IOP PUBLISHING JOURNAL OF GEOPHYSICS AND ENGINEERING

J. Geophys. Eng. 9 (2012) S70–S80 doi:10.1088/1742-2132/9/4/S70

Improving archaeological site analysis: arampart in the middle Orkhon Valleyinvestigated with combined geosciencetechniquesC Grutzner1, J Bemmann2, J Berking3, M Frechen4, R Klinger3,N Klitzsch5, S Linzen6, S Mackens7, M Oczipka8, H Piezonka9,S Reichert2, M Schneider6 and B Schutt4

1 Neotectonics and Natural Hazards Group, RWTH Aachen University, Lochnerstr. 4-20, 52064 Aachen,Germany2 Vor- und Fruhgeschichtliche Archaologie, Rheinische Friedrich-Wilhelms-Universitat Bonn,Regina-Pacis-Weg 7, 53113 Bonn, Germany3 Department of Earth Sciences, Physical Geography, Freie Universitat Berlin, Malteserstr. 74-100,12249 Berlin, Germany4 Leibniz-Institut fur Angewandte Geophysik, Stilleweg 2, 30655 Hannover, Germany5 Institute for Applied Geophysics and Geothermal Energy, RWTH Aachen University, Mathieustr.,52074 Aachen, Germany6 IPHT Jena, Albert-Einstein Str. 9, 07745 Jena, Germany7 Geotomographie GmbH, Am Tonnenberg 18, 56567, Germany8 German Aerospace Center (DLR), Rutherfordstr. 2, 12489 Berlin, Germany9 Historisches Institut, Ernst-Moritz-Arndt-Universitat Greifswald, Hans-Fallada-Str. 1,17487 Greifswald, Germany

E-mail: c.gruetzner@nug.rwth-aachen.de

Received 16 January 2012Accepted for publication 14 May 2012Published 9 August 2012Online at stacks.iop.org/JGE/9/S70

AbstractThe Orkhon Valley in the Central Mongolia was included in the World Heritage list in 2004. Ithosts multiple archaeological sites from Palaeolithic to recent times, which can contribute tothe reconstruction of settlement history in this part of the Eurasian Steppe landscape. Almost100 archaeological sites from prehistoric and historic times including ramparts and khirigsuurswere investigated in five field campaigns from 2008 to 2010 in the middle and upper OrkhonValley. One site, MOR-2 (Dorvolzhin), proved especially difficult to date due to the lack ofsufficient archaeological surface finds, and its role within a manifold of walled enclosuresfrom different times in the study area remained unclear. Therefore, different techniques ofarchaeology, geophysics and geoarchaeology were combined at MOR-2 in order to determinea comprehensive picture about its timing, archaeological meaning, and environmental history.Information on topographical setting and morphometry of the rampart was gathered by anoctocopter equipped with a high-resolution range finder camera. We achieved ahigh-resolution DEM that allowed us to map the rampart in detail and this served as a basemap for all other investigations. SQUID magnetometry, ground-penetrating radar, and electricresistivity measurements (capacitive coupled geoelectrics) were subsequently used to detectarchaeological remains and to characterize the sediment distribution of the inner part of theenclosure and the ramparts themselves. The data show that the construction of the walls issimilar to well-known Uighur neighbouring sites. Man-made sub-surface structures or bigger

1742-2132/12/040070+11$33.00 © 2012 Sinopec Geophysical Research Institute Printed in the UK S70

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

Improving archaeological site analysis

finds could not be detected. Sediment cores were drilled in a nearby meander, covering3000 years BP. The analysis of the strata in terms of elemental composition (P, N, Mn, Fe, etc)revealed an increase of organic content in Medieval times, whereas the allochthonous filling ofthe back water must have started around the beginning of the 6th century AD. Usinggeophysical, archaeological and geological observations, we assume a dating in theTurk/Uighur period (6th–9th century AD) and a re-use under Mongolian reign (12th–17thcentury AD). This would mean that this site is the furthermost walled structure in theperi-urban area of Khar Balgas. However, the specific usage of this walled enclosure remainsunclear and needs further analysis.

