3D geometrical modeling of excavation units at the archaeological site of Tell ‘Acharneh (Syria)

3D geometrical modeling of excavation units at the archaeological site of Tell ‘Acharneh (Syria)

Journal of Archaeological Science 34 (2007) 272e288 http://www.elsevier.com/locate/jas 3D geometrical modeling of excavation units at the archaeologi...

4MB Sizes 2 Downloads 29 Views

Journal of Archaeological Science 34 (2007) 272e288 http://www.elsevier.com/locate/jas

3D geometrical modeling of excavation units at the archaeological site of Tell ‘Acharneh (Syria) L.-M. Losier a,b, J. Pouliot a, M. Fortin b,* a

Geomatics Department, Laval University, Pavillon Louis-Jacques-Casault, Door 1351, Quebec, Canada G1K 7P4 b History Department, Laval University, Pavillon de Koninck, Door 5324, Quebec, Canada G1K 7P4 Received 27 February 2006; received in revised form 6 May 2006; accepted 8 May 2006

Abstract 3D geometric modeling consists of representing geometric and spatial relationships of volumetric objects. We think it could be helpful in the context of archaeological excavation units representation and analysis. This article presents a procedure developed to generate 3D models from GPS positions taken at the top and the bottom of the excavation units boundaries on the archaeological site of Tell ‘Acharneh (Syria). It shows and discusses two geometrical modeling approaches (voxel and tetrahedral) used in the Gocad 3D modeling tool. Once excavation units are geometrically modeled, it is possible to refer them within a trench or the entire archaeological site, to handle them in various ways (zoom, rotation, translation), to perform on them 3D spatial analysis such as volumetric calculus or intersection computation, to make various kinds of queries such as to find out excavation units that have a certain number of artefacts, to generate sections anywhere in the 3D model, and finally to publish it with VRML (Virtual Reality Modular Language). As well as improving data analysis techniques, we think that if this 3D modeling operation can be done during the excavation, it could greatly help archaeologists to plan more efficiently their daily excavation strategy. Ó 2006 Published by Elsevier Ltd. Keywords: Excavation units; 3D geometric modeling; Voxel; Tetrahedron; 3D data

1. Introduction Nowadays, almost all archaeological excavations use the concept of ‘‘excavation units’’[11], often referred to as ‘‘stratigraphic units’’ since most [3,10,15], if not all, archaeological sites are composed of several superimposed levels/layers of archaeological deposits/debris. Several other terms are also employed by field archaeologists: loci/locus [3,20], lot [30], layer [3,10] and anthropogenic units [3]. Basically, all have a similar meaning referring to a 3D volumetric component of archaeological remains/debris within a site under excavation [4e7,9,26,31,39]. This concept of a 3D volumetric component, utilized to ease the excavation process itself, constitutes the basic element of the recording procedure on a dig [9,34]. * Corresponding author. Tel.: þ1 418 656 2131; fax: þ1 418 656 3603. E-mail address: [email protected] (M. Fortin). 0305-4403/$ - see front matter Ó 2006 Published by Elsevier Ltd. doi:10.1016/j.jas.2006.05.008

Field archaeologists have made use of this 3D volumetric concept for their excavations for years. Nevertheless, surprisingly, most of them are still relying heavily on 2D representations of the excavation units for their analysis and interpretation, and consequently, publications [31,34]. These 2D representations take the form of drawings showing excavation units in section/cut, normally superposed onto each other in a stratigraphic sequence. Moreover, archaeological ‘‘section drawing’’ is a stylistic and arbitrary type of representation. Furthermore, only the sides of excavated areas are usually drawn and therefore, not all excavation units are shown in those drawings. We believe that this unrealistic situation still prevails as a result of the lack of an appropriate, easy-to-use, and cost-effective tool specifically designed for use by field archaeologists. Beyond the representation of excavation units and their usual components (such as soil, color, inclusions, and texture),

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

273

Fig. 1. Tell ‘Acharneh is located on the right bend of the Orontes River, in the Ghab valley, to the west of modern Hama (reproduced with permission of the Tell ‘Acharneh archaeological mission).

the spatial relationships (also called topology1) of the excavation units, be either horizontal or vertical, is crucial to field archaeologists. By definition, the vertical plan corresponds to the orientation of the trench section. Because it allows the relative dating of excavation units in a trench, vertical topology is being considered as the most important one. Indeed, stratigraphy, that is the superposition of layers, is the key concept from which the modern/scientific archaeological excavation process originated [34]. Currently, to represent the topological relationship between excavation units, field archaeologists rely heavily on the diagram matrix [13,14]. Whilst this system seems to meet their needs, we suggest it only does so in part because it is based on conventional unrealistic icons arranged in 2D arbitrary diagrams.

1 ‘‘Topology, used in the context of cartography, concerns those characteristics of geometric objects that do no depend on measurement in a coordinate system.topological relationships are built from the connections and contiguities of objects.’’ [8].

