Combining GeoEye-1 Satellite Remote Sensing, UAV Aerial Imaging, and Geophysical Surveys in Anomaly...

870 IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, VOL. 4, NO. 4, DECEMBER 2011

Combining GeoEye-1 Satellite Remote Sensing,UAV Aerial Imaging, and Geophysical Surveys inAnomaly Detection Applied to Archaeology

Albert Yu-Min Lin, Alexandre Novo, Shay Har-Noy, Nathan D. Ricklin, and Kostas Stamatiou, Member, IEEE

Abstract—This paper describes a method of combinedultra-high resolution satellite imaging, unmanned aerial vehicle(UAV) photography, and sub-surface geophysical investigation foranomaly detection, which was employed in a non-invasive surveyof three archaeological sites in Northern Mongolia. The surveyedsites were a Bronze Age burial mound, a Turkish period tomb,and a steppe city fortification of unknown origin. For the satellitesurvey, 50 cm resolution pan-sharpened imagery was generatedthrough a combination of multispectral and panchromatic data,collected from the GeoEye-1 earth-sensing satellite. The imagerywas then used to identify the location of the aforementioned sitesin an approximate area of 3000 km . Aerial photographs of thesites were obtained with two customized electric-powered UAVs:a fixed flying wing rear-propulsion plane and a multi-propeller“oktokopter” helicopter system. Finally, geophysical investigationwas conducted with a GSM-19 Overhouser gradiometer, an EM38electromagnetometer, and an IDS Detector Duo ground pene-trating radar. The satellite imagery and aerial photographs werecombined with the geophysical survey results and on-site surfaceobservations to provide insight and contextual information aboutanomalies in multiple layers of data. The results highlight theeffectiveness and robustness of the employed method for archaeo-logical investigation in large, rugged and scarcely populated areas.

Index Terms—Aerial, anomaly detection, archaeology, electro-magnetic survey, GeoEye-1, geophysics, ground penetrating radar,magnetic survey, remote sensing, satellite, UAV.

I. INTRODUCTION

A ERIAL images have long been utilized in archaeologicalresearch to provide a perspective that accentuates ground

features, not otherwise apparent, and a greater understandingof their spatial context [1]–[4]. Buried features can producesmall changes in surface conditions such as slight differences

Manuscript received December 06, 2010; revised March 04, 2011; acceptedMarch 22, 2011. Date of publication June 13, 2011; date of current version De-cember 14, 2011. This work was supported by the National Geographic Society,the Waitt Institute for Discovery, the GeoEye Foundation, and the National Sci-ence Foundation under IGERT Award DGE-0966375.A. Lin is with the California Institute for Telecommunications and Informa-

tion Technology, University of California at San Diego, La Jolla, CA 92093USA (e-mail: a5lin@ucsd.edu).A. Novo is with Geostudi Astier, Livorno 57121, Italy (e-mail:

novo@geoastier.it).S. Har-Noy and N. Ricklin were with the University of California at San

Diego, and are now with Tomnod Inc., San Diego, CA 92126 USA.K. Stamatiou was with the University of California at San Diego, and is now

with DEI, University of Padova, 35122 Padova, Italy.Color versions of one or more of the figures in this paper are available online

at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/JSTARS.2011.2143696

in ground level, soil density and water retention, which in turninduce vegetation patterns (cropmarks), create variability in soilcolor (soilmarks) or even shadows (shadowmarks) that can beseen from above [5]. The overhead perspective can also providevaluable insight on the results of geophysical surveys, throughthe referencing of surface anomalies to generated subsurfacemaps [6]–[9].The introduction of earth-sensing satellites has further con-

tributed to the integration of remote sensing in archaeology[10], [11]. The ability of detecting features on the ground fromspace is largely dependent upon the ratio of feature size todata resolution. As sensor technologies have improved, thepotential to utilize satellite imagery for landscape surveys hasalso improved [12]–[14]. In September 2008, the GeoEye-1ultra-high resolution earth observation satellite was launchedby GeoEye Inc., to generate the world’s highest resolution com-mercial earth imaging (at the time of launch) [15]. Generating41 cm panchromatic and 1.65 m multispectral data, this sensorfurther expanded the potential of satellite based archaeologicallandscape surveys. However, government regulations restrictpublic use of satellite data exceeding 50 cm resolution.Earth-based aerial imaging technologies have also seen sig-

