multimethodological approach to investigate chamber tombs in the sabine necropolis at colle del...

Archaeological ProspectionArchaeol. Prospect. 16, 111–124 (2009)Published online 24 March 2009 in Wiley InterScience

351
(www.interscience.wiley.com) DOI: 10.1002/arp.
* Correspondence to: S. PirMonterotondo Sc., Rome, IE-mail: salvatore.piro@itab

Copyright # 2009 John

Multimethodological Approach toInvestigate ChamberTombsin theSabine Necropolisat Colle del Forno(CNR,Rome,Italy)

SALVATOREPIRO* ANDROBERTOGABRIELLI

ITABC^CNR, POBox 10 ^ 00016Monterotondo Sc., Rome, Italy

ABSTRACT Non-destructive geophysical prospecting methods are increasingly used for the investigation ofarchaeological sites, especially where a detailed physical and geometrical reconstruction of struc-tures is required prior to any excavationwork.Often, due to the limited size and depth of an archaeo-logical structure, it may be rather difficult to single out its position and extent because of thegenerally lowsignal-tonoise (S/N) ratio.This canbeovercomeby improvingdataacquisitionandpro-cessing techniques and integrating different geophysicalmethods. In thiswork the results of amulti-methodological surveys, used with the aim of detecting sharp discontinuities (boundary of cavitiesand fractures in the host medium) at the Archaeological Test Site of Sabine Necropolis at ResearchArea of National Research Council of Rome (Montelibretti, Italy) are shown. For the survey a combi-nation of passive andactivemethods (magnetic, ground-penetrating radar (GPR), anddipole^dipolegeoelectric (DDG)), topographical and three-dimensional laser scanner surveysandarchaeologicalexcavationswereused to study the state of conservationof underground tombs.With allgeophysicalmethodsahigh-resolutiondataacquisitionwasadoptedwith theaimofreconstructingaglobalvisionof the study area. Signal processing and amplitude time-slice representation techniques were usedfor theanalysisof GPR data.Thebi-dimensionalcross-correlation techniquewasapplied to enhancethe S/N ratio of the magnetic data. An example of the integration (both qualitative and quantitative)of these results is presented for a portion of the investigated area in the Sabine Necropolis at Colledel Forno (Rome, Italy). Archaeological excavations were then conducted systematically after com-pletingthegeophysicalsurveysandinterpretations (from2000 to 2006),whichconfirmedthelocationand shape of the individual chamber tombswith associated corridors.Copyright# 2009 JohnWiley& Sons, Ltd.

Keywords: SabineNecropolis;magnetic;ground-penetratingradar; resistivity; integration;cavity;excavation

Introduction

Geophysical prospecting methods are con-sidered valuable techniques for non-destructivedetection and mapping of shallow buriedfeatures of archaeological interest. Usually detec-

o, ITABC–CNR, PO Box 10 – 00016taly.c.cnr.it

Wiley & Sons, Ltd.

tion is possible because the physical property ofthe target features contrast with the surroundingmedium. Generally the limited dimensions anddepth of the archaeological remains and thepresence of structures made with the samematerial of the surrounding ground, as well assoil inhomogeneity and other environmentaldisturbances, can yield low signal-to-noise (S/N) ratios. This makes it difficult to define thespatial position and extent of the buried features.

Received 3 February 2009Accepted 13 February 2009

112 S. Piro and R. Gabrielli

Some approaches to overcome this problemhave been considered during the past twodecades (Weymouth, 1986; Scollar et al., 1990;Brizzolari et al., 1992a; Bozzo et al., 1994;Camerlynck et al., 1994; Piro, 1996; Gaffney andGaffney, 2000). They generally consist of improv-ing data acquisition techniques (the recenttechnical instrumental evolution involves theapplication of the high-resolution two-dimen-sional and three-dimensional acquisition tech-niques) and by improving data processingmethods (modelling, other processing techniquesto enhance S/N), (Bernabini et al., 1988; Brizzolariet al., 1992b; Tsokas and Hansen, 1995, 2000;Tsokas and Tsourlos, 1997; Piro et al., 1998, 2007b;Papadopoulos et al., 2006). Our recent approach,discussed in previous papers and applied here, isrelated to the integration of the results obtainedwith different geophysicalmethods (Cammaranoet al., 1997a,b; Piro et al., 2001, 2007a; Kvamme,2003; Gaffney et al., 2004; Diamanti et al., 2005;Ciminale et al., 2007; Finzi et al., 2007; Campanaand Piro, 2008).There is no doubt that a multimethodological

survey, which compares and contrasts magnetic,ground-penetrating radar (GPR) and resistivemethods at the same site, could provide theopportunity to gather significant information.This is possible because each of these methodsevaluates different physical properties of thetarget features with respect to the ground(Neubauer and Eder-Hinterleitner, 1997; Cam-marano et al., 1998; Piro et al., 2000; Kvamme,2006; Cardarelli et al., 2008).All common near-surface geophysical

methods, such as magnetic and resistivity, yieldmeasurements that are average values over agiven volume. In contrast the GPR methodproduces three-dimensional values in the formof radar traces, which cannot be handled in thesame way as most other data. Another aspect ofcombining and contrasting multiple method datais that spatial sampling intervals need to be aslow as possible. In studying archaeologicalremains, an integrated survey must thereforebe performed at a spatial sampling interval that isin accordance with the penetration depth of eachmethod used, with respect to the total integrationvolume. These varying parameters includeelectrode distance in geoelectric methods, and