We are of the opinion that for the grouping of excavation units sharing common affinities, as described above, field archaeologists would greatly benefit from Computer-Aided Design (CAD) and Geographical Information Systems (GIS) softwares. These categories of systems offer valuable tools for capturing, modeling, storing, sharing, analyzing, and showing geographically referenced data [37]. Being able to model 3D ‘‘excavation units’’ in realistic 3D representations, at the same time as managing their relationships, would be an asset for field archaeologists because it would allow them to carry out topological analysis and visualize the results in a more realistic manner. Displaying three-dimensional excavation units could be an important aid in understanding stratigraphical relationships and identifying potential patterning [28]. These kinds of systems would also allow us to perform metric analysis, to validate interpretations or to formulate new ones since archaeologists would be able to revisit their site in immersive reality [18,21]. Furthermore, it would even stimulate and put to contribution a larger proportion of their brain in the process of problems resolution.

274

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

Fig. 2. Aerial photography taken in 1999 of Tell ‘Acharneh (reproduced with permission of the Tell ‘Acharneh archaeological mission).

Only recently promising attempts have been made either to undertake 3D modeling on archaeological sites [1,5,7,32,33, 38,39] or to combine it to a GIS [24]. It is a difficult task since CAD tools and GIS systems perform differently in 3D spatial modeling context. GIS software is usually employed with larger

data sets whereas CAD tools are exploited in local applications. GIS systems manage cartographic projection of spatial data while CAD tools do not offer functions to transfer from one coordinate system to another. Furthermore, whilst GIS is usually integrated within a database management system, CAD tools

Fig. 3. Position of trench TEW opened in 2004 to the northwest of the main Tell (reproduced with permission of the Tell ‘Acharneh archaeological mission).

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

275

Fig. 4. Data acquisition on the upper surface of an excavation unit with a GPS Trimble 5800.

have to be connected by way of attributes linking. Finally, GIS systems do not manage volumetric data, despite of the research done in this area [2]. Using experiments made in geology as a starting point [25], we have explored CAD tools capabilities with Gocad software (http://www.earthdecision.com/) in the context of solid modeling of excavation units on an archaeological site (Tell ‘Acharneh, Syria). This was part of a more general project program designed to rethink and redefine the entire excavation process in integrating Geomatics techniques and methodologies and new information technologies (http://archeogeomatique.crg.ulaval.ca). The final product will be an integrated 3D spatial database management system. After a brief presentation of the archaeological site that served as a test case and a description of the data upon which the experiment is based, we will review the 3D geometric modeling process, with a special emphasis on a comparison between voxel and tetrahedron models, and discuss the results obtained. 2. Tell ‘Acharneh: a case study 2.1. A promising site The archaeological/historical significance of Tell ‘Acharneh, in western Syria (Fig. 1), suggests that it may correspond to the location of the ancient city of Tunip which was a major town, if not the capital, of the Ghab valley in Syria during the Bronze Age (Third to Second millennium B.C.). The name of Tunip is mentioned in several ancient texts attesting that it was involved in diplomatic relationships with the Egyptians, the Hittites and the Hurrians, in other words, the major empires of that time. It would explain why it was defended by an impressive rampart, typical of the Middle Bronze Age major settlements in the Levant. In the Iron Age, it was also an important city, the name of which remains unknown, within an Aramean state formed around modern Hama, which also incorporated the Ghab valley; it was so important that it was destroyed by the Assyrian king Sargon II in 720 B.C. as revealed

by an inscribed stela he erected close to the site, which was found by chance in 1924. Archaeological excavations, undertaken by a Canadian team directed by one of the authors [12], have just started: two field seasons in 2001 and 2002, preceded by a brief exploratory campaign in 1998. In June 2004, two 5  5 m squares (TEW) were opened on the edge of the summit of the main hill (Fig. 2). They revealed remains of a substantial fortification wall from the time of the Crusaders’ occupation of the site in 1111 AD, associated with several layers containing artefacts and ceramic sherds from this historical period. Tell ‘Acharneh comprises a very uneven geomorphological landscape: two large hills (tells, in Arabic) to the north of the site, a main one (200  300  45 m) and a secondary one (200  200  30 m), a flat ground to the south mostly covered by modern houses but which corresponded in the antiquity to the lower town, and a mighty defensive wall (50 m wide at the base  12 m high) simply made of earth, surrounding the site, preceded by a moat on two sides, the two others being naturally protected by the course of the Orontes river (Fig. 2). In addition to the complexity of the archaeological field, there is a temporal difficulty since this site has been occupied from 2500 to 720 B.C as well as during the Crusades and Ottoman epochs. Thus, there are several periods of occupation to distinguish from stratigraphic layers. Because several spatial and non-spatial data were already acquired and afterward interpreted and argued by archeologists, Tell ‘Acharneh is a promising site for the implementation of new information technologies. It also offers result’s comparison capability, i.e. with and without information technologies implementation. 2.2. Spatial and descriptive data acquisition In the field, spatial data were gathered in different ways over the years: with a theodolite during the 1998 and 2001 campaigns, with a total station Leica TCR 705

276

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

Fig. 5. Illustration of geometric objects with different dimensions (0D, 1D, 2D, 3D).

