|
|
Summary of Software and Techniques Used to Construct 3D Visualizations of the Harappa Archaeologic Site, Punjab Province, Pakistan
Software EarthVision has been developed by Dynamic Graphics, Inc. (DGI) of Alameda, California. EarthVision provides 2D and 3D modeling capabilities. DGI was one of the first companies to develop tools to help geologists build visual models. By using EarthVision, 2D and 3D grids of scattered data can be constructed and then used to produce solid graphical bodies. These solid bodies can then be integrated into the visual model. EarthVision can be used to perform mapping, modeling, visualization, and analyses. Earthvision was used to construct thhe gridded surfaces for topography, paleotopography and stratigraphy. EarthVision uses a minimum tension (minimum curvature) gridding technique to produce 2D and 3D grids from scattered data. The specific minimum tension technique used by EarthVision is a bicubic spline algorithm. Minimum tension gridding is an iterative process which seeks to honor the input data when calculating the evenly spaced grid. Initially a coarse grid is calculated and the cubic function is fitted to the grid nodes using the original input data. After each iteration, the value of the grid node is evaluated and compared to the input data If an original data point is located within one-half of a grid cell width, this point is compared to the grid node. As gridding progresses, the residual between the grid node and the original input data point decreases. If the residual increases, the grid node is reset to the a value that is closer numerically to the input data point. EarthVision uses a voxel technique of 3D gridding and isosurfacing. A 3D extension of the minimum tension gridding algorithm is used to calculate a regularly spacing 3D grid from scattered point data of a physical property. The grid then represents the modeled distribution of the input property throughout a defined volume. The property should vary continuously in space. EarthVision fits isosurfaces through the 3D grid and then this isosurface is converted to a triangulated vector data structure that can be viewed in a variety of manners. Intervals can be specified for which isosurfaces or onion peels" are computed within the grid. These surfaces can then be color-shaded to reveal the internal geometry of the geologic body. Graphic manipulation of the grids permits slicing the model to reveal its intemal structure. EarthVision also allows for a basic geologic model to be built as a stratigraphic sequence corresponding to the geologic history of the volume being modeled. The various geologic horizons and discontinuities are imported as 2D or 3D grids. Users have the ability to use these surfaces to influence the intemal property distribution by distributing properties within each of the geologic zones. By graphically removing geologic zones, property variations within the model can be seen. Another unique feature of the EarthVision system is the Geologic Structure Builder. This module enables users to define three-dimensional geologic models where complexities such as faults, unconformities, and complex geologic intersections exist. These complexities can make interpretation and analysis difficult. A common methodology to model these complexities is the creation of composite surfaces such as a combination of topography and a geologic surface where drainages have incised into bedrock. The Geologic Structure Builder eliminates the need for the construction of composite surfaces by establishing the geologic relationships (such as sequences and processes) between geologic surfaces. The Geologic Structure Builder can build fault surface models, fault block models, lithologic surface models, and lithologic zone models. Several geologic rules can be added to various surfaces used to construct a 3D stratigraphic model. These geologic rules determine the appropriate geologic operation which a surface will use to interact with other surfaces within the model. Geologic surfaces can be designated as either depositional, unconformal, or channel erosional. Depositional surfaces do not truncate or modify underlying surfaces or horizons. Unconformities do truncate underlying or older surfaces and can act as depositional surfaces where the unconformal surface lies above the underlying surface. Channel erosional surfaces truncates the underlying surfaces, but does not act as a depositional surface when above an underlying surface. Microstation 95 Software Archaeologic structure models were constructed in geologic modeling software (Earthvision) and in a 3D CAD system. Earthvision software models acted as preliminary models and represent a new functionality for Earthvision software. In order to produce a more realistic model of the archaeologic structures and trenches than could be in Earthvision software, a 3D CAD approach was taken. Microstation 95 software was used to construct the 3D geometries of the archaeologic structures and trenches. These wireframe models were then and rendered using texture patterns appropriate for the various structures and trenches. Microstation 95 software, distributed by Bentley Systems, is a high end CAD system, representing a comprehensive drafting and design package. Microstation software was initially developed from Intergraph Corporation design software and runs on a variety of platforms. Microstation software is a 3D CAD package that can represent and manipulate points, lines, and areas in 3D space. A 'design' file contains all the graphical and geographical elements (Sahai, 1996 and Haque and others, 1996). Microstation software contains 63 design levels, allowing the display of different types of data on each level. Microstation software was used to model the archaeologic structures and trenches, as well as representing the final gridded surfaces from Earthvision software. It serves as the integrating program for merging the variously scaled visualizations and models and linking them to a database. Microstation software also allows the display of other design files within an active