Copyright # 2009 John Wiley & Sons, Ltd.

sensor height in magnetic surveys. As GPRreflections can be considered as nearly accurate,with respect to the dimension of the searchedbodies, the risk of spatial aliasing is muchhigher. Therefore with GPR spatial samplingneeds to be compatible with the wavelength andappropriate to the dimensions of the targetfeatures in the ground.

The purpose of the work presented here is todemonstrate the results and the advantages ofcombining geophysical methods that measurethe variations of potential magnetic field (gradio-metric method) with active methods thatmeasure the variations of physical propertiesdue to the body’s geometry and volume (GPRand resistive methods). These methods wereused to investigate previously unexplored por-tions of the Sabine Necropolis at Colle del Forno(Roma, Italy).

With all methods a high-resolution dataacquisition has been adopted with the aim ofreconstructing the real shape of the individualstructures in the portion of the area investigated.Signal processing and time-slice representationtechniques have been used for the analysis ofGPR data. The bi-dimensional cross-correlationtechnique has been applied to enhance the S/Nratio of the magnetic data. A representation ofhorizontal apparent resistivity slices at differentpseudodepths is also displayed to show theshape of the individuated anomalous bodies. Anexample of an integration of the variousmethods’ results is presented.

Archaeological site characteristics

The area of Sabine Necropolis at Colle del Forno(Montelibretti, Roma, Italy) is characterized bythe presence of numerous unexplored tombs(Piro and Santoro, 2001; Piro et al., 2001; Godioand Piro, 2005; Cardarelli et al., 2008). From theresults of previous archaeological studies thesetombs are known to be cavities with averagedimensions of about 2� 2� 1.5m3 (Santoro,1977). Their ceilings are about 0.80–1.0m belowthe surface of the ground. Each also contains adromoi (corridor) with average length of 6m,which is between 0.8m and 1.0m in width(Figure 1).

Archaeol. Prospect. 16, 111–124 (2009)

DOI: 10.1002/arp

Figure 1. Horizontalcross-sectionofthechamber tomb (cavity) andtheundergroundcorridor (dromoi); schematicverticalcross-section of the unexplored tomb (after Piro et al., 2001).

Multimethodological Approach to Investigate Chamber Tombs 113

Geologically the bedrock in this area ischaracterized by a series of lithified tuffs, withaverage resistivity values of about 30–80V�m.These units are about 10m thick, lying onQuaternary sand and clay sediments. At theground surface the tuffs are covered with a layerof topsoil between 0.20m and 0.30m in thickness.

Since 1998 a scientific collaboration betweenthe Institute of Technologies Applied to CulturalHeritage (ITABC–CNR) and the Institute ofEtruscan and Italic Archaeology (now ISCIMA–CNR) has been developed with the aim ofsystematically locating and excavating unknownburied structures and to study individual tombs.This project includes the following technologies:(i) topographic surveys, based on the applicationof integrated methodologies (differential globalpositioning and total station systems) with thepurpose of constructing a topographical micro-model; (ii) geophysical surveys using high-resolution geophysical methods with the aimof defining the location, the shape, the dimen-sions and depth of unknown chamber tombswith corridors; (iii) archaeological excavation ofthe individual tombs that were discovered; (iv)survey of the excavated tombs using a laserscanner, with the plan to obtain a very high-density of measurements to construct geometri-cal models of the monumental tombs.


Ground-based remote sensingsurveys

From 1993 to 2005 a series of topographicand high-resolution geophysical surveys tookplace at Sabine Necropolis site, which provideda detailed map of this ancient necropoliscentre. The topographic survey permitted exca-vated tombs and contemporary features to beregistered within a common grid, and remotesensing surveys allowed the location of unknownchamber tombs with associated corridors.

Topographic survey

Taking into account the location of the SabineNecropolis, the Colle del Forno Hill wassurveyed topographically using a Leica D-GPS(Differential Global Positioning System) ModelSR530, integrated with a Trimble ElectronicTotal Station (Model 5600), Figure 2. Duringthe D-GPS survey, in order to reduce errorsdue to the different synchronism between thesignal received from the satellite and the signalemitted from the receiver, the distance betweensatellite and receiver was correlated to the phaseof the signal (Piro et al., 2007a).