(http://www.leica-geosystems.com) in 2002, and finally, with a Trimble 5800 Real-Time Kinematic (RTK) GPS Surveying System (http://www.trimble.com/) during the 2004 season. The altimetric precision of the Trimble 5800 is 2 cm and the planimetric precision is 1 cm. For the experimentation in 2004, two adjacent 5  5 m squares labelled TEW were opened along the crest of the steep western slope of the main Tell (Fig. 3). The criterions that were used to determine the limits of each excavation unit were soil texture, composition, consistency, cohesion, and density. The methodology employed for data acquisition during the 2004 campaign consisted of taking three-dimensional points with GPS on the top of each excavation unit and feature (walls

and pits) and recording the results in a Microsoft Access database (see Fig. 4). The bottom surface of an excavation unit corresponds to the top of one or more excavation units below. The GPS helps to avoid the loss of time and resources during data acquisition. First, we recorded points along the limits of each excavation unit; 20 in all were thus recorded. Second, we took points within each unit as close as possible to each other, according to the same methodology described above. Finally, the topology of each three-dimensional point, about 20e50 points in all for each excavation unit, was entered into the database. Moreover, during the excavations, we acquired descriptive data such as artefacts (flint, terracotta, metal objects, seals, and so on), ecofacts, ceramics, elements of construction and human remains.

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

277

Fig. 6. Classification pattern of 3D geometric modeling approach [25].

This operation in the field with a GPS RTK did not take much longer time than the usual procedure, with a theodolite or a total station, which then consisted of taking a point at each corner of an excavation unit and one or two in the middle of it. The work done with the GPS RTK did not slow down the pace of the excavation, but it allowed archaeologists to take many more points of an excavation unit within the same time period. 3. 3D geometric modeling 3.1. State-of-the-art In order to understand the limitations and outcomes of using three geometric dimensions in the process of archaeological site modeling and analysis, it is important to be familiar with concepts linked to geometric modeling such as model, object, dimension and universe. According to the geographic information standard ISO TC 211 (http://www. isotc211.org/), a model is an abstraction of some aspects of the reality where by, for a specific purpose, some components of the system are approximated, others are simplified, and a few are ignored. Geometric modeling relates to the presentation of geometric properties of the system. For example, representation of the boundaries of an excavation unit could be modeled by assembling geometric primitives2 (e.g. points and lines) to create the surface boundary. The result is an object with spatial attributes corresponding to its shape and location in the embedding space [26]; it is known as a spatial object. 2

Geometric primitives are non-decomposed objects that present information about geometric configuration. They include points, curves, surfaces, and solids [36].

The representation of such an object refers to the basic concept of dimension (D). Some confusion may exist though about the D in the expression 3D since for some it refers to the coordinate system while for others it corresponds to the length, the width and the height of the object itself [17]. Here, dimension refers to the amount of measurements needed to represent the portion of space occupied by an object or its geometric representation. Thus, a punctual object has no dimension (0D); a linear object has a single one, that is its length (1D); a surface object has two dimensions, such as a width and a length, and, consequently, a computable surface (2D); a solid object has three dimensions, since it is made of stocks of length, width and height (3D), and, consequently, a computable volume. Fig. 5 shows examples of objects with different number of dimensions. To represent and store those geometric objects, various geometric modelings are available, especially in the context of the representation of the third dimension. The system of classification applied by Pouliot et al. [25] regarding geometric approaches is used in this article (Fig. 6). Like in 2D, 3D geometric modeling approaches could be split into two groups [19]: (1) space-oriented representations, modeling entirely the world of interest, for example an occupied space, (2) object-oriented representations, allowing the subdivision of space into objects with well-defined boundaries. The partitioning step of space-oriented model could be regular, for example using volumetric elements, or irregular, using tetrahedral elements. The partitioning step of an object-oriented model could be based on the boundary of objects (point, line, polygon or polyhedron) or on parametric figures: sphere, cones, cubes, etc. It is important, especially in the context of 3D geometric modeling, to understand that 3D modeling could be done from surface primitives or by solid primitives, which will have an impact on further

278

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

Fig. 7. (A) An assemblage of voxels representing an excavation unit at Tell ‘Acharneh. (B) One voxel with three axes dx, dy and dz.