design file. These other design files, called reference files, are externally referenced design files whose contents can be displayed on top of or beneath the active design file. The displayed graphical elements can be manipulated without changing the original referenced design file. Microstation software can also integrate with a variety of relational data base systems, allowing geographic features to be related to one or more data tables. A relational database has the ability to access data organized in tabular files. These files maybe related together by a common item. The database can then recombine the data items from different files without repeating information. This provides a powerful tool for data analysis. Model Construction Techniques Topographic Surface Construction The topographic surface is a 2.5D representation of the modern topography at the Harappa site. EarthYision was used to perform the gridding and editing operations to construct the topographic surface. The majority of the data for this surface came from M.S. Vats' topographic map published in 1940, with some additional HARP topographic survey data. During the gridding process, the xyz scattered data are converted to a regularly-spaced grid. This process involves the interpolation and extrapolation of the scattered data into points on a regular grid. The grid represents the discretization of the continuous topographic surface. There are various interpolation schemes to accomplish this, such as minimum tension interpolation or kriging (EarthVision uses a minimum tension gridding algorithm). The gridding of scattered data by modeling programs involves the specification of the spatial range of the grid and the grid spacing. The spatial range must be chosen with care in order that the entire range of interest is contained within the grid. A critical step is the selection of the grid spacing. Tradeoffs must be considered between data realizations, computational speed, and file storage size. Various size grids should be experimented with considering these factors. The spatial limits (in the local site coordinates) of the topographic surface grid (and the other surface grids) are N750, E2350 to N2050, E 2850. A grid spacing of 5 meters was used, resulting in a grid of 261 rows and 301 columns. As mentioned previously, the topographic data was digitized from several maps used by HARP and previous excavations. Using adjacent data points lying on the same digitized contour line the average distance between data points was approximately 6.6 meters (minimum distance between data points was 0.25 meters and the maximum distance was 29.64 meters). Because the grid cannot resolve topographic features less than 5 meters in size, this is considered to be the horizontal resolution of the digital terrain model at this scale. Since the contour interval of Vats is 1.64 meters (5 feet), this could be considered the vertical resolution of the this digital terrain model. All surfaces generated in this research (paleotopography and the stratigraphic layers) maintain the spatial parameters (area and grid spacing) of the topographic surface grid. Paleotopographic Surface Construction In order to construct a stratigraphic model of the Harappa site, it was necessary to determine the base of the model. Since the stratigraphy of the Harappa site is primarily anthropogenic, it was determined that the base of the stratigraphic model should be the topography as it existed prior to occupation by the early Harappans. The elevations of natural soils obtained from HARP excavations from 1986 through 1995, as well as the Master's thesis of Elise Pendall (U.C. Berkeley, 1989). The current HARP interpretation of ancient Harappa existing in a fluvial floodplain was also used to reconstruct the paleotopography. Geomorphic features and depositional environment considerations were also considered. From these data, and the professional judgment of archaeologists and geoscientists, the original topography was reconstructed. Initial data were "ridded at a 5 meter spacing (to match the spatial discretization of the modem topography) and then edited using EarthVision's interactive grid editor. Several iterations were performed with editing and consultation with HARP archaeologists to reach the final paleotopographic surface. Pendall's work was useful in obtaining generalized terrace locations. Her work has produced a soil map which identifies the soil types comprising the various fluvial terrace deposits apparently laid down by the River Ravi. The concentric nature of the soil deposits around me central cultural deposits suggest a migrating meander bend of a river. Because of the pattern of the soils, it was interpreted that a rive channel was present to the south of the present day mounds in ancient times. These soils to the south are either older than or concurrent with the site occupation (avulsion of the Ravi River subsequent to site occupation would have obliterated the mounds). The elevations used to construct the paleotopographic surface are the elevations of the natural (i.e., non-cultural) soils beneath the archaeological deposits. These elevations, especially in the eastern part of Mound E may be influenced anthropogenic activities. These areas may have been locations of raw material excavation during the settlement period of Mound E. These areas may have provided clays for the manufacture of bricks used in the construction of ancient structures. HARP archaeologists believe that the data collected beneath Mound E probably represents the actual topography, but other data points may reflect human influences. The natural soil beneath the southwestern corner of Mound AB possess truncated soil horizons, probably due to human activity removing much of the soil profile. Stratigraphic Surfaces Construction The surfaces produced to construct the 3D stratigraphic model were the paleotopography (the base of the visualization), the tops of the Periods 1, 2, 3A, 3BC, and bottom of Period 4/5 deposits, and the current topography. It should be noted that the original topographic surfaces for the