DOI: 10.1002/arp

Figure 2. Plan view of Colle del Forno hill (Research Area of National Research Council, near Rome), showing, on thetopography, the positions of the explored tombs and the location of all areas investigated with geophysical methods. Onthe left there is the detail of areas A and B with 9� 9m dimensions, selected as test sites to use the integrated approach.This figure is available in colour online at www.interscience.wiley.com/journal/arp


These measurements were made to constructthe topography of the studied area, and to createa three-dimensional image of Colle del FornoHill as well as to locate both the old and recentlyexcavated tombs and place the geophysicaldata into that database. The topographicalsurvey was also used to construct a three-dimensional model of the terrain, whichrepresents the morphological aspect of the studyarea (Figure 3a–c). For the processing of thesedata SKIPRO software was used and the result-ing maps were constructed using Surfer, Auto-Cad and ArcView softwares.


Geophysical investigations from1993 to 2003

GPR survey (1993)

Taking into account the results of previousresistivity surveys in the southern portion ofthe necropolis (Orlando et al., 1987), a GPRinvestigation was made in May 1993, in a20m� 24m grid (Figure 2). The data werecollected using a GSSI SIR-10 system and500MHz bistatic antennae with a constant off-set. Data were collected with a time window of


DOI: 10.1002/arp

Figure 3. SabineNecropolisat Colle del Forno. (a) three-dimensionalrepresentationof topographicalacquisitiongrid; (b) planofdigital terrainmodel (DTM); (c) shaded representation of DTM.

Figure 4. Sabine Necropolis at Colle del Forno. Area20m� 24m. Time slice from a GPR survey made using a500MHzantenna (GSSI), which corresponds to 25^29 ns (twoway travel time) at an estimated depth of 108 1̂20 cm. Thearrows indicate theposition of the cavityand the corridor.


70 ns, 512 samples per scan, and 16 bit data. Thegrid consisted of 19 reflection profiles 20m inlength with a 1-m profile spacing. Scans werelater adjusted and stacked for a trace spacing of0.20m. These traces were then resampled andamplitudes of reflected waves were placedwithin a volume with the aim of producingslice-maps of all reflective features (Goodmanet al., 1995; Conyers and Goodman, 1997;Conyers, 2004). These amplitude slices werecalculated by creating two-dimensional horizon-tal contour-maps of the averaged absolute valueof the wave amplitude from a specified timevalue across parallel profiles and then placed inthree-dimensional space using x and y horizontalcoordinates for the grid, with z being two-waytravel time (twt).

As an illustration one amplitude slice from 25to 29 ns (estimated depth 108–120 cm) shows areflective feature at one specified depth interval(Figure 4). This map identifies the position andthe extension of a tomb (both its cavity andcorridor) with x coordinates 10–13m and ycoordinates 18–22m within the surface grid.This map is characterized by some linear andisolated artefacts due to the interpolationmethods used that are a product of gridding.Even with these artefacts, this amplitude slice is


in spatial agreement with the location of theburied cavity, which was previously known fromresistivity survey using different electrode arrays(Orlando et al., 1987). This buried feature was


DOI: 10.1002/arp


selected for the further quantitative integrationpresented here.

Integration of different geophysical methods(1996)

Methodological approachWith the aim of applying the integrated

geophysical approach, the GPR data from 1993was used to select two test-areas (areas A and B),with dimensions of 9m� 9m where two tombswere believed to exist. The first was centred onthe position of the detected tomb described aboveand the second was directly adjacent (Figure 2).In these two areas measurements were re-collected in June 1996 using GPR, resistivityand gradiometric methods; much smaller spatialintervals were used to provide greater resolutionof this buried feature, with the aim of developinga quantitative integration.It is important to remember that to have a

possible quantitative integration of the resultsobtained using different geophysical methods itis necessary, after the elaboration developed foreach data set, to make all obtained valuesnumerically comparable (Piro et al., 2000). Todevelop this approach it is necessary to make thefollowing assumptions.Let us indicate byM the number of geophysical

methods applied in the surveying of a given area.Even if not all the adopted methods lead to atomographic reconstruction of the geophysicalproperties underground, we suppose that theresult of the surveymi, with i¼ 1,. . ..,M, is a datafunction fmi

rð Þ, where r is a three-dimensionalposition vector in a fixed Cartesian referencesystem with the x–y plane on the ground surfaceand the z axis pointing downwards. In otherwords, fmi

rð Þ is an indicator of the presence of theanomaly source observed with the method mi.Further, let us indicate by zm the depth at

which the archaeological target is to be expectedobserving the behaviour of fmi

rð Þ and withfmi

x; yð Þ the restriction of fmirð Þ on the plane

z¼ zm.The aim of this elaboration was to perform a

quantitative integration of different methodsused in the study of the necropolis area. In orderto achieve this purpose, the single data-functionsfmi

x; yð Þmust be numerically comparedwith each


other. This means that we had to work with non-dimensional values of the parameters defineddirectly from the experimental measurements.

Let us start with the definition of ~fmias the

‘undisturbed’ value of the data function fmix; yð Þ,

i.e. the response of the mi-th method if noanomalous bodies are present underground.