spatial analysis capabilities. For example, the boundary representation approach is based on the assemblage of surfaces, which means that there is no volumetric object in the system. In this context, in order to get a volume, geometrical algorithms will have to be exploited. As such, it is feasible although it could be complex, depending on the archaeological system to be modeled. One of the well-known models in the context of space-oriented approaches is the voxel model (volumetric element). It is an extension in 3D of the 2D raster model [7,24,35]. The voxel model involves a regular decomposition of the space into voxels (Fig. 7). The size of a voxel could vary since the spatial resolution in x, y, and z of a voxel can be unequal. Those values have to be determined before modeling the object. Therefore, it is important to choose for the voxels a size appropriate to the spatial extent of the objects to be modeled, which is not necessarily easy for inexperienced modelers. Tetrahedron modeling could be either space-oriented or object-oriented, depending on the methodology used. In the space-oriented approach, instead of using regular decomposition of the space like the voxel model, we could also exploit irregular shapes such as tetrahedron which are polyhedron composed of four triangular faces. Tetrahedron could be used to represent space or boundary of objects. In the object-oriented approach, an object is populated with tetrahedrons constructed from group of points or from the interior of surface model boundaries (Fig. 8). 3.2. An experimentation on Tell ‘Acharneh excavation units In order to perform 3D geometric modeling, we had to choose a 3D modeling tool adequate to our needs and adapted to the specific spatial and descriptive data acquisition procedure (GPS points). In our context, those needs were: volume computation of excavation units, spatial relationship analysis (to display all excavation units at distance range of a specific location), spatial queries (to find all excavation units that have specific qualitative properties, such as its dating, and quantitative ones, such as a certain number of artifacts) and last, easy and rapid 3D model construction. Some commercial softwares can construct 3D models: Autocad and Microstation. Nevertheless, our past experiments of 3D geometric modeling with

Autocad and Microstation softwares showed several limitations, mainly related to volumetric computation. On the other hand, a 3D modeling tool, Gocad (Geological Object CAD), especially designed for geological modeling, has demonstrated relative good performance for 3D model construction and spatial analysis [16,22,29]. The Gocad reference system, like several CAD softwares, is composed of three orthogonal axes (x, y and z) needed to edit, display and manage spatial objects in a 3D universe. The primitives in Gocad are points, curves, surfaces and solids (composed of voxels or tetrahedron). Depending on the hardware used, this system can handle millions of points for the modeling of a surface although the number of voxels is limited (12 millions). Gocad architecture does not incorporate a database management system (DBMS), but rather it stocks in text files the properties of quantitative material such as quantities of artefacts, ceramic fragments, and so on. However, it does not permit directly the recording of qualitative properties, such as dating, consistence of the soil or munsell color; they have to be converted into numbers. 3.2.1. The modeling process on Gocad To construct with Gocad valid 3D objects (excavation units, in our case), connected and closed, we followed a procedure in four steps:3 (1) Data importation: an operation aimed at making available, in the modeling tool, acquired data (points, object identification and recorded attributes). Geometric objects are constructed either from punctual data and/or curve data. Following the 2004 campaign, we created a Pointset4 with GPS points representing upper surfaces of excavation units. For the 1998, 2001 and 2002 campaigns, we imported curves data representing the boundary of excavation units that had been modeled in Microstation software.5 In this 3 Developed by K. Be´dard in her master’s thesis ‘‘La construction de mode`les ge´ologiques 3D: de la standardisation au tutoriel’’, Department of Geomatics, Universite´ Laval, Quebec, Canada, 2006. But her procedure has been transformed for the present research in order to suit the archaeological stratigraphical context. 4 A Pointset corresponds to punctual objects and is made up of a group of points. 5 This work was done by Eve Grenier.

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

279

Fig. 8. (A) Tetrahedron model of an excavation unit, at Tell ‘Acharneh. (B) A tetrahedron composed of four triangular faces.

Fig. 9. Surface validation. (A) Surface model of a construction. (B) Picture of a construction (reproduced with permission of the Tell ‘Acharneh archaeological mission).

Fig. 10. (A) Voxel model of TEW 1 and 2 (20 excavation units), at Tell ‘Acharneh. (B) Tetrahedron model of TEW 1 and 2.

280

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

Fig. 11. Referenced excavation units (294 in all) and squares at Tell ‘Acharneh. An aerial picture has been draped on a digital elevation model.

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

281

Fig. 12. Excavation square CT 119 (Tell ‘Acharneh): (A) top oblique view; (B) peeled off excavation units 2, 3 and 7 from top oblique view; (C) bottom oblique view.

process, we have to check the coherence of the data for possible errors such as points wrongly labelled or located. (2) Surface object creation: a multi-step process using imported points and curves to create a triangle mesh. The latter forms a close and convex surface associated, in our context, to the top and the bottom of each excavation unit. The appearance of surfaces is smoothed in densifying the points with which the triangle mesh was assembled. But, those new points thus generated cannot serve as control points (as a snap position). Then, we apply constraints on the borders so that the surface could only moves vertically during the interpolation. Afterwards, we define

corresponding Pointset as control nodes of the surface to force it to pass by those points. To smooth the surface more, we make sure the triangle mesh is the most equilateral as possible. Finally, we perform an interpolation algorithm on the surface in order to compute the final upper and lower surfaces. For that purpose, Gocad software offers a particular algorithm, the Discrete Smooth Interpolation (DSI) [23]; it is a local linear interpolator (in a least square sense) designed to work in an n-dimensional space. With the two surfaces thus obtained, we are able to create the entire border of an excavation unit, its sides, and we merge them with the upper and lower surfaces to obtain a final

Fig. 13. Transparency on an excavation square, CT 119, at Tell ‘Acharneh.