Keywords: archaeogeophysics, Mongolia, Orkhon Valley

(Some figures may appear in colour only in the online journal)

1. Introduction

The international project ‘Geo-Archaeology in the Steppe—Reconstruction of Cultural Landscapes in the OrkhonValley, Central Mongolia’ is devoted to the study ofprehistoric and historic settlement patterns at this UNESCOWorld Heritage site. In this BMBF (German Ministryfor Education and Research) funded project, we seekto enhance knowledge of ancient cultures, such as theTurks, Uighurs and Mongols, and their geographic footprintin the centre of Mongolia. The multidisciplinary projectbrings together archaeologists, geographers, geophysicists,geologists, physicists and engineers. One of the aims wasto survey large areas using photogrammetric and geophysicalmeasurements. This would allow for the assessment of not onlyancient settlement areas and their immediate surroundings,but also the wider environs in relation to human impact(Bemmann et al 2011). Almost 100 archaeological sites wereinvestigated within the project, among them walled enclosures,ramparts and khirigsuurs (round or square structures, oftenwith stone frames and a central mound, typical of Mongoliaand mostly dated to the Bronze Age). Most of them weresubject to geophysical, geographical and remote sensinganalysis (Ahrens et al 2008, Oczipka et al 2009, Bemmann et al2011, Klinger et al 2011). Large transects crossing the entireOrkhon Valley were surveyed with geoelectrics and georadar.Apart from geo-archaeological questions, the developmentand adaptation of new technologies for the special workingconditions in the steppe was a central focus for the project.

The site MOR-2 (Dorvolzhin, Durvuljin), an area of 220 m× 170 m enclosed by a rampart of about 1.5 m height, proveddifficult to date due to the lack of sufficient archaeologicalsurface findings. Its role within a manifold of walled enclosuresin the study area remained unclear. This paper presents amethodological approach to cope with such uncertainties. Weused remote sensing, magnetic and geoelectric measurements,ground-penetrating radar (GPR) and shallow drillings togather a comprehensive dataset from MOR-2 in the middleOrkhon Valley (N47◦37’, E102◦34’, figure 1). We showhow the combined application and interpretation of variousgeoscience techniques allows us to characterize the site andsupport archaeological assumptions. Our data prove that thisapproach has the potential to become a standard procedure

in geoarchaeological research, especially in similar studyenvironments.

1.1. Geological background/surface conditions

Mongolia comprises a great variety of geological and tectonicunits. The Ural–Mongolian synclinal belt consists of severalstratigraphic units of Proterozoic to Cambrian age, dividingthe Siberian craton from the Sino–Korean massive. In one ofthe southern ridges, the Khangay mountains have developedand largely consist of Devonian to Mid-Permian sedimentsof marine facies (Sladen and Traynor 2000, Walther andGegeennsuvd 2005).

Since Neogene times, tectonic faulting was accompaniedby intense volcanism generating the vast flood basalts in theupper Orkhon Valley (Walther and Gegeennsuvd 2005). Themiddle Orkhon Valley consists of thick deposits of Quaternaryunconsolidated debris, fluvial pebbles and sand accumulations.Fluvial terraces developed on both sides of the middle OrkhonValley, accompanied by structures that have resulted frommeandering. The terraces show features of fluvial and aeoliansediment deposition with corresponding grain sizes rangingfrom silt up to medium rounded gravel.

The prevailing cold steppe climate is characterized bylow precipitation and low temperatures in winter and viceversa during the summer months. The average total annualprecipitation is around 300 mm year−1. Soils in the regionconsist of regosols in upslope areas and well-developedkastanozems in the lowlands, which sometimes containhigh organic carbon and subsequent calcrete embeddings(Schwanghart et al 2009, 2010). Areas with high groundwatercontent like the Orkhon floodplain are dominated by fluvisols(Haase 1983). However, in the areas where low soil watercontents prevail, the typical shallow vegetation coverage ofthe steppe region is present (Barthel 1983).

1.2. Archaeological background

The middle Orkhon Valley has been a highly favoured areafor human habitation from the Palaeolithic to modern times,as shown by the wide range of findings gathered duringour field work. Between 2008 and 2010, we registered 96archaeological sites in the middle Orkhon Valley. A range

S71

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

C Grutzner et al

Figure 1. The study area MOR-2 in the Central Mongolia is situated on the western bank of the Orkhon river. Green dots representimportant archaeological sites. Geomagnetic measurements have been performed in the entire inner part of the walled enclosure and acrossthe ramparts in the NW and SE, cf figure 2. The marked area has been subject to 2.5D geoelectric measurements. The four black linescrossing the ramparts in the NW of MOR-2 indicate the GPR and OhmMapper profiles, cf figures 2 and 5.

of different structures were identified at these sites with themost common being walled enclosures. One of these walledenclosures is MOR-2, which is listed in Perlee’s catalogueof Mongolian settlements from antiquity to the Middle Ages(Perlee 1961). The site lies on the edge of the western terraceabove the Orkhon river, about 15 km to the NNW of the ancientUighur capital Khar Balgas (figure 1). Above ground it isvisible as a large rectangular rampart measuring approximately220 m × 170 m. The enclosure is orientated WNW to ESEalong its long axis with an entrance in the middle of theeastern bank towards the river. The banks rise up to 4 min height. Although the site had been repeatedly revisited

by the archaeological survey, there was a substantial lack ofdiagnostic finds to allow the dating of the site.

