ReviewShallow geophysical survey at the archaeological site of San Miguel Tocuila,Basin of MexicoA. Arciniega-Ceballosa,*, E. Hernandez-Quinterob, E. Cabral-Canob, L. Morett-Alatorrec, O. Diaz-Molinab,A. Soler-Arechaldeb, R. Chavez-SegurabaDepartamento de Vulcanologia, Instituto de Geosica, Universidad Nacional Autonoma de Mexico. Cd Universitaria, Mexico D.F. 04510, MexicobDepartamento de Geomagnetismo y Exploracion, Instituto de Geosica, Universidad Nacional Autonoma de Mexico. Cd. Universitaria, Mexico, D.F. 04510, MexicocUniversidad Autonoma Chapingo, Carr. Estatal Texcoco, Mexicoarti cle i nfoArticle history:Received 31 October 2008Received in revised form20 January 2009Accepted 28 January 2009Keywords:Seismic refraction tomographyGround penetration radarMagnetometryMoundTlatelTexcocoabstractAcombinednearsurfacegeophysical surveyconductedinSanMiguel Tocuilashowthatgeophysicalmethodsoffer thepossibilitytocharacterizeandreconstruct thegeometryof subsurfacestructureswithout destroying the deposits, providing a way to nd solutions to the questions of archaeological orengineeringsignicance. Thesurveyconsistedof theapplicationmagnetometry, seismic refractiontomography(SRT)andgroundpenetratingradar(GPR)withinadepthrangeof10 m. BeforeSpanierconquest San Miguel Tocuila was a very prominent suburb of the main Aztec ceremonial complex locatedontheeasternmarginof LakeTexcoco, central Mexico, whereseveral moundsknownasTlatelesinNahuatllanguagehavebeenidentied. Nowadays, therapidexpansionofMexicoCitysmetropolitanarea within the last four decades has strongly inuenced Tocuilas environment and has compromisedseveral of its archaeological and ancient human settlements. This study shows how the high resolutionimaginingof non-invasivegeophysical methodologiesinadditionwithsurfacearchaeological studies[Parsons, J.R., 1971. Prehistoric Settlement Patterns intheTexcocoRegion, Mexico. Memoirs of theMuseum of Anthropology, University of Michigan, Number 3] provide different kinds of information thatcharacterize the subsoiland a buried structure. Basedon the history of theancient settlementsin thezone andconsideringthecharacteristicsofshapeandheight ofthestructure, we interpretedthatthesubsurface images obtained depict a buried Tlatel which corresponds to a ceremonialcivic center of LateAztec times. 2009 Elsevier Ltd. All rights reserved.1. IntroductionInnear surface explorationthe combinationof geophysicalmethods have proved extremely useful across a wide spectrum ofdisciplines such as archaeology, geotechnical engineering andenvironmental geology. The application of geophysical techniqueshas become more accessible since the development of newportabletechnologies;theiruseinmodernarchaeologyincludesmagnetometry, electrical resistivity tomography, ground pene-tratingradar(GPR)andmorerecentlyseismicrefractiontomog-raphy(SRT)(e.g. Chavezetal., 2001;DeDomenicoetal., 2006;Leucci, 2003; Leucci et al., 2007). In this work we describea methodology to characterize subsurface structures applying non-destructivetechniques whichallowtoidentifyshallownaturaland/or man-made structures providing useful information forarchaeological studies and urban development. We gave anexampleofhowusegeophysical techniquestoinferPrehispanicburiedstructures. TheidenticationofPrehispanicburiedstruc-tures is relevant because they affect the geotechnical characteristicsof the surrounding subsoil andthe applicationof geophysicalmethodscanbeusefulonfuture modernstructuredesigninthearchaeological site of San Miguel Tocuila in the Texcoco region. Theapproach includes the application of three geophysical techniques:magnetometry, SRT and GPR in order to measure the variability ofgeophysical properties at different depth ranges.The Texcoco region is located towards the eastern portion of theBasin of Mexico, 30 km away from Mexico City (Fig. 1). This regionlies to the southern end of the North-American grasslands withina volcanic area active during the Late Pleistocene and Early Holo-cene (Arroyo-Cabrales et al., 2006; Rueda et al., 2006; Meier et al.,2007). The intense volcanic activity contributed to the morphologyandclosureof thebasinthat