282

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

Fig. 14. Surface lots of TEW 1 and 2 at Tell ‘Acharneh. Each color simply represents a different excavation unit.

closed surface. Once all the surfaces are computed, we validate the surface model in comparing the result to a photo of the excavation taken during the excavation (Fig. 9). (3) Solid object creation: a process which constructs solid object from closed surfaces. This step depends directly on the topological coherence (no undershoot, no overshoot) of the surface model. Beforehand, we need to choose one of the two solid modeling approaches: voxel or tetrahedron. In voxel creation, we have first to determine the size of the voxel (in our case, it was 7 mm, by 19 mm and 5 mm), and then to decide which surface will limit the regions. Afterwards, Gocad assigns automatically an identification number which will be used to form a region. Finally, we have to clean up regions (usually small and located at the intersection of surfaces) which have no archaeological meaning in filtering then. For the tetrahedron modeling, we simply have to identify surfaces (triangles) associated to an object and Gocad will, still based on DSI method, create the corresponding tetrahedron. (4) Object properties assignment: addition of qualitative and quantitative properties to every excavation unit of the 3D model, in order to be able to do attributes querying. In a voxel model, because the concept of object does not exist, the properties are assigned to a region of voxels. In the case of a tetrahedron model, because every tetrahedron delimited by a boundary surface constitutes an archaeological excavation unit, the properties are assigned to each tetrahedron. 3.2.2. Voxel and tetrahedron models comparison The two 3D geometric modeling approaches share common characteristics, such as allowing queries on attributes, volume

analysis on 3D objects (excavation units), and sectioning on 3D objects with the use of a slicer. But there are also several differences. The solid model voxel (Fig. 10A), which is the most usual one in archaeology [5,7,24], allows the automatic and rapid creation of the solid model for all excavation units within an excavated area without having to partition the space in objects. Properties assignment is relatively easy to accomplish and depending on the topological coherence of the surfaces, data cleaning could be simple. Nevertheless, voxel model produces a stairway effect and shows holes on the excavation unit when its thickness (related to voxel size) is too thin. It also has the inconvenience of being very heavy (675 MB) when the spatial coverage is large and when the voxel accuracy is necessary; its manipulation become thus very slow. Finally, due to the size of the voxel model, querying takes time (56 s). In contrast, the tetrahedron model (Fig. 10B) is faithful to the limits of the surface model because nodes used to create the tetrahedron pass exactly by the control points since tetrahedrons honour complex geometry of objects.6 Visualisation of excavation units with the transparency option is especially interesting. The tetrahedron model requires small files (5 MB), easy to handle, and displays quick results (2 s) once queried. 4. Discussion As previously stated, the aim of our experiment was to evaluate the interest of using 3D geometric modeling in the 6

As suggestd by T. Frank, in his Ph.D. thesis ‘‘Advanced visualization and modeling of tetrahedral meshes’’, Institut National Polytechnique de Lorraine, Nancy, France, 2006.

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

283

Fig. 15. Metric analysis on PN trench (Tell ‘Acharneh).

context of solid modeling of excavations units on an archaeological site. We compared two 3D geometric modeling approaches available on a specific software and we built 3D models from points data and properties information, a procedure which would take only a few hours for someone familiar with this kind of software and with some knowledge in archaeological data management. It depends also on the number of excavation units, their respective geometric complexity and the time taken for the topological data cleanup at the intersections of excavation units. Regardless of those constraints, we have been able to experience several advantages of using such 3D models: 4.1. To understand the site For archaeologists, to be able, with the help of a digital elevation model draped with a rectified aerial picture,7 to reference geographically, at a macro scale, each excavated square/ trench on the site and, at a micro scale, each excavation unit within squares/trenches, would be a tremendous help to get closer to a realistic representation which in turn would improve the understanding of the site (Fig. 11). It is important for archaeologists to see, in 3D, the exact positions in the field of trenches, the regional context around it, and the excavation units dug up in those trenches. 7

In our test, this photo was taken by the French army in 1935; reproduced with permission of the French Institute of Archaeology in the Near East, Damascus.

4.2. To investigate excavation units Every excavation unit can be seen from all angles and scales, as well as their topology (Fig. 12). The 3D modeling tool (in our case Gocad) is user-friendly because with a mouse click, one can zoom, rotate and translate on excavation units in the same click action. Archaeologists might be interested in putting transparency on excavation units, without having to peel them off, in order to see their topological relationships with each other (Fig. 13). With the possibility of changing the color, they could attribute one to an excavation unit in regard to a specific chronology, for example. Surface model can be used for visualization with the aid of visual properties such as colors and transparencies (Fig. 14). Colors used to differentiate excavation units do not follow a standard since such a standard does not exist yet.