2. Remote sensing

Most of the archaeological sites in the study area arehard to recognize in the field and appear as intenselydegraded ramparts and walled enclosures. The majorityof these structures have a rectangular outline and yieldrampart heights ranging from a few centimetres up to severalmetres. In many cases, adobe has been used for the wallconstruction. Therefore, during the hundreds of years withoutmaintenance, the walls turned into hill-shaped rises. In addition

S72

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

Improving archaeological site analysis

(A) (B)

(C)

Figure 2. (A) Aerial photo (stereoscopic image) of MOR-2 collected with the octocopter. Black lines indicate the position of the 2Dgeoelectric and georadar data crossing the ramparts. The example dataset presented in figure 5 is marked. The rectangle shows the positionof inlet B. (B) Detailed view of the position of the archaeological trench that cut the rampart, including contour lines. Trench data arepresented in figure 6. (C) Photograph of MOR-2 from its western edge to the north. Note the tracks caused by the OhmMapper.

to these enclosures, several grave fields and an ancient quarrywere recorded.

With aerial imagery, it is possible to detect additionalstructures by slight colour changes in overview photos. Theimages are recorded at an altitude of up to 400 m above groundand vertical exaggerations of low features can be observedwhen viewed in 3D. Using such techniques, remote sensinghas become increasingly important for archaeological researchin recent years (Wiseman and El-Baz 2007).

The Leica M8, a 10 mega-pixel range finder camerawith a Voigtlander 21 mm/F4 Color Skopar lens was used tosurvey the archaeological sites and to detect formerly unknownstructures. The camera was installed on a stabilized mountcarried by an octocopter, an electrically powered multicopte(figure 3A). From the image data, which were recorded withan overlap of 85% along and across the track, digital elevationmodels, orthophoto-mosaics and 3D models were derived. Inaddition to that, the surroundings of several of the sites wererecorded with oblique photos and stitched to panoramic views.

By using remote sensing and photogrammetrictechniques, impressing models can be created, and

furthermore, archaeological and GIS documentation analysiscan be executed. We used the octocopter data not only as abase map (figures 1 and 2), but also for topographic correctionof the geophysical data (figure 5). We achieved a geometricresolution of 2–5 cm in X, Y and Z at 80 m flight altitude and anaccuracy of less than 1 cm (Oczipka et al 2009, Bemmann et al2011). The mean standard deviation of terrain points (X, Y, Z)was (0.015, 0.015, 0.024) m; the maximum standard deviationof terrain points was (0.066, 0.060, 0.113) m.

3. Geophysical methods

The octocopter images were used to create a DEM ofMOR-2 and they allowed us to precisely map thearchaeological remains with surface expression. For imagingsmall and/or buried objects, we applied geophysicaltechniques. Since most artefacts known from the OrkhonValley have different magnetic properties than the surroundingsediments, magnetic measurements were applied forprospection. Electrical resistivity and georadar were used toretrieve depth information of larger objects and structures

S73

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

C Grutzner et al

(A) (B)

Figure 3. (A) The octocopter was used to gather high-resolution imagery and terrain models by applying photogrammetric techniques.(B) Magnetic prospection with the SQUID gradiometer system in the middle Orkhon Valley in autumn 2010. The SQUID gradiometers arelocated inside three tubular helium cryostats mounted on the left end of the cart. The lateral distance from the sensor position to the back ofthe off-road car is 6 m. The data acquisition unit, as well as the attitude and position measuring unit, is mounted at about one-third of thedistance (white cover box).

(Arlsan et al 1999, Neal 2004, Samouelian et al 2005, Porebaet al 2007). Imaging the internal structure of the rampartswill increase knowledge of the building style, and also allowMOR-2 to be compared to adjacent sites in the study area.We also expected to gain information about the geologicalbackground, which should help us to characterize the entirearchaeological site.

3.1. Magnetics: the SQUID gradiometer system

To enable precise magnetic prospection of large areas in theorder of tens of hectares within passable time, a motorizedmeasurement system was developed at the Institute ofPhotonic Technology (IPHT), Jena, Germany (Linzen et al2009). The instrument is based on superconducting quantuminterference device (SQUID) gradiometer sensors, whichprovide the highest magnetic field gradient resolution, evenwhen used at high speeds over the ground. Currently, we reachup to 60 km h−1 by the use of a metal-free cart carrying theSQUID gradiometers and an off-road car as a pulling vehicle(figure 3B). This technique allows magnetic prospection,progressing at 1–3 hectares h−1, to be performed. Seventeensites with a total area of more than 300 hectares correspondingto 2000 driven line kilometres were prospected within 32measurement days. After analysing strengths and weaknessesthat became clear during the first campaign, 2009’s sensorconfiguration of three SQUID gradiometers was improvedin 2010 with a setup of 12 gradiometers. The most recentsystem allows the instantaneous recording of several tensorcomponents of the magnetic gradient in comparison to onlyone component with the 2009 setup. In 2010, a measurementpoint spacing of 2–10 cm and a spacing of 50 cm between themeasured lines were achieved. However, two datasets recordedwith a partially damaged system at the end of the 2009 missionhave exceptionally lower lateral resolution as the line spacingis doubled (100 cm). One of these sets was recorded by thesystem at MOR-2 over 3 h. The corresponding magnetogramis shown in figure 4.