various periods do not necessarily represent the actual topography at that particular time. Rather, they represent the top/bottom of the various deposits. The cross section and cultural deposit map from HARP were used to construct the various surfaces needed to produce the 3D stratigraphic model. Vertical profiles were taken from the cross section to obtain elevation changes along the central part of the mounds. Cultural deposits that extended across the most of the cross section, such as Periods 3A, 3BC, and 4/5, were assigned the appropriate Northing coordinate from the cross section line (either N1380 if west of E2050 or N1240 if east of E2050) and an Easting coordinate from the digitization. Deposits that were primarily contained within a single mound, such as Periods 1 and 2, were assigned Northing coordinates based on a section line connecting the centerline of the period mound's topography (i.e., the line of highest elevation). The lateral extent of each deposit was obtained from the cultural deposit map by digitizing the outline of the particular deposit as a polygon. These elevations and polygons were then used to outline the extent of the Period 1, 2, 3A, 3BC, and 4/5 mounds and construct the topography of each mound. Some special consideration were taken in the construction of the cultural deposits surfaces. It was assumed that gullies oriented along the cardinal directions were the surface expressions of ancient streets. The shape and elevations of the gullies were added to the appropriate surface in locations where these gullies dropped below the surface defined by the Kenoyer's urban growth model. The Period 3BC surface also reflects the excavation of a 'tank' or depression to hold drain water (located between the ancient mounds and modern Harappa). Stratigraphic Model Construction The cultural layers are based upon HARP's chronology at ancient Harappa. The stratigraphic layers are those cultural remains deposited during the periods delineated in the chronology. In order of decreasing age these layers are Period 1, Period 2, Period 3A, Period 3BC, and Period 4/5. Period 3B and 3C and Periods 4 and 5 were combined into single layers (respectively) due to their very similar natures. The base of the model is formed by the paleotopography. Period 1 deposits represent the earliest occupation of the site and is characterized by its pre-urban nature. Period 2 represents the urbanization of the area, with distinctive mud-brick architecture. Period 3 (subdivided into A, B, and C) represents the peak of the urban stage at Harappa. Periods 4 and 5 represents deposits acquired during a cultural change at ancient Harappa, until eventual abandonment of the site. To construct the 3D bodies that depict the cultural deposits at Harappa, the stratigraphic surfaces were used to delineate the top and bottom of each deposit. For the Period 1 deposits, the bottom surface was the natural topography with the top surface being the top of Period 1 deposits. Period 2 deposits were delineated on the bottom by the top of Period 1 deposits and on the top by the top of Period 2 surface. The bottom surface of Period 3A deposits was represented by the top of Period 2 surface and the top surface was formed by the top of Period 3A grid. The bottom of the Period 3BC layer was formed by the top of Period 3A deposits and the top of Period 3BC surface. Period 4/5 cultural deposits were delineated by the bottom of Period 4/5 layer (essentially the top of Period 3BC surface for Mounds AB, E, and ET) and the modern topography surface. Figure 10-2 presents a diagrammatic explanation of the surfaces used to construct the stratigraphic model of the Harappa site. EarthVision contains a Geologic Structure Builder that allows users to model complex 3D geologies with faults, unconformities, and geologic intersections. Surfaces and their geologic relationships can be designated and from these, layers constructed for a 3D stratigraphic model. With these geologic relationships in mind, the various surfaces generated to show the topography of the Harappan mounds through time were used to generate a 3D stratigraphic model. Some surfaces were slightly modified to clip the surface beyond the extent of the deposit, to avoid the construction of strata (even infinitely thin layers) where none was present. This was done to avoid overlap in the construction of the layers. The paleotopography surface was used as the base of the model, with succeeding layers deposited on it. The Period 1 surface was designated as a depositional layer on the paleotopography. The Period 2 surface was designated as a depositional surface and allowed deposition of Period 2 layer on the Period 1 mound and the paleotopography. Period 3A topography was likewise designated as a depositional surface depositing the period 3A layer on the Period 2 mound and paleotopography. The Period 3BC surface, which contained much common topography with the modern topography, was designated as an unconformity. This would allow for the deposition of Period 3BC deposits across the landscape and the incisement of drainages from the modern topography. The Period 4/5 surface was designated as a depositional surface to deposit the Period 4/5 layer on the top of the Period 3BC mounds. Three-Dimensional Archaeologic Structure Model Construction Because of the interrelated nature of the architectural structures and the surrounding stratigraphy, a methodology was designed to construct archaeologic structures using Earthhvision. EarthVision was only used to model the bastion/gateway complex. The visualizations were constructed using the archaeological data from the plans and section drawings. Plans of the surface expressions of the various brick architecture were colored to outline the various structures of different periods and rebuilding phases. These outlines were then digitized to form definitional polygons of the various brick architecture structures. An approach to geologic modeling called "volumetric modeling" allows the construction of irregularly shaped geologic bodies. The volumetric approach essentially