4.3. To do metric and topological analysis on an excavation unit An advantage of this 3D modeling is that it is easy to take measurements on an excavation unit, to determine the distance in 3D, in 2D and vertically (Fig. 15). It is easier to measure distances on a 3D model than on 2D drawings. Surface area can be performed to know the areas of the exposed sides of an excavation unit. 3D solid models allow users to perform volume calculation on each object, excavation units in our

284

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

Fig. 16. Volume calculation on tetrahedron model of construction 1 in square TEW 2 (Tell ‘Acharneh).

case (Fig. 16). For archaeologists, it might be useful to know if the volume of a soil sample is representative of the entire excavation unit volume. Further, an archaeologist can practically evaluate the amount of work to do for the next campaign by calculating the volume of earth to dig. Another advantage is to be able to do topological analysis with GisTool plug-in (http://www.geo.tu-freiberg.de/~apelm/gistools.htm): users could display all excavation units which touch one in particular.

4.4. To realize section cuttings anywhere on the solid model Vertical sections and their drawings in 2D are used as a primary source of recording the stratigraphic sequence [39] and do not permit flexibility. Gocad can do this task easily with a slicer, taking horizontal or vertical section everywhere on the 3D model, and then displaying on-thefly results (Fig. 17).

Fig. 17. Making of a section on the tetrahedric model, (A) before sectioning with the slicer; (B) model cut; (C) visualization of a stratigraphical section.

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

285

Fig. 18. Query on TEW 1 and 2 (Tell ‘Acharneh). (A) Script query applied on the 3D model. (B) Results of the query shown in red.

4.5. To do queries on qualitative or quantitative properties As mentioned above, we can perform spatial analysis of 3D model on qualitative and quantitative properties added to 3D objects. For example, to show excavation units of the Early Bronze Age which have more than a certain number of sherds and between X1 and X2 flint flakes. This type of queries allows archaeologists to do ’in situ’ multi-variate analysis, from which they can then put forward hypotheses about the site. With these

queries, they can rapidly locate the results and see the distribution of the results in a three-dimensional space (Fig. 18). Nevertheless, users need to know the querying language, which is, in the case of Gocad, close to SQL (Sequential Querying Language); a well-known language in database operation. 4.6. To help planning future excavations With the help of a digital elevation model of a site on which previously excavated areas are located, archaeologists

Fig. 19. Trench planned for the next campaign of excavation at Tell ‘Acharneh.

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

286

Fig. 20. Visualisation of the VRML from a bottom view of TEW 1 and 2 (Tell ‘Acharneh) with the Cortona viewer.

can determine the exact location of future areas of excavation and even calculate the volume of earth to be removed (Fig. 19). 4.7. To publish excavation data Gocad offers the possibility to do 3D navigation of a site with Virtual Reality Modeling Language (VRML), which can then be put on a Web page (Fig. 20).

4.8. To explore the excavation from a new perspective (3D) Following our experiment, we propose to use 3D modeling as a tool to help field archaeologists to acquire new information by enabling them to explore alternative perspectives. With a 3D model, one can observe more quickly and from different angles the relationship between excavation units without consulting multiple recording sheets. When one takes a measurement on

Fig. 21. Comparison between (A) a traditional recording method (reproduced with permission of the Tell ‘Acharneh archaeological mission) and (B) the 3D model of square TE (Tell ‘Acharneh).

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288

a 3D model after a zoom, one does not have to calculate the map scale for each measurement (Fig. 21). However, we do not go as far as to recommend to field archaeologists to abandon their traditional recording systems but to try using the 3D modeling process as a complementary tool. One should not forget that to feel comfortable with a new technology is a long process.

287

We also envision to experiment other categories of systems, comparable to CAD or GIS systems, in the context of archaeological units representation and analysis, such as the SOLAP (Spatial On-Line Analytical Processing) system which is more friendly-user and whose architecture, a multi-dimensional database, allows to perform complex analytical and ad-hoc queries very quickly [27].

5. Conclusion Acknowledgments It is obvious that field archaeologists have numerous needs for 3D modeling, since archaeological data, and especially excavation units, are complex, multivarious and three-dimensional. Several attempts have been made in the past to apply CAD tools to 3D geometric modeling in archaeological field contexts. In our experiment, we used the Gocad software to reference excavation units within a trench and the whole of an archaeological site, to handle them in various ways, to perform on them volumetric, metric and spatial analysis, to generate sections anywhere on a 3D model, and to publish it with VRML. The 3D modeling procedure adapted to field archaeologists’ needs on archaeological excavations we proposed here, although quick and easy to follow, necessitates some experience in 3D modeling tools and knowledge in mattering, concepts and methods related to geomatics such as geometric modeling, topology, geometric transformation and triangulation. CAD systems of this type could be at the heart of a future 3D GIS.8 We are actually in the process of building an integrated 3D spatial database management system. For that purpose, we tend to favour the tetrahedron model which is a vector-based object boundary approach. This kind of modeling allows users to work with objects instead of arbitrary limits (such as voxels) and will eventually permit to attach properties and behaviour at those objects by way of a database as Oracle spatial 10g. In doing so, we will also benefit from database management system capacities: data security, data sharing, validation, and so on. There are many more developments which would further optimise the procedure of geometric modeling here described: a topological structure of solid objects, automatic tools for 3D data cleaning, 3D models with VRML interface for the web, an immersive reality experiment, as already tested in geology. We are also aware that the procedure to acquire spatial data (points, in our case) has to be improved since it has a direct impact on the 3D modeling construction steps and the accuracy of the final model; the amount and the distribution of points taken on the upper surface of an excavation unit have therefore to be taken into account. Overlapping problems can occur if the lower surface of an excavation unit and the upper surface of the following one differ from each other in the density of points taken. To this effect, we are looking forward to testing a CallidusÒ 3D Laser Scanner (http://www.trimble.com/) in the field, as soon as it will provide for each excavation unit millions of points which will be processed with the PolyWorks software.9 8 As suggestd by M. Apel, in his Ph.D. thesis ‘‘A 3D geological information system framework’’, Technische Universita¨t Freiberg, Freiberg, 2004, p. 102. 9 Developed by the company InnovMetric (http://www.innovmetric.com/) which has generously given us permission to use it for our experiments.