Magnetic signatures indicating building structures cannotbe found within the inner area of MOR-2. However, a

pronounced grainy magnetic area caused by a stony surfacelayer of 0.5–1 m thickness is visible in the magnetogram insidethe enclosure in the NW corner. This area is about 65 m indiameter and has the lowest altitude. Similar signals can befound directly on top of the ramparts, especially on the NWand SE ones. In contrast, the downslope part of the rampart ismagnetically smooth in many places and without any structure.Here, homogeneous colluvial layers are accumulated, whereaserosion dominates the rampart’s top. The erosion has revealedwhat is most likely to be the ramparts core with its magneticallydetectable granularity. Sharp, horizontally oriented magneticcontrasts in the upper part of the slopes were detected on otherramparts in the vicinity, but not at MOR-2. This fact indicatesa high level of distortion of the original wall geometry, whichis in good agreement with archaeological test trench data(section 4), as well as the subsequently described geoelectricresults.

3.2. Geoelectrics: OhmMapper

The geoelectrical measurements were carried out with theOhmMapper (Geometrics Inc., San Jose, CA, USA). Thisdevice transmits the electrical field by capacitive (CR) insteadof traditional galvanic coupling (Timofeev 1974) via dipoleline electrodes (Ziekur and Grelle 2009, Neukum et al 2010,Vadillo et al 2012). The benefits are a higher data collectionrate compared to traditional galvanic-coupled geoelectricmeasurements (DC) and the possibility of gauging sealedor frozen surfaces with high resistivities. The method doesnot allow detection of small objects but is able to detectlarger structures (e.g., walls or wells). Our system operatesat 16 kHz using a dipole transmitter and up to five receivers.The investigation depth depends on the distance betweenthe transmitter and the receivers (through experience themaximum investigation depth is about 20 m). A practicalresolution limit is set by the fact that only five data pointscan be registered in the entire depth range. For archaeologicalpurposes, a smaller offset needs to be chosen, thus reducing thepenetration depth to around 4 m. Data inversion was carriedout with RES2DINV (2D data) for most of the investigation

S74

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

Improving archaeological site analysis

Figure 4. SQUID magnetogram of MOR-2, showing the tensor component dBx/dz with x in the north-east direction and z in the downdirection. UTM coordinates and contours were calculated from differential GPS data, which are recorded by the SQUID system. Contourspacing is 1 m. Two areas are characterized by an accumulation of magnetic anomalies that point to archaeological remains: an oval-shapedpart in the northern part of the enclosure (white ellipse) and the top of the NW and SE walls (black arrows). Black lines indicate thelow-resistivity anomaly detected with geoelectric measurements. Linear structures of the magnetogram outside the western part of theenclosure as well as inside in the north and south-west reveal currently used trails with slightly compacted topsoil. The positive gradientregion on the western sidewall of the rampart in the south is most likely a measurement artefact caused by the strong tilting of the cart.

sites. At MOR-2, we carried out the 2D inversion of theprofiles that cross the rampart, but no 3D inversion of theparallel 2D lines recorded for 2.5D data interpretation wascarried out (figure 1). We took into account the specialrequirements for line electrodes (Neukirch and Klitzsch 2010)and used a virtual point-electrode dipole length of 80% fordata inversion to address these special requirements. Nearbydrillings and former measurements in the Orkhon Valleyrevealed more or less horizontal layering in the undisturbedsubsurface. This is typical for areas of the Orkhon Valleydominated by alluvial and floodplain deposits. Here, sedimentsranging in grain size between silt and medium gravel wereencountered (drillings are discussed in detail in section 5).Clayey sediments would have yielded a high conductivity (thesame as for saturated deposits). However, due to permafrostoccurrence, in most places, the pore fluid was frozen thatcaused far lower sediment conductivity.

Five receivers with dipole cables of varying lengthshave been used for the investigations at MOR-2. We testedthree different dipole cables in order to determine the bestconfiguration. A penetration depth of approximately 5 m couldbe reached at MOR-2 with the 2.5 m cables. The north-western part of MOR-2 (one-third of the entire inner area)

has been imaged to reveal subsurface structures (figure 1).Using the parallel profiles with 2 m spacing, we then mappedthe resistivity distribution in the NW part of the enclosure.