Thus the function fmix; yð Þ � ~fmi

indicates thedeparture of the physical properties investigatedby the mi-th method, at the pseudodepth zm fromthe uniform half space. Carrying out a normal-ization of this function, wemay define a new datafunction

Fmix; yð Þ ¼

fmix; yð Þ � ~fmi

��

max fmix; yð Þ � ~fmi

��

(1)

that obviously satisfies the condition

0 � Fmix; yð Þ � 1 (2)

It is clear that Fmix; yð Þ is equal to 1 at those

points where the departure of the data-functionfrom the undisturbed value is maximum. The useof the normalization (Equation 1) cancels out anydependence of Fmi

x; yð Þ on the strength coeffi-cients in the anomaly geometry. Further, thecondition (Equation 2) permits us to assert thatthe new data-function, being deprived of thephysical dimensionality, is a suitable indicator ofsource occurrence (ISO) of archaeological targets.

These assumptions allow for the proposal ofthe hypothesis that each geophysical methodinvestigates the same event, i.e. the presence ofanomalous volumes underground. We are there-fore able quantitatively to integrate the results ofthe different methods. More precisely, we pre-sent the information obtainable by performing amean of the ISO functions, that is:

F x; yð Þ ¼ 1

M

XM

mi

Fmix; yð Þ (3)

which is an indicator of the spatial distribution ofthe anomaly’s source occurrence detected by atleast one of the methods used. In fact, F x; yð Þ isequal to 1 only at those points where themethodsgive the maximum anomaly pattern and is equalto 0 where all methods reveal no geophysicalanomalies. At the other points F x; yð Þ, being just a


DOI: 10.1002/arp


mean of the ISO functions, takes into account thecontributions of theMmethods (Piro et al., 2000).

Integrated geophysical surveyIn the two selected test areas, the GPR data

were collected using a new three-dimensionalacquisition method that employed a very close-spaced sampling interval within a regular gridcomposed of 46 south–north parallel profiles,each 9m in length. Along each profile radartraces were regularly spaced every 0.20m with atransect spacing also of 0.20m. Reflection traceswere collected within a 100 ns time window intrue amplitude mode with no recorded gain orfiltering. The velocity of radar travel wascalculated at 0.75mns�1 from previous investi-gations (Malagodi et al., 1996). All data werecollected with a GSSI SIR-10A with 16 bitsdynamic range, 1024 samples per scan and atrace stacking of 8. Bistatic 100MHz antennaewere offset 0.95m. The GPR data were processedusing Radan-3 software and then converted toSEG-2 seismic format and further processedusing Seistrix 3 Interpex software (Malagodiet al., 1996). All the reflection traces containedlow-frequency noise, most probably due tochanges in ground–antenna coupling, and con-sequently producing a low S/N ratio. This noiseseems to be related directly to the nominalfrequency of the antenna. To reduce the low- andhigh-frequency noise we applied a filter thatmaintains the most important frequency contri-butions for the signal used. To select that filter aspectral analysis of the traces collected corre-sponding to the centre of the cavity and outsidethe cavity along each profile was made (asdescribed in a previous paper, Malagodi et al.,1996). The amplitude spectra of all traces, alongeach profile, show that many components are inthe interval between 0 and 350MHz (Malagodiet al., 1996). To decrease these disturbances due tolow- and high-frequency noise and enhance theS/N ratio a band-pass filter was applied to allfield profiles. Finally, with the aim of obtaining aplanimetric vision of the anomalies, the ampli-tude slice representation technique was appliedusing all processed profiles (Piro et al., 2000,2003). A time-slice represents a cut at a specifiedsample value, for instance a time-value, acrossthe radar scan, reported as a function of the x and


y horizontal coordinates. The final image wasobtained by contouring the radar intensities atthe specified time value across parallel profiles.Also, to have the depth progression of thereflection features that approximates the three-dimensional body geometry to create a pseudo-image, a sequence of time slices was calculatedevery 5 ns along the time window, for increasingtimes with such a constant time interval in orderto realize the most complete image reconstruc-tion. Figures 5a and 6a show the results of thenormalized amplitude slices corresponding tothe time 24–28 ns (twt), respectively for Area Aand Area B.In each 9� 9m test area ten parallel dipole–

dipole resistively profiles were collected with aconstant separation of 1m. The measurementswere performed with the BRGM SYSCAL/R2resistivity meter (Version 1990), which canoperate in the stacking signal enhancementmode. Each apparent resistivity datum wasobtained with dipole–dipole Geoelectric arrays(DDG) of 0.5m of length and attributed at apseudodepth equal to half the spacing betweenthe centres of the emitting and receiving dipolesalong the median vertical axis through the linejoining them. The step-by-step continuous dis-placement of the dipoles along each profileprovided a dense network of about 152 datapoints in the vertical pseudosection. Themeasured values of the apparent resistivities inthe ten sections of each areawere then consideredtogether in order to obtain a three-dimensionalmatrix. In this way we were able to producehorizontal normalized apparent resistivity slicesat different pseudodepths, which are displayedin Figures 5b and 6b, respectively for Area A andArea B (Cammarano et al., 1998).For the magnetic survey the measurements

were carried out using a GEOSCAN FM 36fluxgate gradiometer. This instrument measuresthe vertical gradient of the magnetic componentZ, with a fixed intersensor spacing of 0.5m. Datawere collected along parallel S–N profiles with a0.5m intertraverse distance, after taking intoaccount the depth and dimensions of theexpected targets. The measurements were col-lected with a sampling interval of 0.5m alongeach profile. The bottom sensor of the gradi-ometer was kept at a constant height of 0.3m