nificant improvement in sensor and platform technology. Earlytechniques included balloon and kite photography, which havebeen significantly enhanced over the years to enable high-reso-lution data collection [16]–[18]. However, these techniques gen-erally require a tether to the ground and either helium gas or suf-ficient wind conditions to enable flight, thus limiting the areasand conditions where they can be deployed and their mobilityin the field. Recently, radio-controlled aircraft have been em-ployed in archaeology for the purpose of obtaining aerial im-ages [19], [20], due to their advantages of mobility and largedeployment range.Geophysical methods, such as magnetic [21], [22], electro-

magnetic (EM) [23]–[25], and ground penetrating radar (GPR)[26] prospection, have been used to detect andmap undergroundfeatures in a fast and non-invasive manner [27]–[31]. Each ofthese methods exploits different physical properties of the ter-rain. Magnetic survey is a passive detection of contrasts in themagnetic properties of differing materials, whereas EM surveysmeasure the conductivity and magnetic susceptibility of soilby inducing eddy currents through a generated electromagneticfield. GPR transmits an electromagnetic pulse and measures areflected signal that is dependent upon the dielectric propertiesof the subsurface material [32]. Through GPR, it is possible

1939-1404/$26.00 © 2011 IEEE

LIN et al.: COMBINING GEOEYE-1 SATELLITE REMOTE SENSING, UAV AERIAL IMAGING, AND GEOPHYSICAL SURVEYS IN ANOMALY DETECTION 871

to reconstruct high-resolution 3D data visualizations of sub-surface anomalies, whereas EM and magnetic surveys provide2D ground maps. Subtle features in geophysical survey mapscan indicate archaeological features or objects otherwise unde-tectable on the surface.In this paper, we describe a multi-stage method which was

employed in the archaeological investigation of three sites innorthern Mongolia, namely, a suspected Bronze Age burialmound, a Turkish period tomb and a steppe city fortificationof unknown origin. Data was collected via high-resolutionsatellite imaging, unmanned aerial vehicle (UAV) photography,and sub-surface geophysical measurements. The satellite, aerialand geophysical data were then overlayed, in order to providecontextual information for features and anomalies observed inall three layers of data. The results of the study indicate theeffectiveness and robustness of the employed method in thearchaeological investigation of large and remote areas.The rest of the paper is organized as follows. In Section II, we

describe in detail the three-layer method and the tools involved,and in Section III, we present the results of the archaeologicalstudy. Finally, Section IV summarizes our main conclusions.

II. METHOD

A. General

The area of investigation was roughly 3000 km in size,scarcely populated and rugged, presenting several challengesfor travel, as sites were often only accessible on foot or onhorseback. The survey proceeded in three stages: Firstly,GeoEye-1 satellite imagery, in combination with local nomadicknowledge, was used to locate and identify the sites of interest.Subsequent on-site investigation involved the deployment oflow-cost UAV imaging platforms, in order to identify subtleterrain features, often not visible from the ground. Two electricpowered UAVs were customized for this research: a fixed flyingwing rear propulsion plane and a multi-propeller “oktokopter”helicopter system. The aircraft were radio-controlled andcollected geo-referenced images in both the visible and near-in-frared wavelengths. Finally, magnetic, EM and GPR surveyswere performed in order to obtain subsurface measurements. Inthe following, we describe each of the stages of investigationin detail.

B. Satellite Imaging

Satellite imagery from the GeoEye-1 sensor was collectedduring the summer season and used to survey an area approxi-mately 3000 km in size in NorthernMongolia, which consistedmostly of grassy steppe with limited tree cover. The GeoEye-1provides 50 cm pan-chromatic and 1.65 m multispectral im-agery in 15.2 km (9.4 mi) swaths. Using a custom pan-sharp-ening algorithm based on the YCbCr colorspace transforma-tion, the structure information from the pan-chromatic data wascombined with the red, green and blue bands of the multispec-tral data resulting in 50 cm pan-sharpened color imagery [33].Due to the massive scale of the imagery data, a visualizationtool known as the Highly Interactive Parallelized Display Space

Fig. 1. Highly Interactive Parallelized Display Space (HIPerSpace) at UCSD’sCalit2 was used to analyze GeoEye-1 satellite imagery.