We found no indications of any archaeological remainsinside MOR-2. The underlying geology, however, ischaracterized by a curved low-resistivity pattern (resistivityvalues not exceeding 300 � m compared to a backgroundresistivity of at least 1000 � m). Since the structure is visiblefor tens of metres, has a diameter of several tens of metresand taking into account that we do not expect archaeologicalremains at depths of more than 1.5 m, the pattern indicatesthe presence of a palaeo-river stream now filled with silty orclayey sediments. Stones were also present on the surface inthat area. Interestingly, the low-resistivity structure more orless follows the outlines of the magnetic anomaly, which ismarked by a white ellipse in figure 4.

In addition, we crossed the north-western ramparts withfour profiles in order to determine the internal structure ofthe walls (map in figures 1 and 2). We found low resistivitiesof <100 � m in the ramparts and below 50 � m in somelocations. This indicates that clay (adobe) may have been usedas a building material instead of mud bricks (figure 5).

S75

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

C Grutzner et al

210

-1-2-3

NW Distance [m] SE0 10 20 30 40 50 60 70 80 90

]m[ htpe

D

30 60 120 240 500 1000 2000 4000Resistivity [Ωm]

outer part inner part

outer part rim inner part

Model resistivity with topography

highattenuation

hyperbolae

210

-1-2-3-4-5-6

]m[ htpe

D

Figure 5. Geophysical data from the ramparts. Both electric and GPR data are from the westernmost profile that crosses the rampart; seefigures 1 and 2 for the location. Geoelectric measurements (OhmMapper data) in the upper image reveal a low resistivity anomaly inside theramparts. Note the logarithmic colour scale. Values do not exceed 100 � m, which is typical for adobe. Five iterations were used for datainversion, the resulting absolute error was 8.2%. The GPR data (lower image) also reveal high conductivities in the ramparts, herecharacterized by a high attenuation of the radar waves. Several hyperbolae occur in the inner part of MOR-2.

3.3. Ground-penetrating radar

We used a 270 MHz antenna with the SIR-3000 datacollection unit from GSSI (Salem, USA), a survey wheeland a hand-held GPS. This equipment allowed a penetrationdepth of up to 4 m and a spatial resolution of approximately7 cm, depending on the underground conditions (Bereset al 1995, Neal 2004). All data processing was performedwith ReflexW by Sandmeier Software, Karlsruhe, Germany.The main processing steps applied were move starttime, background removal, gain adjustment, averaging andtopographic correction. No migration was computed. Fromhyperbola analyses, we derived wave velocities of 0.15 m ns−1

(εr = 4), which were used for time–depth conversion. Fourprofiles crossing the north-western rampart were collectedalong the geoelectrics tracks. The high attenuation of the radarwaves implies high conductive materials inside the ramparts,most probably adobe (figure 5), confirming the results ofthe OhmMapper survey. The construction scheme is knownfrom neighbouring Uighur sites (Jaquin et al 2008). Althoughthis alone does not prove an Uighur dating of MOR-2, thegeophysics revealed that the ramparts are constructed in asimilar manner. Other constructions in the area were madeup of fired bricks, which are characterized by lower electricalconductivities. This would have led to a deeper penetration ofthe radar waves and to higher resistivities in the geoelectricsurvey.

Diffraction hyperbolae found in the inner part of theenclosure are a hint of small-scale structures in the subsurface,such as walls, boulders or even burrows. The signals are,however, not very strong and it is unlikely that they are due

to larger archaeological remains, such as walls or foundations.Burrows were not visible in the field and the occurrence oflarger boulders in that area is unlikely. Therefore, the origin ofthe diffraction hyperbolae remains unclear.

4. Archaeological trenching

Based on the results of the geophysical surveys, a test trenchwas dug into the rampart close to its western corner in autumn2010. We aimed to verify the geophysics results, to examinethe construction style of the rampart and to extract an opticallystimulated luminescence (OSL) sample (figure 6). Subsoil wasreached at a depth of 2.3 m. No soil samples were taken for thegeotechnical analysis or for the determination of geophysicalproperties. The cut shows the remains of an impressive earthenrampart. It consists of horizontal bands mixed from white tolight yellow, compact silty loam partly containing fine gravel,and soft middle brown silt lenses. The borders between thebands are very distinct. This indicates that the rampart waserected from layers of stamped earth. The structure of thelayers shows some discontinuities. If these discontinuitiesare interpreted as collapses, the original ramparts wouldhave been substantially higher. Large amounts of animalbones were found, especially within animal burrows. However,the trench yielded no datable archaeological artefacts. Thedating of this enclosure must therefore rest on the OSL datingmethod for which a sample was taken from the lower partof the undisturbed rampart, approximately 1.75 m below thesurface. As yet, there are no results from the OSL sample;therefore, the precise dating of the site itself remains unclear.