facilitatedthedevelopment of anextensivelacustrineenvironment whereLakeTexcocowas themost prominent feature. DuringPrehispanicandColonial times*Corresponding author.E-mail address: maac@geosica.unam.mx (A. Arciniega-Ceballos).Contents lists available at ScienceDirectJournal of Archaeological Sciencej ournal homepage: ht t p: / / www. el sevi er. com/ l ocat e/ j as0305-4403/$ see front matter 2009 Elsevier Ltd. All rights reserved.doi:10.1016/j.jas.2009.01.025Journal of Archaeological Science 36 (2009) 11991205human settlements were intimately related to the Texcoco lacus-trinesystem. Humanactivitypromotedmodicationsto thelakesystem during the early stages of occupation, later on, the intensehumanactivitydevotedsubstantial resourcesthatdisruptedthelacustrine environment resulting in the death of the lake system.ThevillageofTocuilaislocatedonwhatusedtobetheeasternmarginoftheformerLakeTexcoco. Thisareaischaracterizedbya late Pleistocene volcano-sedimentary sequence particularly richin both archaeological and paleontological deposits (Morett-Alatorre et al., 1998, 2001; Siebe et al., 1999; Gonzalez et al., 2006;Arroyo-Cabrales, 2006). Tocuila became one of the most prominentsuburbs of the main civic ceremonial complex of the Aztecs, duringthe Late Post classic between 1300 and 1519 (Parsons, 1971).Over the last 50 years, the demise of the lake system togetherwith theclose proximity andrapidexpansion oftheMexicoCitymetropolitan area strongly inuenced Tocuilas environment. Theurbandevelopmenthascompromisedseveralarchaeologicalandancient human settlements (Menze et al., 2005). However,unknown buried Prehispanic structures may affect the behavior ofmodernstructuresduringearthquakesduetothesoil-structureinteraction (Barba, 2003). In consequence the application ofgeophysical methods has become relevant to aid in the exploration,conservation and better urban development of the region.1.1. The importance of a Tlatel in Prehispanic Central MexicoMounds are important structures identied around the world asareas of ancient human settlements and are considered a classicalplaceforarchaeological exploration. Naturalorarticial moundsaretopographicallyhigherelevationsof different shapes, whichhavebeenbuilt or adaptedfor manysocial purposesincludingceremonial, burial, residential anddefensive. Thestudyof theirspatial occurrence reveals the organizationof the rst humansocieties (Squier et al., 1998; Menze et al., 2005).IntheBasinof Mexico, mounds knownas a Tlateles, fromNahuatl language, were sites of temporal human occupation sinceancient times. They were built on the lakebed through successiveconstructionof conning walls andtheir inll of surroundingmaterials that included silty sand, rocks and heterogeneousbuildingdebris (Apenes, 1943; Parsons, 1971). Dimensions andgeographic density of suchfeatures represent the engineeringsolutions from a well-developed society living on the shoreline ofalakethatexhibitedahighlyseasonallyuctuatingwaterlevel.Tlatelesalongwithchinampas (anothernativestructuredevel-oped for agricultural purposes) were the dominant type ofconstructionandlandformmodicationintheBasinof Mexicoprior tothe Spanishconquest in1521. Nowadays, Tlateles aremostlyremainsofthePrehispanicsettlementsystemsbutaresopervasive in the Texcoco region, that fromthe engineeringperspectiveisimportanttoidentifyburiedTlateles;denetheirdepth, extension, andcontrast of geophysical propertiesduetotheir inuenceonsurfaceproperties andconsequentlyoncivilstructure stability (Barba, 2003; Santoyo-Villa et al., 2005).The relevance of Tlateles in Central Mexico was rst investigatedby Apenes (1943) and Parsons (1971), making their investigation animportant archaeological activity. The archaeological study ofmounds and aerial photography by Parsons (1971) reports severalTlateles in the Texcoco region. Parsons classication criteria is onlybasedontheheightandstair-steppedshapedividingTlatelesintwo types: domesticresidential or ceremonialcivic center. Fig. 2shows theremainsofaTlatelwhichwaspart ofthemainurbanAztec center in Huexotla, Texcoco (Parsons, 1971). During the timesof Parsons study, Tocuila was mostly a rural area of intense agri-cultural activity primarily based on corn and alfalfa crops. Over thelast40 yearstheTexcocoregion hasbeenurbanized,althoughinTocuila residential units were sparsely distributed until 1997, whenthe installation of a centralized drainage systems, paved streets andanother urbanservices wereintroduced. This shift inlandusepromotedthemodicationof therural environment andmanyTlateleshavebeenmodied, partiallyleveled, erodedorsimply19 15Ixtacihuatl19 4598 35TocuilaMexico CityHuexotlaPopocatepetl99 20MABCNFig. 