Fig. 2. UAV platforms: (a) RP FlightSystems’ “Spectra” unmanned flyingwing system; (b) the Mikrokopter’s “Oktokopter” eight-blade helicoptersystem.

(HIPerSpace) Wall at the California Institute for Telecommu-nications and Information Technology (Calit2) of the Univer-sity of California at San Diego (UCSD) was used to allow re-searchers to interact with imagery at its original data resolutionin an environment where the context of image features is main-tained due to the large imaging real-estate on the display (Fig. 1)[34].Analysis of the satellite data provided a preliminary map of

potential archaeological sites to be further investigated via aerialand geophysical surveys. These were conducted on-site in Juneof 2009 and July of 2010.

C. Aerial Imaging

Two UAV technologies were tested for usability duringon-site investigations: (a) RP FlightSystems’ “Spectra”unmanned flying wing system and (b) Mikrokopter’s “Ok-tokopter” eight-blade helicopter system (Fig. 2(a) and (b)).

872 IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, VOL. 4, NO. 4, DECEMBER 2011

Both systems were electrically powered with lithium ionbattery packs and recharged in the field using Brunton Solaris52 flexible solar panels. Imagery was captured via PanasonicLumix LX3 digital cameras, which are capable of acquiringdata at a 10.1 Megapixel resolution. One camera was modifiedfor near-infrared wavelength capture, acquiring data in the712 nm range. Both aerial platforms were GPS guided andtracked to enable geo-referenced data acquisition. Multipleimages of sites were stitched together to form an image of thearea of interest in its entirety. Reference markers placed on theground were used to accurately scale and orient images. Usingan “Eyefi” WiFi enabled memory card, data could be deliveredwirelessly to the ground in near-real time to help guide theaerial image acquisition.Due to the constant forward motion required to maintain lift

on the Spectra aircraft, images were captured through a seriesof “fly-by”s over the target site. This required the image sensorto be limited to a fast shutter speed to reduce motion blurring[35]. The Oktokopter on the other hand could hold its positionfor extended periods of time (on the order of several minutes).Neither system required a runway or landing space. Neverthe-less, the Oktokopter only required a single operator for verticaltake-off while the “Spectra” aircraft required both an operatorand launcher.

D. Geophysical Surveys

For the geophysical surveys, a GEM Systems GSM 19 Over-hauser gradiometer (Fig. 3(a)), a Geonics Limited EM38 elec-tromagnetometer (Fig. 3(b)), and a IDS Detector Duo groundpenetrating radar (Fig. 3(c)) were employed.The gradiometer used in this study has a sensitivity of

0.022 nT Hz with 0.01 nT resolution. Its magnetic rangeis between 15,000 nT and 120,000 nT and it has a gradienttolerance less than 10,000 nT/m. During the survey, the dis-tance between sensors was set at 1.5 m and the distancebetween the lower of the two sensors and the ground wasmaintained at 0.2 m. Data was collected in “fast walking” modeat 0.5 s cycling rate following parallel North-South transectsapproximately 1 m apart. The internal sub-meter GPS of thegradiometer was employed for data positioning. Data pro-cessing was conducted with the Oasis Montaj software createdby Geosoft Inc. A de-spiking filter was used to remove extremeoutlying values followed by application of the Kriging method[36] to generate 2D maps of subsurface anomalies. Lag andheading corrections were applied to remove anomalous GPSpositioning and directional effects.The electromagnetometer surveys provided measurements of

ground conductivity (quad-phase) in milliSiemens per meter(mS/m) and magnetic susceptibility (in-phase) in parts per mil-lion. The maximum effective depth range (1.5 m) was achievedby collecting data in the vertical dipole mode. Data collectionwas performed in walking mode at a cycling rate of 2 readingsper second following parallel transects approximately 1m apart.The instrument was kept as close to the surface as possible whilemaintaining a constant height over surface anomalies. An in-ternal sub-meter GPS recorded geospatial positions of scans andan external data logger allowed the operator to view position andraw data in real time. The 2D matrices (typically cells of 0.5 m

Fig. 3. Geophysical survey conducted with: (a) GEM systems GSM 19 over-hauser gradiometer; (b) geonics limited EM38 electromagnetometer; (c) IDSdetector duo ground penetrating radar.

