.Wathelet, M., Jongmans, D., and Ohrnberger, M.: Surface wave inversion using a direct searchalgorithmanditsapplicationtoambientvibrationmeasurements,NearSurf.Geophys.,2,211221, 2004. 10Yuncha, Z. A. and Luzon, F.: On the horizontal-to-vertical spectral ratio in sedimentary basins,B. Seismol. Soc. Am., 90, 11011106, 2000.2625NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|Table1. Measurement points, UTMcoordinates, frequenciesof thesignicant peaks, H/Vratios.ID Longitude Latitude Frequency (Hz) Amplitude Cluster1 15.066871 38.125908 0.73 4.49 12 15.063414 38.126892 1.23 7.64 13 15.059912 38.127903 0.76 3.80 13 15.059912 38.127903 1.13 4.23 14 15.057073 38.129833 0.77 8.12 15 15.055261 38.132241 0.82 5.96 16 15.055074 38.127563 1.03 6.17 17 15.056796 38.126292 1.00 5.35 18 15.059567 38.124235 1.00 5.10 18 15.059567 38.124235 1.78 3.19 29 15.062717 38.125018 0.84 6.08 19 15.062717 38.125018 1.10 7.09 110 15.065124 38.124719 0.77 6.48 110 15.065124 38.124719 1.10 4.30 111 15.065406 38.121772 0.83 4.21 111 15.065406 38.121772 1.43 4.51 112 15.062658 38.122990 0.93 4.68 113 15.064252 38.119654 1.53 6.00 114 15.060934 38.121026 1.10 3.56 114 15.060934 38.121026 1.37 2.55 114 15.060934 38.121026 1.73 2.71 215 15.057181 38.121650 1.83 3.22 216 15.054749 38.118991 0.90 4.28 316 15.054749 38.118991 1.53 4.35 216 15.054749 38.118991 10.14 3.10 417 15.051740 38.122412 0.90 2.57 317 15.051740 38.122412 1.70 3.24 217 15.051740 38.122412 2.67 4.94 218 15.049392 38.119468 0.87 2.60 318 15.049392 38.119468 1.50 2.67 219 15.046962 38.114408 0.87 3.12 319 15.046962 38.114408 2.20 4.23 219 15.046962 38.114408 6.08 2.80 420 15.051695 38.116305 0.97 4.00 320 15.051695 38.116305 1.43 4.54 221 15.054697 38.114280 0.87 3.31 321 15.054697 38.114280 1.37 4.41 221 15.054697 38.114280 1.73 4.31 222 15.050217 38.111984 0.87 4.10 322 15.050217 38.111984 1.23 5.39 223 15.058505 38.118050 0.77 3.83 123 15.058505 38.118050 1.30 3.63 12626NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|16 Fig. 1-Geologicalmap ofOliveri town (from SERVIZIOGEOLOGICO DITALIA (2011)-Foglio Geologico 600 BarcellonaPozzodiGotto1:50.000CARG-ISPRA):a)slopedebris;ba)activealluvialdeposits;bb)alluvialand litoral deposits; b2) eluvium and colluvium; bn) inactive alluvial terraced deposits; g2) active strandlines (beaches);gn1-5)terracedmarinedeposits;ROEBiodetritalcalcarenites,claysandsandyclays(RomettaFm.,UpperPliocene- Middle Pleistocene); ASI) Argille Scagliose (Upper Cretaceous);CODb-c) Capo dOrlando Flysch: areanaceous facies (arkose)withalternationofclay-marlylevels(UpperOligocenelowerBurdigalian);PMAa-b)paragneissand gneissicmicaschists,silicatemarbles,quartzitesandaplo-pegmatiticdykes(PMPa)(ApromonteunitPaleo-Proterozoic Permian); MLEa-b-c) Paragneiss, micaschists, marbles, eclogites, quartzites (Mela unit Paleozoic). Fig. 1. Geological map of Oliveri town (from SERVIZIO GEOLOGICO DITALIA (2011) FoglioGeologico 600 Barcellona Pozzo di Gotto 1 : 50000 CARG-ISPRA): a) slope debris; ba) activealluvial deposits; bb) alluvial and litoral deposits; b2) eluvium and colluvium; bn) inactive allu-vial terraced deposits; g2) active strandlines (beaches); gn15) terraced marine deposits; ROEBiodetrital calcarenites, claysandsandyclays(RomettaFm., UpperPlioceneMiddlePleis-tocene); ASI) Argille Scagliose (Upper Cretaceous); CODbc) Capo dOrlando Flysch: areana-ceous facies (arkose) with alternation of clay-marly levels (Upper Oligocenelower Burdigalian);PMAab) paragneiss and gneissic micaschists, silicate marbles, quartzites and aplo-pegmatiticdykes(PMPa) (ApromonteunitPaleo-ProterozoicPermian); MLEabc) Paragneiss, micas-chists, marbles, eclogites, quartzites (Mela unitPaleozoic).2627NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|17 Fig. 2 - Distribution of epicenters of instrumental earthquakes located by the INGV between 1981 and 2011. Fig. 2. Distributionof epicentersof instrumental earthquakeslocatedbytheINGVbetween1981 and 2011.2628NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|18 Fig. 3 HVSR analysis of the microtremor record nr. 2. Fig. 3. HVSR analysis of the microtremor record nr. 2.2629NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|19 Fig.4 HVSR average curves of the clusters in which have been considered to be dominant the structural effects of 23 measures in the area of Oliveri. Fig. 4. HVSR average curves of the clusters in which have been considered to be dominant thestructural eects of 23 measures in the area of Oliveri.2630NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|20 Fig. 5 Dendrogram of the clustering process. Fig. 5. Dendrogram of the clustering process.2631NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|21 Fig. 6 Scatter plot of amplitude versus peak frequency for the four HVSR clusters and their spatial distributions. Fig. 6. Scatter plot of amplitude vs. peak frequency for the four HVSR clusters and their spatialdistributions.2632NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|22 Fig. 7 HVSR peak frequency maps of: cluster B G (left) and cluster R (right). Fig. 7. HVSR peak frequency maps of: cluster BG (left) and cluster R (right).2633NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|23 Fig. 8 Stratigraphic column of the borehole. Fig. 8. Stratigraphic column of the borehole.2634NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|24 Fig. 9 Inversion of the HVSR ellipticity curve of the measurement point n.2. Fig. 9. Inversion of the HVSR ellipticity curve of the measurement point n.2.2635NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|25 Fig. 10 Oliveri. Left) Map of the thickness of the sedimentary cover. Right) Map of the altitude of the seismic bedrock above sea level. Fig. 10. Oliveri. (Left) Map of the thickness of the sedimentary cover. (Right) Map of the altitudeof the seismic bedrock a.s.l.2636NHESSD2, 25972637, 2014Integration of HVSRmeasures andstratigraphicconstraintsP. Di Stefano et al.Title PageAbstract IntroductionConclusions ReferencesTables Figures Back CloseFull Screen / EscPrinter-friendly VersionInteractive DiscussionDiscussionPaper|DiscussionPaper|DiscussionPaper|DiscussionPaper|26 Fig. 11 Oliveri. Sections representing the seismic bedrock and the soft cover. Fig. 11. Oliveri. Sections representing the seismic bedrock and the soft cover.2637