DOI: 10.1002/arp

Figure 5. SabineNecropolisatColledelForno: testareaA(9m� 9m). (a)NormalizedGPRtimeslice (100MHz) correspondingtothe timewindow24^28 ns (twt). (b) Normalizedresistivity tomography. (c) Normalized two-dimensionalcross-correlationofmag-netic survey. (d) Integration (sum) ofallnormalizedimages.The imagescorrespond to the estimateddepthof1m.Thenormalizedcolour scale corresponds to:Min (0),Max (1).


from the ground surface. After common pre-processing techniques (despiking, filtering andrearranging), the results were represented as acoloured contour map of the residual values ofthe gradient of the Z component (Figure 7,anomalies with x coordinates 10–20m and ycoordinates 35–55m).Even if the magnetic measurements, in this

particular portion of the site, present a good S/Nratio, to use this data set for the development ofthe quantitative integration approach, the bi-dimensional cross-correlation technique wasapplied to achieve a better estimate for the centreof the prospected body. The problem of recover-ing the anomalies masked by the noise is tochoose and to apply suitable techniques toimprove the S/N ratio. Treitel and Robinson(1969) showed in detail that if we have at least arough estimation of the shape, dimensions andphysical properties of the expected body, the bestfilter (operator) is the theoretical anomaly of the


structure itself. This operator is ‘an absoluteoptimum’ in the case of Gaussian noise. In theparticular case of coherent noise (autocorrelated)the best operator is still the theoretical anomalythat minimizes the prevalent frequencies of theautocorrelated noise. The application of thisoperator, in the space domain, consist in thecross-correlation of the raw field data with thecalculated theoretical anomaly resulting in across-correlation function that is a measure ofsimilarity between sets of data (Piro et al., 1998).

If the target anomaly has a shape close to thatof the theoretical one, the cross-correlationtechnique produces a signal with a shape similarto that of the autocorrelation of the theoreticalanomaly. Whenever the two sets of input dataare close, their cross-correlation usually willbe positive and their value large. In the oppositecase, some of the products will be positivewhile others are negative, and the sum tends tobe smaller. It is evident that above all it is


DOI: 10.1002/arp

Figure 6. Sabine Necropolis at Colle del Forno: test area B (9m� 9m).Forother information see Figure 5.


necessary to calculate the synthetic magneticanomalies for the three-dimensional anomalousbodies to be used as operators in the bi-dimensional cross-correlation, i.e. their dimen-sions, physical parameters and depths (Piro et al.,1998).

In order to make the correlogram more mean-ingful it is advantageous to normalize it in someway. As the maximum output from the auto-correlation is normalized to unity when the dataset is matched, the interpretation of the cross-correlograms is clearer by dividing the outputcross-correlated values by the maximum autocor-relation values for each of the applied operators. Ifthe normalized cross-correlated values approachunity for one output, the best synthetic operatorwill fit the field data. From this informationand our assumptions about the other source/field parameters, it seems possible to determinethe location and limit the depth-range ofthe investigated area (for details see Piro et al.,1998).

The theoretical magnetic anomalies for a bodywith dimensions of 1� 1� 1 (grid units) were


computed using the relations proposed in theprevious paper by Piro et al. (1998). Thecalculations were performed with geomagneticparameters: F¼ 45000 nT, I¼ 558, D¼ 08 anduniform magnetization M of the body. Foruniform susceptibility contrast, different valueswere used in relation to the different archae-ological site conditions. In Figures 5c and 6c thenormalized two-dimensional cross-correlationare shown.As described above, in the methodological

approach section, the obtained data in these newsets without dimensional parameters can beconsidered as a sign of the presence or absenceof the anomalous source. These values rangebetween 0 and 1 and then can be represented asan image. Taking into account that each geophy-sical method measures the variations due to thesame target, it is possible to integrate eachnormalized data set to produce one map thatcontains the contribution related to each methodused.The results of this multisensor integration are

shown in Figures 5d and 6d. It should be noticed


DOI: 10.1002/arp

Figure 7. SabineNecropolisatColledel Forno.Fluxgategradi-ometergriddedgreyshademap of the total surveyedarea.


that these images represent the integration F(x,y)(as the sum) of all methods and the results showthe position of the buried features of interest. Inparticular the result of Area A is very impressive,where both the corridor and chamber of the tombappear clearly (Figure 5d). It is also interesting to


compare this result with the result in the adjacentarea, Figure 6d. This second result still indicatesthe position of a tomb, but we argue that thepreservation state is quite different. Indeed, it ispossible to note the clear presence of the chamberonly with the geoelectrical method, whereas themagnetic and GPR methods indicate only theanomaly due to the dromos. So the integrationavoids misinterpretations, easily possible with aone-method survey.