This research was supported by generous grants from GEOIDE (GEOmatics for Informed Decisions) (http:// www.geoide.ulaval.ca/), a Canadian federal-funded Network of Centres of Excellence (http://www.nce.gc.ca/) and from the Social Sciences and Humanities Research Council of Canada, with its RDI program (Research Development Initiatives). The authors also wish to acknowledge contributions from Karine Be´dard, for her helpful comments on a previous version of this article; from Eve Grenier, for her work on Microstation; from Ve´ronique Dufort, Ce´dric Kinnard, and Patrick Robinson for their undergraduate essay on GPS data acquisition ‘‘La cartographie 3D: Une ressource adapte´e aux chantiers de fouille arche´ologique’’, and from Alison Eardley, for her linguistic proof-reading. References [1] P. Allen, S. Feiner, A. Troccoli, H. Benko, E. Ishak, B. Smith, Seeing into the past: creating a 3D modeling pipeline for archaeological visualization, in: Proceedings of the 3D data Processing, Visualization, and Transmission, 2nd International Symposium on (3DPVT’04), IEEE Computer Society, Thessalonika, 2004, pp. 751e758. [2] C. Arens, J.E. Stoter, P.J.M. Van Oosterom, Modelling 3D spatial objects in a GeoDBMS using a 3D primitive, Computers and Geosciences 31 (2) (2005) 165e177. [3] E.B. Banning, The Archaeologist’s Laboratory: The Analysis of Archaeological Data, Kluwer Academic/Plenum, New York, 2005. [4] J.A. Barcelo´, Visualizing what might be: an introduction to virtual reality techniques in archaeology, in: J.A. Barcelo´ (Ed.), Virtual Reality in Archaeology, Archaeopress, Oxford, 2000, pp. 9e35. [5] J.A. Barcelo´, O. De Castro, D. Traver, O. Vicente, A 3D model of an archaeological excavation, in: M. Doerr, A. Sarris (Eds.), The Digital Heritage of Archaeology, Proceedings of CAA2002, Hellenistic Ministry of Culture, Archive of Monuments and Publications, Athens, 2003, pp. 85e87. [6] J.A. Barcelo´, O. Vicente, Some problems in archaeological excavation 3D modelling, in: K.F. Ausserer, W. Borner, M. Goriany, L. KarlhuberVockl (Eds.), Enter the Past: The E-way into the Four Dimensions of Cultural Heritage, CAA2003, BAR International Series 1227, Archaeopress, Oxford, 2004, pp. 400e404. [7] M. Cattani, A. Fiorini, B. Rondelli, Computer applications for a reconstruction of archaeological stratigraphy as a predictive model in urban and territorial contexts, in: K.F. Ausserer, W. Borner, M. Goriany, L. Karlhuber-Vockl (Eds.), Enter the Past: The E-way into the Four Dimensions of Cultural Heritage, CAA2003, BAR International Series 1227, Archaeopress, Oxford, 2004, pp. 299e303. [8] N. Chrisman, Exploring Geographic Information Systems, second ed., John Wiley & Sons, New York, 2002. [9] M. Doneus, W. Neubauer, Digital recording of stratigraphic excavations, in: K.F. Ausserer, W. Borner, M. Goriany, L. Karlhuber-Vockl (Eds.), Enter the Past: The E-way into the Four Dimensions of Cultural Heritage,

288

[10] [11] [12] [13] [14] [15] [16] [17]

[18]

[19] [20] [21]

[22] [23] [24]

[25]