0.5 m size) of data points generated through EM surveys wereprocessed into 2D subsurface models using the Kriging method[36] through the software MVS/EVS (CTech, USA).The IDS GPR system utilizes a dual frequency antenna at

250 MHz and 700 MHz for simultaneous investigation of deepand shallow targets, respectively. A standard procedure for3D GPR data acquisition in archaeological prospection was

LIN et al.: COMBINING GEOEYE-1 SATELLITE REMOTE SENSING, UAV AERIAL IMAGING, AND GEOPHYSICAL SURVEYS IN ANOMALY DETECTION 873

followed [26]. Parallel profiles 0.25 m apart were followedusing string as a guideline, in order to assist the operator inpushing the GPR antenna across a generated surface grid. Thismethod, along with 3D visualization techniques, have beenwidely applied in GPR surveys for archaeology [37], [38].Data processing was performed within GPR-Slice v7.0 (GAL,US) in order to produce amplitude maps at different levelsof depth [39]. A dewow filter was employed to eliminate thecontinuous or low frequency component of raw traces recordedby the radar. A manual gain curve was used to amplify thedata signal and diminish any attenuation effects. Finally, abackground removal filter was applied to eliminate bandingnoise in the radargrams. For 3D data processing a set of 30time-slices 5 ns thick was generated by calculating the squaredamplitude of each trace and then mapped using the Krigingmethod [36] to interpolate data points between scans. Finally,vertical interpolation between time-slices was applied in orderto create a continuous and visually smoothed 3D volume.For the specific site investigated in this study a velocity of0.08 m/ns determined from hyperbola analysis was used for thetime to depth conversion. Iso-surfaces rendering was generatedfrom the 3D volume [40].

III. RESULTS AND OBSERVATIONS

In this section, we present in detail the results of our investi-gation, using the techniques described in Section II.Fig. 4 depicts the results of the investigation of a cluster of

rectangular burial sites, believed to originate from the BronzeAge based on the comparison to other identified sites in Mon-golia [41]. Fig. 4(a) presents GeoEye-1 0.5 m resolution pan-sharpened imagery of the identified rectangular anomaly andFig. 4(b) shows a photo of the site, which consists of rocks or-ganized in a rectangular shape with an opening on the southernaspect. A high resolution aerial image in Fig. 4(c) shows sev-eral burial structures approximately 25 m east and 15 m westof the central feature not obvious in the satellite imagery. Thecomposite aerial image was stitched together from several im-ages taken from the “Spectra” aircraft with a near-IR imagesensor. Overlaying a semi-transparent survey map of multipleEM scans over this image correlates subsurface anomalies withsurface features (Fig. 4(d)). Two points of high magnetic sus-ceptibility (likely related to metal objects) are observed as redmarkers within the central rectangular tomb near its southernaspect.Fig. 5 shows the results of the investigation of the second

archaeological site, speculated to be a Turkish period tombbased on the comparison to other identified sites in Mon-golia [41]–[43]. As shown in the ground and aerial view, thesite comprises a small cluster of stones and a line of imagestones (“babal”, see [41]) that starts at the former and extendseastward. The aerial image was captured from the modifiedOktokopter in the visible spectrum. Not seen from the ground,but easily observed from the aerial image, is a subtle variancein the surface vegetation creating the appearance of a squarering around the main rock cluster. If near-inrared data had beencollected at this site, the Normalized Difference VegetationIndex (NDVI) could have been applied to further highlightany variations in surface vegetation [13]. Unfortunately, at

Fig. 4. Bronze age tomb cluster: (a) GeoEye-1 satellite imagery with anomalycentered in image; (b) observations from the ground; (c) aerial survey usingfixed flying wing UAV and near IR sensor; (d) overlay of EM survey subsurfacemap on aerial survey.

this point in the survey, the NIR UAV sensor had been dam-aged, rendering data collection impossible. The square feature,caused by subsurface soil changes, is apparent in the magneticscan overlaid on the aerial image in Fig. 5(c). Furthermore,two more square-shaped highly magnetic features are detected

874 IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, VOL. 4, NO. 4, DECEMBER 2011

Fig. 5. Turkish period tomb: (a) observations from the ground; (b) aerial surveyusing the modified Oktokopter UAV and visible spectrum camera; (c) overlayof magnetic survey subsurface map on aerial survey (note: square features high-lighted at SE corners with white arrows).