1. Top: Landsat TM image of southern portion of the Basin of Mexico showing theTexcoco region (green boxed area) and the location of the San Miguel Tocuila (yellowarea and panel below). The Mexico City metropolitan area is shown in purple colors.Theyellowlineshowstheextensionof theformerLakeTexcocosystem. Theblackcolorareasarethecurrentremnantsof theformerLakeTexcoco. Bottom: satelliteimage of the Tocuila area. Previously described Tlateles are shown by letters A, B, andC. M shows the megafauna locality.Fig. 2. Image of the remains of a ceremonial Tlatel considered part of the Aztec centerin Huexotla, Texcoco. Photograph taken from Parsons report (1971).A. Arciniega-Ceballos et al. / Journal of Archaeological Science 36 (2009) 11991205 1200buriedundermodernurbanstructures. ParsonsmentionedfourTlateles considerably destroyed by plowing and erosion within thetown of Tocuila, fromwhich one was cataloged as ceremonialciviccenterof approximately2.5 mhighand80 80 mlong. Inourinvestigation, following the oral information from elder locals welocated three Tlateles A, B and C shown in Fig. 1. Tlateles B and C arecompleteor partiallycoveredbysmall houses, their preservedcharacteristics of size and height suggest that they were domesticresidential units; therefore, locality A (Fig. 1) may correspond withthe ceremonial Tlatel of Tocuila. This is the area where we carriedout a shallow exploration survey.2. Site description and geophysical surveyTocuila is on a at plain near the Lake Texcoco shore,bordered between two streamlets. The town of Tocuila actually isa looselyconnectednetworkof corncrops andsmall houses(Fig. 1). Duringits urbanizationinthelast decades paleonto-logical deposits of megafauna (location M in Fig. 1) were found ina house backyard (Siebe et al., 1999) and other Mammothslocalities have been reported (Morett-Alatorre et al., 1998;Gonzalez et al., 2006; see Fig. 1). An efcient exploration designshoulduse of several independent geophysical techniques. Ingeophysical prospectiondeningobjectives suchas horizontaldimensions anddepthisthegroundworkfor asurveydesign.Theareaof studyis aat openareaof 115 115 mimmersewithinavolcano-sedimentarygeological environment (Fig. 1).Consideringthecharacteristicsof theareaof studythesurveywas conductedinthreesteps: rst, weappliedthemagneticmethodtoobtainaquickreconnaissanceof theareain21/2dimensions (Fig. 3). This is a low cost method, instrumentation isrelatively simple to use, and is highly recommended for the rstrecognition and evaluation to address further explorations. Basedon the resulting total magnetic eld anomaly pattern, we appliedthe seismic refraction technique along proles P1, P2, P3, P4 andP5(seeFig. 3). Seismicproleswereobtainedperpendiculartothemaximumvariations of themagneticanomaly. Finally, wesurveyedthe same seismic proles using GPRtechniques forfurthercomparisonandreconstructionof thesubsurfacestruc-turegeometry(depthandlayerthickness).The area is located far enough of high voltage power lines thatcouldaffect themeasurementsof themagneticeldorseismictraces. The GPR method was carried out during the dry season toavoid as much as possible water content and thus improvepenetration. Inthefollowingsectionswediscussedthebasisofeachgeophysicalmethod, thedetailsofeachgeophysicalsurvey,the data processing, the analysis and interpretation.2.1. MagnetometryGeomagneticmethodsinshallowexplorationcanbeusedinapplicationssuchasmappingburiedobjectsandpipesandhavebeen considered a primary geophysical technique in archaeologicalprospecting. Magnetometry is an effective, fast and non-destructivemethodfortheinvestigationbasedonthemeasurementof