involves the construction of geologic bodies using many irregular components to represent the complex 3D shape. These irregularly shaped components can then be merged into a single complex body by assigning the same attribute to each component. ln the same way, various "archaeologic bodies" were constructed using various bounding surfaces, such as top and bottom elevations and the outlines of the individual structures that comprise the entire structure. These individual bodies or volumes can then be assigned attributes according to their archaeologic feature designation and merged together into a single visualization. In this way, different phases of the construction of the architecture or different building materials can be represented. Section drawings were used to determine the top and bottom elevations of each of the structures. These elevations were used to construct clipping grids (usually flat or with very gradual slopes). The 3D visualizations were constructed using the polygons to construct the structure outline and the clipping grids to construct the top and bottom of the architecture. These individual visualizations of each structure were then merged to form an overall visualization of the various baked brick and mud-brick structures. The various period structures and rebuilding phases are represented with different colors. Microstation software was selected to construct another, more realistic model of the gateway complex. Incorporation of the digitized structure footprints into Microstation software required conversion of Earthvision software footprint polygon files to AutoCad DXF file formats that were readable by Microstation software. This process was accomplished using software conversion routines available at CSM. For each archaeologic component of the gateway complex, the following five step procedure was undertaken within Microstation software: 1) Import the DXF file containing the structure footprint polygon; 2) Convert the structure footprint polygon to a Microstation software surface feature, as yet unrelated to model elevations; 3) Locate the surface at the appropriate elevation; 4) Extrude surface to form a 3-D solid corresponding to the height of the structure and incorporating an appropriate taper as necessary (for example, taper is typical of perimeter walls); 5) Assign the resulting 3D structure model to its appropriate Microstation software design file level and associate a color code to it for rendering purposes. Completion of these steps for all the individual archaeologic structures produces a wireframe model of the gateway complex. Rendering is the process of adding realistic lighting and textures to a 3D model. Photo- realistic rendering, while useful in representing an image that is readily understood, was not an objective of this research. For example, a rendered brick pattern on an archaeologic structure, shown in subsequent figures, does not represent the actual position of bricks on the exposed 3D surfaces. Complex wireframe models are not readily interpreted by persons who are unfamiliar with the details of the structures being visualized. A simple form of rendering, incorporating basic colors, textures, and lighting, was utilized by this study. In order to render an image, the lighting and material assignments must be set. Microstation software offers several ways to manipulate global (ambient) and source (point) lighting. Material assignment involves the selection of texture maps that will be used to color graphical elements. Texture maps are 2D graphical patterns that represent textures such as water, wood, and masonry. The texture maps are wrapped or tiled around the 3D graphical image (similar to applying wallpaper to a wall) and are stored in a palette file. The palette file actually stores settings for the image and a pointer to the location of the image file. Various texture maps from palette files can be assigned to graphical elements and these settings are stored in a material allocation file. Microstation software allows users to render objects in a realistic manner by adding color, texture, and lighting to graphical elements and views. Slides taken at the Harappa site that showed appropriate representations of Harappan brick patterns were scanned and saved as TIFF graphics files. These images were edited and clipped to contain a 0.5 meter by 0.5 meter brick pattern that could be repeatedly tiled across a 3D graphical element. No acceptable slide of the mud brick was found due to the erodibility of the relatively soft mud bricks. Therefore, the color of the reddish baked brick patterns was adjusted to be the yellowish color of the mud bricks, producing a brick pattern that resembled the mud brick pattern. These image files of the baked and mud brick were then introduced into a rendering palette file. In order to tile images onto 3D representations of the baked and mud brick structures, the images were scaled in the palette to the 0.5 meter by 0.5 meter area represented by the image. This enables tiling of the image across the structure to reproduce the size of the image in the real world. Once the palette has been constructed, the materials are assigned to appropriate structures. The reddish brick pattern was used to render structures containing reddish colored baked-brick and mudbrick and the yellowish brick pattern was used to render structures composed of yellowish mudbrick. It should be noted that the brick patterns probably do not accurately reflect the size of the bricks used by the Harappans. Several brick sizes were used for both the baked and mud brick structures by the Harappans during various cultural periods. These brick patterns should not be considered accurate in a scientific sense. The placement of a rendered brick does not represent the position of an actual brick in the structure. The brick patterns should be considered a visual representation of the brick patterns that give a sense of the look of the structure and not the actual locations of individual bricks. by Wayne Belcher |
|
| SLIDESHOW | CREDITS |
| CONTENTS | | HOME | © Harappa 1997 |