S76

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

Improving archaeological site analysis

Figure 6. Trench log of the archaeological excavation conducted in autumn 2010; see figure 2 for location. Clear-cut contacts between thesingle layers indicate a construction with layers of stamped earth; distorted layers point to a collapse of the rampart, and hence an originallyhigher construction. OSL sample datings are under preparation.

Table 1. Dating results of RK10

Lab-n Technique Material Depth 14C Age 2-sigma Cal age 2-sigma

Hv-25925 C14 bulk/soil 19 cm 340 115 1606 AD 138Hv-25926 C14 bulk/soil 48 cm 540 70 1372 AD 52Hv-25927 C14 bulk/soil 155 cm 3260 125 1560 BC 137POZ-28504 C14-ams wood 183 cm 2925 35 1133 BC 65

Judging from the type of construction, the layout and theorientation of the complex, we suppose a construction duringthe 8th/9th century, i.e. the time of the Uighur Empire.

5. Drillings

In addition to the non-invasive measurements, near-surfacedrillings provide an insight into the sediment setup. Therefore,five sediment cores inside the rampart and three from nearbymeanders were taken and described. Cores were taken witha percussion driller to obtain undisturbed sediment samples.The drilling locations are marked in figure 1. Because of alack of datable material (C14 dating), only one core gavesufficient data regarding datable material and structure. Radio-carbon dating was undertaken using four samples from coreRK10 (table 1). Samples were treated with standard methodsfor conventional dating and calibrated using the CalPal Huludataset (Danzeglocke et al 2011).

Cores from the centre of MOR-2 compliment theresults of GPR and geoelectrics. No brick material or otheranthropogenic objects were found, and significant stratificationwas missing. The drilling core RK10 is located in a nearbymeander on the northern edge of the rampart. The lithologyshows a threefold setup (figure 7): sands and silt in the

upper part of the core with brownish colours; finer materialin the part below 78 cm characterized by greyish colours;and fluvial deposits ranging from large pebbles to coarsesand in the lowest section below 180 cm. Inductivelycoupled plasma–optical emission spectrometry measurementsof element concentrations in 10 cm intervals (standard treatedsamples) show major variances in the three layers. Themiddle and lowest parts show no significant concentrationsin measured elements except an increased concentration ofiron and Fe/Mn ratio compared to the upper layer. The upperlayer is characterized by a strong increase of concentrationsregarding organic and inorganic carbon, sulfur, strontium andhigh Sr/Mg ratios at approximately 50 cm below the surface.

Samples for dating purpose were taken at depths of 19,48, 155 and 183 cm below the surface level. Since 1000 calBC, the location has been a backwater body as indicated bygrain size, colour and weak soil formation process indicatorsas a result of meandering processes (Stølum 1996, Hartmannand Wunnemann 2009). The samples show an onset of thethird phase of sedimentation prior to the 12th century: thefilling with coarser material and soil formation started and,consequently, inorganic/organic carbon content increased.Peaks of phosphate concentrations indicate intensified humanusage around the drilling site in modern times and at about

S77

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

C Grutzner et al

Figure 7. Elemental concentrations in RK10 and dating results.

1200 cal AD. This goes along with maximized erosion aroundthe site at 1350 cal AD, as indicated by mineral peaks. Usingthe small number of datings, we can estimate the onset of soilformation (beginning of the upper layer) using a regressionanalysis and assuming a constant sedimentation process. Thecorresponding date of this onset at 78 cm is given by 537 ±279 cal AD, which falls into the time period of the Turk/Uighurreign. The statistical result shows a beginning at about 537 ±279 cal AD, which also falls into the Turk/Uighur reign. Thecombination of all data suggest a filling during constructionof MOR-2 in Uighur times and a re-use of the site during theMongol dominion (13th century). The significant variationsin the mineral content strongly point to an anthropogenicinfluence on the sediments, especially since these effects werenot observed in other parts of the Orkhon Valley, where thereis a significant distance to archaeological sites.

6. Discussion

The lack of archaeological information at the beginning ofour study made it hard to determine the age and purpose ofthe spectacular structure MOR-2. Octocopter imagery did notreveal new archaeological remains but was used to provide aDEM for topographic corrections, which served as the mapbase. From other sites, it is known that this technique canadd valuable archaeological information (Oczipka et al 2009)due to its high precision and accuracy. Large areas can besurveyed relatively fast but post-processing is time consuming,the machinery is expensive and the octocopter cannot take offduring bad weather.