thegeomagnetic eld which is modied by the magnetic properties ofthe surrounding materials. Buried structures or objects withdistinctive magnetic properties to those of the surrounding rocks orsoilsaffectthelocalmagneticeldyieldingamagneticanomaly.Furthermore, human activities may also affect the originalmagneticpropertiesofrocksandsoilsthroughactivitiessuchasburning, harvesting or excavation. These activities modify themagnetic orientation generating variations on the distribution andintensity of the local Earths magnetic eld. These variations in themagnetic eld intensity indicate the presence of distinctive mate-rials or deposits and are key for the analysis and interpretation ofthe geophysical data in archaeological sites.The magnetometry survey was conducted using a Gem-SystemsGSM-19 Overhauser magnetometer; with a 0.015 nT/Hz sensitivityomnidirectional sensor equipped with a GPS to georeference eachmeasurement. The magnetic measurements were taken everysecond thus yielding a sampling rate of roughly one measurementevery meter (Fig. 3). The survey consisted of 990 referenced pointsof total intensity magnetic eld, at 1.0 mof height above the groundsurface.Thedataset wascorrectedbylatitudeanddiurnal variationapplying the IGRF-2005 epoch model. Dst Geomagnetic Indexreported for the date of the survey (March, 17, 2007) is 6.59. Thiscorresponds to a quiet geomagnetic day (Kyoto WDC-GeomagneticEquatorial Dst Indices), avoidinganyundesirableeffects of thenatural eld variations over the acquired information. Data spikesassociated with electronic noise were removed and a compressionof the dynamic range of values was performed. The nal geomag-netic anomaly shown in Fig. 4 was obtained applying Kriging for theinterpolation of the data (Hass and Viallix, 1976); the standard errorestimatedinbothdirections is 7%. InFig. 4 the geomagneticanomaliesrangeapproximatelybetween 160and180 nT, withapositiveeastwesttrendandanaveragevaluebetween0and150 nT. Someof theanomaliesobservedinFig. 4areassociatedwith some exposed steel pipes of the soccer eld facilities. A smallnegative trend was observed westward, running towards thenorthnorthwesternin the west edgeofthestudy area, anditisrelated to the presence of a metallic fence. The area is dominated bya trendof increasing anomaly eldfromwest toeast, whichrepresents a trending rise of depth of a lithologic unit thickness.The magnetic model was generated with the GM-SYS softwarepackage (Northwest Geophysical Associates, Inc, 2004). Thefundamentalsof thecomputational procedurearebasedonTal-wanismethod(Talwani etal., 1959; Talwani andHeirtzler, 1964;Won and Bevis, 1987; Rasmussen and Pedersen, 1979). The inver-sion procedure used is outlined by Webring (1985) and Marqardt(1963). In this data processing Earth is assumed to have relief but nocurvature, considering a two-dimensional at-earth model for themagnetic calculations.2.2. Seismic refraction tomographySeismic refractiontomography(SRT) is aneffectivetool forhorizontal, lateral and vertical characterization of structures, but inP1 P2P3NP4P550 mFig. 3. Conguration of the geophysical surveys arrays in locality A (Fig. 1). Black dotsare magnetic observations; solid lines are the location of the GPR and seismicrefraction proles.A. Arciniega-Ceballos et al. / Journal of Archaeological Science 36 (2009) 11991205 1201archaeology seismic techniques have been considered limited duetoits lowdataacquisitionratecomparedtoother geophysicalmethods. However, in recent years seismic refraction methodolo-gies have been employed to locate buried structures in a variety ofarchaeological conditions (Domenico et al., 2006; Benjumea et al.,2001; Xiaet al., 2002; Leucci et al., 2007; ValentaandDohnal,2007). In our experience, the application of the seismic refractiontomography has yielded successful results due to the high contrastbetween stonewalls and inll layers of the Prehispanic Tlatels.The refraction survey design was based on the total eldmagnetic results shown in Fig. 4. We dened ve seismic refractionproles crossing the geomagnetic anomaly in different directions,three parallel proles (P1, P2andP3) inthe NSdirection, oneprole(P4) along