These outcomes were followed in 2000 byarchaeological excavations, which confirmed thelocation, depth, shape and the geophysicalinterpretation of these two tombs. In particularthe first tomb was found to be an empty cavity,whereas the second tomb (in Area B) was foundwith its cavity filled with collapsed tuff blocksfrom above and with sediments entering thecavity from the dromoi (Piro and Santoro, 2001).

Large magnetic survey (2003)Taking into account the previous geophysical

surveys and the results of the integratedapproach (Piro et al., 2000; Piro and Santoro,2001), we decided to use the gradiometricmethod to make an extensive survey in thenorthern area of the necropolis on an unexploredportion of the hill, Figure 2. For the magneticsurvey conducted in May 2003 the Geoscan FM36 fluxgate gradiometer was used. The bottomsensor was kept at a constant height of 0.30mfrom the ground surface. The 60m� 160m totalsurvey area was subdivided into 10� 10msquares and all profiles were recorded parallelin S–Nprofiles spaced 0.50m apart. Themagneticdata, after the usual preprocessing techniquessuch as despiking, filtering and rearranging (Piroet al., 1998, 2001; Piro and Santoro, 2001; Godioand Piro, 2005), are represented by the residualvalues of the gradient of the Z component for theall assembled squares (Figure 7).

Analysis of this map shows that the largersurvey area is characterized by many dipolaranomalies in a range of �40, þ35 nT. Theseanomalies are spatially organized as pseudo-linear or circular features and are characterizedby a prevalence of the negative component of thedipole (Figure 7). The very strong negativenucleus was probably produced by a buriedstructure that produced a negative susceptibility


DOI: 10.1002/arp


contrast with respect to the surroundingmaterial. This probably was caused by an emptycavity or in the case of a corridor, one full ofsediments with lower susceptibility values of thesediment fill with respect to that of the surround-ing tuff layer.

Taking into account the results of previousmagnetic surveys and the subsequent archae-ological excavations made in the period 2000–2001 (Piro and Santoro, 2001), we hypothesizedthat these magnetic anomalies were produced bychamber tombs with associated corridors. Fromthe analysis of the map shown in Figures 7and 8a and b it is possible to see that the N–Wportion of the investigated area is characterizedby the presence of a large and very intensenegative nucleus of magnetic anomaly (Figure 7,anomaly with x coordinates 2–20m and ycoordinates 125–155m). This anomaly, whichoccupies a large surface, has been interpreted asbeing due to the presence of a combination ofmany cavities associated with a single longcorridor.

This geophysical anomaly was excavated inJuly 2005 and the archaeologists found, exactly inthe position mapped geophysically, a king’s

Figure 8. SabineNecropolisatColledelForno.(a)PositionoftheKphotograpgh of the tombafter the excavation.


tomb of Eretum Sabine town. It is characterizedby three different rooms with squared dimen-sions of about 2� 2� 2m, filled with collapsedtuff blocks from the top, and a very long corridor27m length, filled with sediments (Figure 8c).

Laser scanner surveyAfter the excavation it was found that the

archaeological structure (characterized by threerooms and one corridor) was damaged, withmany hairline fractures of the walls and thesurfaces were in places crumbling away. Takinginto account this situation and the necessity tohave the possibility to study the complexgeometry of this Sabine tomb in the future, wedecided to use laser scanner technology to placein space images of the surfaces of the walls of thetomb immediately after the archaeological exca-vations.The three-dimensional laser scanner tech-

nique is a method that can produce very highaccuracy measurements of the surfaces of three-dimensional bodies and their details. Afterthe excavation of the Sabine tomb indicatedabove, these data were collected with a CallidusCP 3200 system (Trimble). The measurements

ing’stomb(n.XXXIV). (b)Detailofthegradiometricmap.(c)Aerial


DOI: 10.1002/arp

Figure 9. Digital view of the excavated tomb XXXIV, after theground-based three-dimensional laser scanner survey.


were carried out from view–points with asampling of 0.1258, for horizontal resolution,and a sampling of 0.258, for vertical resolution.A digital camera was then employed to collect

pictures of the investigated surface and 3D-Extractor software was used to automaticallyplace the pictures into a polygonal model,producing a three-dimensional geometricalreconstruction. This software stores the three-dimensional measurements and produces athree-dimensional image of the excavated tomb(Figure 9). The resulting geometrical modelproduced a high-resolution image of the monu-ment and its details.