along the line of image stones through the magnetic scan (whitearrows indicate southeast corner of square features). Whilelimited geophysical surveys have been performed on turkishtombs in Mongolia, Ates [42], [43] observed similar magneticanomalies at other sites, suspecting the existence of fired tilesunderneath Turkish tombs.The survey of the third site, a fortification of unknown origin,

is outlined in Fig. 6. The fortification was roughly square, withsides approximately 400 m long, and contained structure fea-tures as small as 10 m in size. Geoeye-1 satellite imagery en-abled the mapping of the entire site, while maintaining sufficientresolution (50 cm) to survey features within the site. (Aerialimaging via UAVs of a feature of this size is possible, howeverthe large volume of images required renders image stiching im-practical.) Fig. 6(a) presents the GeoEye-1 pan-sharpened colorimage of the site (left), with a closer examination of a rect-angular mound located approximately in the center of the site(right). In Fig. 6(b) (left), a semi-transparent magnetic scan isoverlayed on top of the satellite imagery. In Fig. 6(b) (right), an

Fig. 6. City fortification of unknown age: (a) GeoEye-1 satellite survey ofoverall site (left) with closer view of central anomaly; (b) magnetic surveyoverlay on satellite imagery of overall site (left) with conductivity map of EMsurvey conducted on top of central anomaly (right).

EM survey, which was conducted on top of the central mound,is overlayed on the close-up shown in Fig. 6(a) (right). The EMconductivity map presents a point of high conductivity (repre-sented in red) near the middle of the southeast side of the moundand is potentially associated with the presence of a metallic ar-tifact.GPR surveys conducted on the top of the central mound con-

firmed the existence of the anomaly identified through the EMconductivity map presented in Fig. 6(a) (right) and provide in-formation about its depth. Fig. 7 provides a vertical slice of GPRdata overlaid on top of a profile view of themound. The anomalyis observed (in red) near the center of the mound. An iso-surfacegenerated from the GPR data outlines regions of maximum re-flected radar signal providing a 3D rendering of the anomaly.

IV. CONCLUSIONS

This paper presented a three-stage investigation method,combining satellite imaging, aerial photography and geophys-ical surveys, as it was applied in the archaeological survey ofthree different sites in northern Mongolia.It was found that geometric features in the 1–10 meter range

could be identified with GeoEye-1 satellite imagery. Moreover,the ultra-high resolution capabilities of the GeoEye-1 sensormade it possible to simultaneously survey the larger featuresof the fortification site, while outlining smaller anomalies suchas the observed central mound. Overall, the results indicate thatlarge scale surveys for anomaly detection are possible with suchhigh-resolution satellite imagery, and are especially useful whenthe area of investigation greatly exceeds the field of view attain-able with cameras on airborne platforms.

LIN et al.: COMBINING GEOEYE-1 SATELLITE REMOTE SENSING, UAV AERIAL IMAGING, AND GEOPHYSICAL SURVEYS IN ANOMALY DETECTION 875

Fig. 7. GPR survey of central mound: Y-cut slice overlaid onto profile photo-graph and generated iso-surface of anomaly.

Of the two aerial systems employed several advantages werefound in the multi-blade helicopter system over the fixed flyingwing. These included ease of operation (in take-off, flight, andlanding), lower susceptibility to wind and turbulence duringflight, and greater repeatability in image capture quality. Thefixed flying wing presented the advantage of longer flight timedue to lower power consumption. Both systems constitute a lowcost, portable solution for rapidly-obtainable aerial imaging ina large range of environments and weather conditions.Finally, in all the sites surveyed, geophysical investigations

noted subsurface anomalies corresponding to, but not obviousfrom, surface features detected through ground, aerial andsatellite observations. A combination of geophysical surveysalso proved important in investigating various aspects of ananomaly.Overall, the results of the proposed method indicate the ef-

fectiveness of combining data from a diverse array of surveytechniques, in order to identify the context of detected featuresand anomalies, in a fully non-invasive manner.