the EW direction and one diagonal prole P5 (Fig. 3).Data were collected using a 48-channel 24 bits GeometricsStratavisorNZseismographand14 Hznatural frequencyverticalOyo-Geospace geophones. Fig. 5 shows the shooting geometry forthe spreads along each seismic line. Each prole consisted of veconsecutive lines of 12 geophones deployedat 2 mintervals, withanoverlapof 2 mbetweenlines. Eachlinehadonelongshot (5 m) andone short shot (1 m) at each end of the spread and one shot in themiddle of the receiver spread, giving a total of 56 measured pointsand25pointsourcesperprole, crossingtheareain118 mlongsections. There was one exception, prole P2 consisted of four linescovering a total length of 96 m. The seismic source used was an 8 Kgsledgehammerverticallyimpactingonan20 Kgground-coupledsteel plate. Five impacts were stacked at each shot point to enhancedata signal-to-noise ratio. Inthe processingandinterpretationof thedata we used theSeisImager Refraction Modeling package whichincludes aninteractive non-linear methodtoperformthe non-lineartravel time tomography in 2D (Oyo Corporation, 2006).SRTis designedtodeneobservedheterogeneities, verticalvelocitygradientsandlateral velocitychanges. Thefundamentalgoal of the interactive tomography method is to nd the minimumtraveltimebetweeneachsource-receiverpairby solvingtheraypathsandtheinversevelocityorslownessusingtheinteractivenon-linear square method (Oyo Corporation, 2006; Redpath, 1973;Gilbert, 1972;Krajewskietal., 1989). Therefractiontomographicmethod is an interactive technique that requires an initial velocitymodel. The method divides the velocity model into quadrangularcells of the same size and constant velocity and traces rays to themodel. A node is the corner of each cell, and along the sides of a cellcan be more nodes. The number of nodes denes the density of therays, andalsodeterminesthedurationof theinversionprocessthus; morenodes, moreraysandinconsequencetheinversionprocess last longer. The travel time between a source and a receiveris determined as the fastest travel time of the ray paths, comparingthe calculatedtravel times tothe measuredtravel times, andmodifying accordingly the model. This iteration is performed untilthe difference between the calculated and observed times isminimized. This process is repeated until the minimum travel timebetween the source and the receiver for each source-receiver pair isfound using an iterative non-linear least-squares approach wherethetraveltimeresidualislessthantheaveragetraveltimepickerror (Oyo Corporation, 2006).Inordertoobtaintheseismictomographicimagingwerstmeasured the travel time of the P-waves rst arrivals for each of the25 shot points along each seismic line (P1P5). In each prole thetraveltimesofeachsourceandreceiverarecombinedtorecon-struct the seismic P-wave velocity distribution on a 2D section. Theinitial velocity model is built up based on the travel time curves ofeach prole considering six horizontal at layers within a velocityFig. 4. Total eld magnetic anomaly of site A in San Miguel Tocuila (Fig. 1).0 11 22 33 44 110Distance(m)point source geophonesFig. 5. Layout of the seismic refraction survey. Consecutive lines of 12 geophones denoted by triangles, spread every 2 m, and ve shot points per line, indicated by stars; givena total of 25 sources and 56 records per prole.A. Arciniega-Ceballos et al. / Journal of Archaeological Science 36 (2009) 11991205 1202rangeof 3002000 m/s, forP-waveandabottomlayerat 10 mdeep. In the initial model we considered a at topography becausethestudyareahasnotsignicantrelief. Theinversionwasper-formed using three to six nodes per cell and 1020 iterations pernode. The same procedure was followed for all proles P1 throughP5.2.3. Ground penetrating radarGPR is a geophysical technique commonly used to imageshallowsubsurface structures based on the propagation andattenuation of electromagnetic energy throughout the subsoil. Itsinvestigation depth is dependent on the dielectrical properties ofthe subsoil, the frequency of the electromagnetic pulses used andthe geometric relationship of strata and other interbedded objects.In general, shallow structures (