Although magnetic signatures indicating buildingstructures were not found inside MOR-2, the SQUIDmeasurements provided information about the distortion status

S78

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

Improving archaeological site analysis

of the ramparts’ core construction. Anomalies concentratedon the inner NW part of the enclosure were detected.Here, interpretation is critical and based on the non-randomdistribution of magnetic anomalies corresponding to small-sized stones. The anomalies observed do not necessarilypoint to archaeological objects, but could also be due toiron-bearing stones. Ore deposits can be found upstreamin the upper Orkhon Valley, which could serve as thesource area. Furthermore, clarification could be given byadditional archaeological trenches at the questioned parts ofthe enclosure. The SQUID technology allows the mapping ofhuge areas in short time (up to several thousands of squaremetres per day) with a previously unreachable precision. Eventhe smallest magnetic anomalies can be detected. However,the SQUID sensors have to be cooled with liquid heliumfor functionality. The instantaneous recording of severaltensor components of the magnetic gradient measured by thelatest setup of the SQUID system will further increase theamount of geomagnetic information that can be obtained andconsequently the relevance of this technique in archaeometry.

Geoelectric measurements revealed a meander structureinside MOR-2 but no foundations, hidden walls or similarfeatures could be observed. However, the geologicalinformation was useful in reconstructing landscape evolution,deciding on the drilling locations and evaluating the generallayout of the site. Data from the ramparts revealed a low-resistivity interior (figure 5). This lead us to conclude thatadobe was the building material rather than fired bricks.Later, archaeological trenching proved this interpretation asvalid. The OhmMapper proved to be especially useful forlong distance profiles, large datasets and frozen (permafrost)surfaces. Its main disadvantage is the lower data densitycompared to classical geoelectrics.

Georadar data confirmed the OhmMapper findings. Wewere able to determine that the ramparts had an adobe core(figure 5). By comparing the ramparts to similar constructionsat other sites where the dating is clear, we got the first indicationof an Uighur timing of the construction of MOR-2 (Jaquin et al2008). Both GPR and geoelectrics revealed a highly conductiveinterior of the ramparts. Metal-bearing material or high watercontent could also cause this phenomenon, but the occurrenceof metal in this geological setting is highly unlikely, and largerobjects would lead to diffraction hyperbolae. An increasedwater content compared to the surroundings could only be dueto the use of silty-clayey building material, which also supportsour interpretation. We show that GPR can successfully beapplied for archaeological purposes in the steppe regions ofCentral Asia. The occurrence of low-resistivity soils (clayeyand wet material) limits the achievable penetration depth, butdespite this GPR is one of the most powerful non-destructivetechniques for geoarchaeological studies.

An archaeological excavation based on the informationof the geophysical survey strengthened the assumption of anUighur construction. The archaeological trench proved thegeophysical data to be helpful for site selection. The featuresobserved in the excavation are in very good agreement with thegeophysical data. Silt as building material accounts for the lowresistivities observed in the geoelectric data and for the high

attenuation of the electromagnetic waves (figure 5). In the GPRdata, thin layers in the ramparts are visible as well as a distortedlayering. The trenching proved that stamped earth was used infine layers and that the ramparts had partially collapsed.

Shallow drillings allowed an insight into the sedimentstructure of the study site and allowed indirect dating(figure 7). Other means of sediment mobilization that wouldlead to the observed variation in the mineral content canbe imagined, but are much less likely compared to ananthropogenic influence, especially taking into account theobvious existence of artificial structures. The possible errorwhen using regression to determine the onset of the layerchange is relatively high, but the results also support ourinterpretation. Since this technique is penetrative, it cannotbe applied to all interesting sites because of the protectionmeasures in cultural heritage areas.

With the help of the aforementioned techniques, we wereable to characterize MOR-2. We found similarities to well-known sites, which gave the first hints for dating the actualstudy area. C14 dating and sediment sample analysis allowedus to narrow down construction and re-use dates; a first fillingduring the Turk/Uighur period and a later re-use in Mongoliantimes (13th century). This would mean that MOR-2 waspart of the peri-urban area of Khar Balgas. It would be thefurthermost walled structure belonging to this ancient capitalwith a distance of around 15 km to the city centre. The re-useduring the Mongol reign is also an interesting outcome as anew capital was built in that time far away from MOR-2. Yet,the specific usage of this walled enclosure remains unclear andneeds further analysis by archaeological means.

The new methodology approach included remote sensing,archaeological, geophysical and geological field methods tocomprehensively investigate a walled enclosure in a CentralAsian steppe setting. We showed that the combination of thesedifferent techniques is able to improve site understanding,especially where classical archaeological research reaches itslimits. The combination of archaeological, geoscientific andengineering techniques allows us to compensate disadvantagesfrom single methods and to verify data. It can be used as astandard procedure for geoarchaeological projects in CentralAsia.