ACKNOWLEDGMENT

The authors would like to thank Prof. S. Bira and Prof.T. Ishdorj of the International Association for Mongol Studies,and Dr. F. Hiebert of the National Geographic Society fortheir collaboration; the International Association for MongolStudies for its support; the VotK expedition team for its fieldefforts; Dr. D. Goodman, Dr. G. Morelli, and Dr. I. Nicolosifor their assistance in geophysics and data processing; O. Paranand M. Hennig for the development and operation of the UAVplatforms; Prof. F. Kuester and the Viz laboratory at UCSDfor their assistance in satellite data visualization; and Prof.T. Levy, Prof. M. Seracini and Prof. R. Rao for their guidancethroughout this research. This effort was supported by theNational Geographic Society, the Waitt Institute for Discovery,the GeoEye Foundation, and the National Science Foundationunder IGERT Award number DGE-0966375.

REFERENCES[1] D. Riley, Aerial Photography in Archaeology. Philadelphia, PA: Uni-

versity of Pennsylvania Press, 1987.[2] R. Bewley, “Aerial survey for archaeology,” The Photogrammetric

Record, vol. 18, pp. 273–292, 2003.[3] L. Deuel, Flights Into Yesterday: The Story of Aerial Archaeology.

New York: St. Martin’s Press, 1969.

[4] T. R. Lyons, Remote Sensing: A Handbook for Archeologists andCultural Resource Managers. Washington, DC: Cultural ResourcesManagement Division, National Park Service, U.S. Dept. of theInterior, 1977 [Online]. Available: http://hdl.handle.net/2027/mdp.39015003448696, for sale by the Supt. of Docs., U.S. Govt. Print. Off.

[5] R. Lasaponara and N. Masini, “Detection of archaeological crop marksby using satellite Quickbird,” J. Archaeol. Sci., vol. 34, pp. 214–221,2007.

[6] S. Campana and S. Piro, Seeing the Unseen. Boca Raton, FL: Taylorand Francis Group, CRC Press, 2009.

[7] D. Gallo, M. Ciminale, H. Becker, and N. Masini, “Remote sensingtechniques for reconstructing a vast neolithic settlement in southernItaly,” J. Archaeol. Sci., vol. 36, no. 1, pp. 43–50, 2009.

[8] I. Scollar, A. Tabbagh, A. Hesse, and I. Herzog, ArchaeologicalProspecting and Remote Sensing. Cambridge, UK: CambridgeUniv. Press, 1990.

[9] G. Leucci, S. Negri, and E. Ricchetti, “Integration of high resolutionoptical satellite imagery and geophysical survey for archaeologicalprospection in Hierapolis (Turkey),” in Proc. 2002 IEEE Int. Geo-science and Remote Sensing Symp. (IGARSS 2002), 2002, vol. 4, pp.1991–1993.

[10] M. Fowler, “High-resolution satellite imagery in archaeological ap-plication: A Russian satellite photograph of Stonehenge region,” Ar-chaeol. Prospect., pp. 667–671, 1996.

[11] S. Parcak, Satellite Remote Sensing for Archaeology. New York, NY:Routledge, 2009.

[12] K. Wilkinson, A. R. Beck, and G. Philip, “Satellite imagery as a re-source in the prospection for archaeological sites in central Syria,”Geoarchaeology, vol. 21, pp. 735–750, 2006.

[13] R. Lasaponara and N. Masini, “Identification of archaeological buriedremains based on the normalized difference vegetation index (NDVI)from quickbird satellite data,” IEEE Geosci. Remote Sens. Lett., vol. 3,pp. 325–328, 2006.

[14] R. Blom, B. Chapman, E. Podest, and R.Murowchick, “Applications ofremote sensing to archaeological studies of early Shang civilization innorthern China,” in Proc. 2000 IEEE Geoscience and Remote SensingSymp. (IGARSS 2000), 2000, vol. 6, pp. 2483–2485.

[15] M. Madden, “GeoEye-1, the world’s highest resolution commercialsatellite,” presented at the Conf. Lasers and Electro-Optics (CLEO2009), Baltimore, MD, 2009.

[16] J. Aber, S. Aber, and F. Pavri, “Unmanned small-format aerial pho-tography from kites for acquiring large-scale, high-resolution multi-view-angle imagery,” in Integrating Remote Sensing at the Global, Re-gional, and Local Scale. Proc. Percora 15, Land Satellite InformationIV and ISPRS Commission I Conf., 2002, pp. 10–15.