L.-M. Losier et al. / Journal of Archaeological Science 34 (2007) 272e288 CAA2003, BAR International Series 1227, Archaeopress, Oxford, 2004, pp. 113e116. P.L. Drewett, Field Archaeology: An Introduction, Routledge, London, 1999. K.L. Feder, Linking to the Past: A Brief Introduction to Archaeology, Oxford University Press, New York, 2004. M. Fortin, Tell ‘Acharneh 1998e2004, Subartu 18, Brepols, Turnhout, 2006. E.C. Harris, Principles of Archaeological Stratigraphy, second ed., Academic Press, New York, 1989. E.C. Harris, M.R. Brown, G. Brown, Practices of Archaeological Stratigraphy, Academic Press, London & New York, 1993. N. Herz, Geological Methods for Archaeology, Oxford University Press, New York, 1997. M. Jessell, Three-dimensional geological modelling of potential-field data, Computer Geoscience 27 (4) (2001) 455e465. S. Larrive´e, Y. Be´dard, J. Pouliot, Mode´lisation conceptuelle des bases de donne´es ge´ospatiales pour des applications 3D, Revue internationale de ge´omatique 16 (1) (2006) 9e27. G.R. Lock, T.M. Harris, Danebury revisited: an English Iron Age hilfort in a digital landscape, in: M. Aldenderfer, H.D.G. Maschner (Eds.), Anthropology, Space and Geographic Information Systems, Oxford University Press, New York, 1996, pp. 214e240. P.A. Longley, M.F. Goodchild, D.J. McGuire, D.W. Rhind, Geographic Information Systems and Science, John Wiley, New York, 2001. J. Magness, The Archaeology of Qumran and the Dead Sea Scrolls, Eerdmans Publishing, Grand rapids, 2003. E.S. Malinverni, G. Gagliardini, G. Fangi, Virtualisation of an archaeological site, in: Commission 5 Symposium Close-range Imaging, Long Range Vision, ISPRS Archives, vol. 35, part 5, Corfu, 2002, pp. 407e 412. J.-L. Mallet, Geomodeling, Oxford University Press, New York, 2002. J.L. Mallet, DSI: Discrete Smooth Interpolation, Association for Computing Machinery, Transactions on Graphics 8 (1989) 121e144. J.D. Nigro, P.S. Ungar, D.J. de Ruiter, L.R. Berger, Developing a Geographic Information System (GIS) for mapping and analysing fossil deposits at Swartkrans, Gauteng Province, South Africa, Journal of Archaeological Science 30 (2003) 317e324. J. Pouliot, B. Lachance, D. Kirkwood, L’importance de la mode´lisation ge´ome´trique 3D lors de l’e´laboration d’un SIG 3D: Exemple du de´veloppement d’une structure topologique pour une application ge´ologique, Revue internationale de ge´omatique 16 (1) (2006) 29e49.

[26] P. Rigaux, A. Voisard, M.O. Scholl, Spatial Database: With Application to GIS, Morgan Kaufmann, San Francisco, 2002. [27] S. Rivest, Y. Be´dard, P. Marchand, Towards better support for spatial decision-making: defining the characteristics of Spatial On-Line Analytical Processing, Geomatica 55 (2001) 539e555. [28] P. Spikins, C. Conneller, H. Ayestaran, B. Scaife, GIS based interpolation applied to distinguishing occupation phases of early prehistoric sites, Journal of Archaeological Science 29 (2002) 1235e1245. [29] K. Sprague, E. de Kemp, W. Wong, J. McGaughey, G. Perron, T. Barrie, Spatial targeting using queries in a 3-D GIS environment with application to mineral exploration, Computers and Geosciences 32 (3) (2006) 396e418. [30] M.Q. Sutton, B.S. Arkush, Archaeological Laboratory Methods: An Introduction, Kendall Hunt, Dubuque, 2001. [31] D. Tsiafakis, N. Tsirliganis, G. Pavlidis, V. Evangelidis, C. Chamzas, Karabournaki-recording the past: the digitization of an archaeological site, in: V. Cappellini, J. Hemsley (Eds.), EVA 2004: Conference of Electronic Imaging and the Visual Arts, Florence, 2004. [32] K. Uotila, C. Tulkki, Three-dimensional excavation plans and 3D Studio Max, in: G. Burenhult (Ed.), Archaeological Informatics: Pushing the Envelope. CAA 2001, BAR International Series 1016, Archaeopress, Oxford, 2002, pp. 427e430. [33] E. Vote, D.A. Feliz, D.H. Laidlaw, M.S. Joukowsky, Discovering Petra: archaeological analysis in VR, IEEE Computer Graphics Applications 22 (2002) 38e50. [34] D.A. Warburton, Archaeological Stratigraphy: A Near Eastern Approach, Recherches et Publications, Neuchaˆtel, 2003. [35] D. Wheatley, M. Gilling, Spatial technology and archaeology, The Archaeological Applications of GIS, Taylor & Francis, London & New York, 2002. [36] K. Wolfgang, F. Kian, ISO Standards for Geographic Information, Springer, Berlin, New York, 2004. [37] M. Worboys, M. Duckham, GIS: A Computing Perspective, Taylor & Francis CRC Press, London, New York, 2004. [38] T. Wu¨st, S. Nebiker, R. Landolt, Applying the 3D GIS DILAS to archaeology and cultural heritage projects e requirements and first results, in: Proceedings of the XXth ISPRS Congress, vol. XXXV, Part B5, Istanbul, 2004, pp. 407e412. [39] M. Zhukovsky, Handling digital 3-D record of archaeological excavation data, in: G. Burenhult (Ed.), Archaeological Informatics: Pushing the Envelope. CAA2001, BAR International Series 1016, Archaeopress, Oxford, 2002, pp. 431e438.