Conclusion

The results obtained in the Sabine Necropolis atColle del Forno Hill indicate that the plan of alarge portion of the hypothesized chamber tombswith associated corridors can be identified andmapped from geophysical data. The location,depth, size and general structure of the buriedfeatures were produced using a multimethodo-logical geophysical approach. Topographicalsurveys, using high-resolution acquisition tech-niques allowed for micromodels of the groundsurface. The use of three-dimensional laserscanner technology allowed for accurate three-dimensional images of the main excavated tomb,producing a digital model with geometrical and


other useful visible information provided bydigital camera images.

The recent archaeological excavations (from2000 to 2005) based on our geophysical mappinghave permitted us to define a new cultural phaseof Eretum town, through the study of the funeralwealth associated the tombs. This phase is relatedto the second half of sixth and the first ten years ofthe fifth centuries BC. We determined that thenorthern sector of the hill is characterized by thepresence of a monumental tomb that is directlyconnected with the location of the ancient SalariaWay. In the monumental tomb (n. XXXVI)(Figure 8), the human remains were in a centralroom inside a wooden box with two ‘bucchero’chalices and a throne in the west room.

This significant discovery was made possibleonly by using the geophysical results presentedhere. When these geophysical results wereintegrated with the archaeological materials itwas possible to conclude that the SabineNecropolis at Colle del Forno was used continu-ously from the latter half of the sixth century tothe fifth century BC.

This project is still in progress and newtopographical and geophysical surveys havebeen planned with the aim to integrate allinformation using GIS for the study area.

Acknowledgements

We thank Professor L.B. Conyers, Editor ofArchaeological Prospection, for his suggestions toenhance the quality of this paper. We also thankPaola Santoro (ISCIMA-CNR) for the usefularchaeological information and Daniele Verrec-chia (ITABC–CNR) for his valuable assistanceduring the geophysical surveys.

References

Bernabini M, Brizzolari E, Piro S. 1988. Improve-ment of signal-to-noise ratio in resistivity profiles.Geophysical Prospecting 36: 559–570.

Bozzo E, Lombardo S, Merlanti F, Pavan M. 1994.Integrated geophysical investigations at an Etrur-ian Settlement in Northern Appennines (Italy).Archaeological Prospection 1: 19–35.

Brizzolari E, Piro S, Versino L. 1992a.Monograph onthe Geophysical exploration of the Selinunte


DOI: 10.1002/arp


Archaeological Park. Bollettino di Geofiscia Teoricaed Applicata XXXIV: 134–135.

Brizzolari E, Ermolli F, Orlando L, Piro S, Versino L.1992b. Integrated geophysical methods in archae-ological surveys. Journal of Applied Geophysics 29:47–55.

Camerlynck C, Dabas M, Panissod C. 1994. Com-parison between GPR and four ElectromagneticMethods for stone features characterization: anexample. Archaeological Prospection 1: 5–17.

Cammarano F, Mauriello P, Patella D, Piro S. 1997a.Integrated geophysical methods for archaeologi-cal prospectings. In Volcanism and Archaeology inMediterranean Area, Cortini M, De Vivo B (eds).Research Signpost Publishers: Trivandrum; 7–34.

Cammarano F, Mauriello P, Piro S. 1997b. High-resolution geophysical prospecting with inte-grated methods. The Ancient Acropolis of Veio(Rome, Italy). Archaeological Prospection 4: 157–164.

Cammarano F, Mauriello P, Patella D, Piro S, RossoF, Versino L. 1998. Integration of high-resolutiongeophysical methods. Detection of shallow depthbodies of archaeological interest. Annali di Geofi-sica 41: 359–368.

Campana S, Piro S. 2008. Putting everythingtogether: GIS based data integration andinterpretation. In Seeing the Unseen. Geophysicsand Landscape Archaeology, Campana S, Piro S(eds). Taylor and Francis: London; 325–330.

Cardarelli E, Fischanger F, Piro S. 2008. Integratedgeophysical survey to detect buried structures forarchaeological prospecting. A case-history atSabine Necropolis (Rome, Italy).Near Surface Geo-physics Journal (EAGE) 6: 15–20.

Ciminale M, Becker H, Gallo D. 2007. Integratedtechnologies for archaeological investigations; theCelone Valley project. Archaeological Prospection14: 167–181.

Conyers LB. 2004. Ground-penetrating Radar forArchaeology. AltaMira Press: Walnut Creek, CA.

Conyers LB, Goodman D. 1997. Ground-penetratingRadar: An Introduction for Archaeologists. AltaMiraPress: Walnut Creek, CA.

Diamanti NG, Tsokas GN, Tsourlos PI, Vafidis A.2005. Integrated interpretation of geophysicaldata in the archaeological site of Europos(northern Greece). Archaeological Prospection 12:79–91.

Finzi E, Praticelli N, Vettore L, Zaja A. 2007.Multitemporal geophysical survey of a Romanbath complex in Montegrotto Terme (Padova,northern Italy). Archaeological Prospection 14:182–190.