[17] G. J. J. Verhoeven, J. Loenders, F. Vermeulen, and R. Docter, “He-likite aerial photography—A versatile means of unmanned, radio con-trolled, low-altitude aerial archaeology,” Archaeol. Prospect., vol. 16,pp. 125–138, 2009.

[18] B.Walkenhorst, G.Miner, and D. Arnold, “A low cost, radio controlledblimp as a platform for remote sensing,” in Proc. 2000 IEEE Int. Geo-science and Remote Sensing Symp. (IGARSS 2000), 2000, vol. 5, pp.2308–2309.

[19] M. Oczipka, J. Bemmann, H. Piezonka, J. Munkabayar, B. Ahrens, M.Achtelik, and F. Lehmann, “Small drones for geo-archaeology in thesteppe: Locating and documenting the archaeological heritage of theOrkhon Valley in Mongolia,” in Proc. SPIE, the International Societyfor Optical Engineering, 2009.

[20] F. Chiabrando, F. Nex, D. Piatti, and F. Rinaudo, “UAV and RPV sys-tems for photogrammetric surveys in archaeological areas: Two testsin the Piedmont Region (Italy),” J. Archaeol. Sci., doi: 10.1016/j.jas.2010.10.022.

[21] H. Becker, “From nanotesla to picotesla — A new window for mag-netic prospecting in archaeology,” Archaeol. Prospect., vol. 2, no. 4,pp. 217–228, 1995.

[22] M. J. Aitken, “Magnetic prospecting. I. The water Newton survey,”Archaeometry, vol. 1, pp. 24–26, 1958.

[23] B. Frohlich and W. Lancaster, “Electromagnetic surveying in currentmiddle eastern archaeology: Application and evaluation,” Geophysics,vol. 51, no. 7, pp. 1414–1425, 1986.

[24] A. Tabbagh, “Applications and advantages of the slingram electromag-netic method for archaeological prospecting,” Geophysics, vol. 51, pp.576–584, 1986.

[25] N. A. Zeid, E. Balkov, M. Chemyakina, A. Manstein, Y. Manstein,G. Morelli, and G. Santarato, “Multi-Frequency ElectromagneticSounding Tool EMS,” Archaeological Discoveries, vol. 5, CaseStories, 2003.

876 IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, VOL. 4, NO. 4, DECEMBER 2011

[26] A. Novo, M. Grasmueck, D. Viggiano, and H. Lorenzo, “3D GPR inarchaeology: What can be gained from dense data acquisition and pro-cessing,” presented at the 12th Int. Conf. Ground Penetrating Radar,Birmingham, UK, 2008.

[27] B. Bevan, “Geophysical Exploration for Archaeology: An Introduc-tion to Geophysical Exploration,” Midwest Archaeological Center,Lincoln, NE, Special Report No. 1, 1998.

[28] C. Gaffney, “Detecting trends in the prediction of the buried past: Areview of geophysical techniques in archaeology,” Archaeometry, vol.50, pp. 313–336, 2008.

[29] K. L. Kvamme, “Geophysical surveys as landscape archaeology,”Amer. Antiquity vol. 68, no. 3, pp. 435–457, 2003 [Online]. Available:http://www.jstor.org/stable/3557103

[30] J. Johnson, Remote Sensing in Archaeology: An Explicitly North Amer-ican Perspective. Tuscaloosa, AL: The University of Alabama Press,2006.

[31] L. Conyers and D. Goodman, Ground-Penetrating Radar: An Intro-duction for Archaeologists. Lanham, MD: AltaMira Press, 1997.

[32] J. Davis and A. Annan, “Ground penetrating radar for high-resolutionmapping of soil and rock stratigraphy,” Geophys. Prospect., vol. 37,pp. 531–551, 1989.

[33] R. Schowengerdt, Remote Sensing: Models and Methods for ImageProcessing. Burlington, MA: Academic Press, 2007.

[34] K. Ponto and F. Kuester, “Digi-Vis: Distributed interactive geospatialinformation visualization,” in Proc. 2010 IEEE Aerospace Conf., Mar.2010, pp. 1–7.

[35] M. Ben-Ezra and S. Nayar, “Motion based motion deblurring,” IEEETrans. Pattern Anal. Machine Intell., vol. 26, no. 6, pp. 689–698, Jun.2004.