Acknowledgments

We thank the Bundesministerium fur Bildung und Forschung(BMBF) for funding this project. Many students fromGermany and Mongolia did a great job during field work.We thank Thomas Wiatr, who helped with GIS. We also thankJack Mason for his help with English.

References

Ahrens B, Bemmann J, Klinger R, Lehmann F, Munkhbayar L,Oczipka M, Piezonka H and Schutt B 2008 ArcheologijnSudlal (Mongolian Academy of Sciences vol 26) pp 311–27

Arlsan E, Biella G, Boniolo G, Caporusso D, de Franco R, Lozej Aand Veronese L 1999 J. Appl. Geophys. 41 169–88

Barthel H 1983 Physisch-Geographische Studien inAsien/Fyzikogeograficke Studie v Asii (Ceskoslovenska

S79

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022

C Grutzner et al

Akademie ved Geographicky Ustav vol 34) ed H Barthel,H Brunner and G Haase (Brno: Ceskoslovenska Akademie vedGeographicky Ustav) pp 3–91

Bemmann J et al 2011 Studia Archaeologica Instituti ArchaeologiciAcademiae Scientiarium Mongolicae (Mongolian Academy ofSciences vol 30) (Ulaanbaata: Mongolian Academy ofSciences) pp 1–14

Beres M, Green A G, Huggenberger P and Horstmeyer H 1995Geology 23 1087–90

Danzeglocke U, Joris O and Weninger B 2011 Calpal-2007software, http://www.calpal-online.de

Haase G 1983 Physisch-Geographische Studien inAsien/Fyzikogeograficke Studie v Asii (CeskoslovenskaAkademie ved Geographicky Ustavvol vol 34) ed H Barthel,H Brunner and G Haase (Brno: Ceskoslovenska Akademie vedGeographicky Ustav) pp 231–367

Hartmann K and Wunnemann B 2009 Quat. Int. 197 28–44Jaquin P A, Augarde C and Gerrard C M 2008 Int. J. Archit.

Heritage: Conservation Anal. Restor. 2 377–400Klinger R, Schwanghardt W and Schutt B 2011 Die Erde 142 41–64Linzen S, Schultze V, Chwala A, Schuler T, Schulz M, Stolz R

and Meyer H G 2009 New Technologies for Archaeology(Natural Science in Archaeology) ed M Reindel andG A Wagner (Berlin: Springer) pp 71–85

Neal A 2004 Earth Sci. Rev. 66 261–330Neukirch M and Klitzsch N 2010 Vadose Zone J. Quat. Sci.

9 882–92Neukum C, Grutzner C, Azzam R and Reicherter K 2010 Advances

in Research in Karst Media ed B Andreo, F Carrasco, J J Duranand J W LaMoreaux (Berlin: Springer) pp 429–34

Oczipka M, Bemmann J, Piezonka H, Munkhbayar J, Ahrens B,Achtelik M and Lehmann F 2009 Proc. SPIE7478 747806

Perlee K 1961 Mongol ard ulsyn ert dundad ujeijn khot suurinytovcoon Catalogue 152, number 207, Ulaanbaatar, Mongolia

Poreba A, Zuberek W M, Nogaj-Chachaj J, Kotyrba A and Siwek S2007 Acta Geophys. 55 640–51

Samouelian A, Cousin I, Tabbagh A, Bruand A and Richard G 2005Soil Tillage Res. 83 173–93

Schwanghart W, Frechen M, Kuhn N J and Schutt B 2009Palaeogeogr. Palaeoclimatol. Palaeoecol. 279 160–71

Schwanghart W, Schutt B and Moller S 2010 Mongolian–GermanQara Qorum-Expedition (Forschungen zur ArchaologieAußereuropaischer Kulturen vol 1) ed J Bemmann,H G Huttel and E Pohl (Wiesbaden: FAAK) pp 14–26

Sladen C and Traynor J J 2000 Lakes during the evolution ofMongolia AAPG Stud. Geol. 6 35–57

Stølum H H 1996 Nature 271 1710–3Timofeev V M 1974 The employment of capacitively coupled

sensors in engineering and geological studies PhD ThesisUniversity of Moscow, Russia

Vadillo I, Benavente J, Neukum C, Grutzner C, Carrasco F,Azzam R, Linan C and Reicherter K 2012 J. Appl. Geophys.82 153–62

Walther M and Gegeennsuvd T 2005 Geogr. Oekol. J. MOLARERes. Centre 2 1–36

Wiseman J R and El-Baz F (ed) 2007 Remote Sensing inArchaeology (Interdisciplinary Contributions to Archaeology)(Berlin: Springer)

Ziekur R and Grelle T 2009 Z. Dtsch. Ges. Geowiss. 4 391–402

S80

Dow

nloaded from https://academ

ic.oup.com/jge/article/9/4/S70/5127632 by guest on 27 July 2022