Gaffney C, Gaffney V. 2000. Non-invasive investi-gations at Wroxeter at the end of twentieth cen-tury. Archaeological Prospection 7: 65–80.

Gaffney V, Patterson H, Piro S, Goodman D, Nishi-mura Y. 2004. Multimethodological approach to


study and characterise Forum Novum (Vescovio,Central Italy). Archaeological Prospection 11: 201–212.

Godio A, Piro S. 2005. Integrated data processing forarchaeological magnetic surveys. The LeadingEdge (SEG) 24: 1138–1144.

Goodman D, Nishimura Y, Rogers JD. 1995. GPRtime-slices in archaeological prospection. Archae-ological Prospection 3: 85–89.

Kvamme KL. 2003. Multidimensional prospectingin North America Great Plains Village Sites.Archaeological Prospection 10: 131–142.

Kvamme KL. 2006. Integrating multidimensionalgeophysical data. Archaeological Prospection 13:57–72.

Malagodi S, Orlando L, Piro S, Rosso F. 1996.Location of archaeological structures using GPRmethod: three-dimensional data acquisition andradar signal processing. Archaeological Prospection3: 13–23.

NeubauerW, Eder-Hinterleitner A. 1997. Resistivityand magnetics of the Roman Town Carnuntum,Austria. An example of combined interpretationof prospection data. Archaeological Prospection 4:179–189.

Orlando L, Piro S, Versino L. 1987. Location ofsubsurface geoelectric anomalies for archaeologi-cal work: a comparison between experimentalarrays and interpretation using numericalmethods. Geoexploration 24: 227–237.

Papadopoulos NG, Tsourlos PI, Tsokas GN,Sarris A. 2006. Two-dimensional and three-dimensional resistivity imaging in archaeologicalsite investigation. Archaeological Prospection 13:163–181.

Piro S. 1996. Integrated geophysical prospecting atRipa Tetta Neolithic site (Lucera, Foggia – Italy).Archaeological Prospection 3: 81–88.

Piro S, Santoro P. 2001.Analisi del territorio di Colle delForno (Montelibretti, Roma) e scavo nella necropolisabina arcaica, Orizzonti – Rassegna di Archeolo-gia, Vol. II. Ist. Editoriali e Poligrafici Internazio-nali: Roma; 197–212.

Piro S, Samir A, Versino L. 1998. Position and spatialorientation of magnetic bodies from archaeologi-cal magnetic surveys. Annali di Geofisica 41: 343–358.

Piro S, Cammarano F, Mauriello P. 2000. Quantitat-ive integration of geophysical methods forarchaeological prospection. Archaeological Pro-spection 7: 203–213.

Piro S, Tsourlos P, Tsokas GN. 2001. Cavity detec-tion employing advanced geophysical tech-niques: a case study. European Journal ofEnvironmental and Engineering Geophysics 6:3–31.

Piro S, Goodman D, Nishimura Y. 2003. The studyand characterisation of Emperor Traiano’sVilla (Altopiani di Arcinazzo – Roma) using


DOI: 10.1002/arp


high-resolution integrated geophysical surveys.Archaeological Prospection 10: 1–25.

Piro S, Peloso D, Gabrielli R. 2007a. Integratedgeophysical and topographical investigations inthe territory of Ancient Tarquinia (Viterbo, Cen-tral Italy). Archaeological Prospection 14: 191–201.

Piro S, Sambuelli L, Godio A, Taormina R. 2007b.Beyond image analysis in processing archaeo-magnetic geophysical data: examples on chambertombs with dromos. Near Surface Geophysics Jour-nal (EAGE) 6: 405–414.

Santoro P. 1977. Colle del Forno, loc. Montelibretti(Roma). Relazione di scavo sulle campagne. 1971–1974. nella necropoli. Atti dell’Accademia Nazionaledei Lincei XXXI: 211–298.

Scollar I, Tabbagh A, Hesse A, Herzog I. 1990.Archaeological Prospecting and Remote Sensing.Cambridge University Press: Cambridge.


Treitel S, Robinson EA. 1969. Optimum digitalfilters for signal to noise ratio enhancement. Geo-physical Prospecting 17: 248–293.

Tsokas GN, Hansen RO. 1995. A comparison ofinverse filtering and multiple source Wennerdeconvolution for model archaeological pro-blems. Archaeometry 37: 185–193.

Tsokas GN, Hansen RO. 2000. On the use of com-plex attribute and the inferred source parameterestimates in the exploration of archaeologicalsites. Archaeological Prospection 7: 17–30.

Tsokas GN, Tsourlos P. 1997. Transformation of theresistivity anomalies from archaeological sites byinversion filtering. Geophysics 62: 36–43.

Weymouth JW. 1986. Geophysical methods ofarchaeological site surveying. In Advances inArchaeologicalMethod and Theory, Vol. 9, AcademicPress: 311–395.


DOI: 10.1002/arp

multimethodological approach to investigate chamber tombs in the sabine necropolis at colle del...

Documents