[36] G. Matheron, “Principles of geostatistics,” Econ. Geol. vol. 58, no. 8,pp. 1246–1266, 1963 [Online]. Available: http://economicgeology.org/cgi/content/abstract/58/8/1246

[37] J. Leckebusch, “Ground-penetrating radar: A modern three-dimen-sional prospection method,” Archaeol. Prospect., vol. 10, pp. 213–240,2003.

[38] N. Linford, “From hypocaust to hyperbola: Ground-penetrating radarsurveys over mainly Roman remains in the UK,” Archaeol. Prospect.,vol. 11, pp. 237–246, 2004.

[39] D. Goodman, Y. Nishimura, and J. Rogers, “GPR time slices in archae-ological prospection,” Archaeol. Prospect., vol. 2, pp. 85–89, 1995.

[40] D. Goodman, Y. Nishimura, H. Hongo, and M. Okita, “3-D GPR am-plitude rendering of the Saitobaru Burial Mound #13,” in Filtering,Optimisation and Modelling of Geophysical Data in ArchaeologicalProspecting. Milano, Italy: Fondazione Carlo Maurilio Lerici, Po-litecnico di Milano, 1997, pp. 93–101.

[41] E. Jacobson-Tepfer, J. Meacham, and G. Tepfer, Archaeology andLandscape in the Mongolian Altai: An Atlas. Redlands, CA: ESRIPress, 2010.

[42] A. Ates, “Archaeomagnetic surveys of the anonymous gravesin Khosho Tsaidam, Mongolia,” Archaeol. Prospect., vol. 9, pp.197–205, 2002.

[43] A. Ates, “Archaeogeophysical investigations around the Bilge Qaganmonument in Khosho Tsaidam, Mongolia,” Archaeol. Prospect., vol.9, pp. 23–33, 2002.

Albert Yu-Min Lin received the B.S. degree inmechanical and aerospace engineering and theM.Sci. and Ph.D. degrees in materials science andengineering from the University of California at SanDiego in 2004, 2005, and 2008, respectively.He is currently an Assistant Research Scientist

with the Center for Interdisciplinary Science in Art,Architecture and Archaeology at UC San Diego,a Fellow National of the Explorers Club, and anEmerging Explorer of the National GeographicSociety. His research interests are remote sensing,

crowdsourcing, and exploration. He is the principal investigator of the Valley

of the Khans (VotK) Project, a multi-disciplinary non-invasive search for thetomb of Genghis Khan.

Alexandre Novo received the Ph.D. degree in envi-ronmental engineering from the University of Vigo,Spain, in 2009.He is with Geostudi Astier in Livorno, Italy, and

is the geophysical survey specialist for the VotKProject. He is an expert in ground-penetrating radar3-D high-resolution reconstruction and imaging.

Shay Har-Noy graduated summa cum laude with aB.S. degree in electrical engineering and economicsfrom Rice University, Houston, TX, in 2004 anda M.Sci. and Ph.D. in electrical and computerengineering from the University of California at SanDiego in 2006 and 2009, respectively.His research interests are in the fields of signal pro-

cessing, crowdsourcing, and digital communication.In addition to his technical expertise, he has been rec-ognized twice in theAmerican Alpine Journal for firstascents in the Sierra Nevada mountain range. He is a

senior member of the VotK Project.

Nathan D. Ricklin received the Ph.D. and M.S. de-gree in electrical engineering from the University ofCalifornia at San Diego in 2010 and 2007, and theB.S. degree in electrical engineering from the Uni-versity of Maryland, College Park, in 2003.His research interests include crowdsourcing,

human-machine data aggregation algorithms, signalprocessing for MIMO wireless communications,and communications over time-varying channels.He is the senior field systems engineer for the VotKProject.

Kostas Stamatiou received his Diploma in electricaland computer engineering from the National Tech-nical University of Athens, Greece, in 2000, and theM.Sc. and Ph.D. degrees in electrical engineeringfrom the University of California at San Diego in2004 and 2009, respectively.He is currently a post-doctoral scholar in the De-

partment of Information Engineering at the Univer-sity of Padova, Italy, where he conducts research onthe performance analysis and design of wireless adhoc and sensor networks. He is a senior member of

the the VotK Project.