Through the spatial arrangements of temples, houses, roads, and more, the built environment provides a window to human interaction (Lawrence and Low, 1990). Spatial configurations influence how people negotiate their surroundings. People “read environmental cues, make judgments…and then act accordingly” (Rapoport, 1990: 139), and these decisions in turn, affect the frequency and intensity of interaction (Fletcher 1981). While many factors influence interaction within landscapes, in this paper I focus on visibility.
The visibility, intervisibility, and invisibility of features communicate information that guides pedestrian movement, and consequently, structures social interaction and community organization (Llobera, 2003, 2006; Gillings, 2015). Building on these ideas, this paper uses Geographic Information Systems (GIS) and 3D visualization to explore the role of visibility in ancient landscapes asking: How might visibility influence where people went, what they did, who interacted with whom, and how did these interactions shape their daily experiences?
The case study is the ancient Maya polity of Copan (Figure 1) ruled for over four-hundred years by a line of dynastic kings who by the late eighth century were facing mounting sociopolitical and environmental problems (Fash, 2001). Copan’s final dynastic ruler, Yax Pasaj, like the rulers of many other Maya polities, was coping with strenuous environmental, demographic, and sociopolitical circumstances that would ultimately lead to the kingdom’s demise. Yet during this time of stress, it seems that Yax Pasaj carried out a major urban renewal project commissioning several new temples in the city center that elevated Copan’s skyline. Given the changes to Copan’s urban fabric, Yax Pasaj’s reign is an ideal case study to investigate the role visibility may have played in the production or reproduction of social interaction among the ancient Maya.
At Copan, as at other Maya centers, imagery on ceramics, murals, and freestanding monuments depicted deities floating over lords who subsequently looked down over lower-ranking persons. Maya architecture replicated this vertical succession by elevating royal compounds above other architecture, and in essence linking Maya rulers to the heavens (Messenger, 1987). In terms of visibility, epigraphic decipherments indicate that “seeing” afforded high status, and sight had an authorizing gaze and witnessing function—similar to Foucault’s (1995) Panoptic gaze—where those who were all-seeing were all-knowing (Houston et al. 2006: 173). In order to be all-seeing or to give such an impression, however, Maya rulers needed to be seen, and so often located themselves in physically high and easily visible places or built tall temples that dominated the landscape.
While we know that Maya kings typically constructed highly visible temples, we actually know very little about the role visibility may have played in structuring social connections and daily interactions among social groups. To do this we need to broaden our view from civic-ceremonial precincts to encapsulate the broader landscape (Doyle et al., 2012; Richards-Rissetto 2010; Landau, 2015).
Early visibility studies in the Maya region focused on astronomical alignments among structures, freestanding monuments, and the sky (Aveni and Hartung, 1986). Later, ethnographic studies showing that contemporary Maya often use sight lines to mark out spaces (Hanks, 1990) inspired researchers to investigate whether non-astronomical lines-of-sight also existed at ancient sites; and in fact, archaeologists identified sight lines between a major temple and outlying stelae at the site of La Milpa, Belize (Hammond and Tourtellot, 1999). Recent research has moved away from lines-of-sight between two objects to study the relationships that an object may have to the many objects or features found within a landscape, referred to as a visualscape (Llobera, 2003). Simple line-of-sight measurements cannot provide data on the relationships among multiple objects because they are done along a fixed line; however, visualscapes can be measured using viewsheds that calculate an object‘s entire 360° field-of-view using GIS.
A GIS links mapped features to attributes stored in a database and overlays different data layers such as land usage, elevation, and buildings to help reveal complex patterns, relationships, and trends that are not readily apparent using other tools such as traditional databases not linked to maps.
Pros: In regard to visibility analysis, GIS allows archaeologists to move beyond line-of-sight analysis to viewshed analysis. A viewshed uses raster data (pixels) to identify all cells visible from one or more viewpoints in a landscape; all non-visible cells are assigned a 0 and all visible cells are assigned a 1. This basic binary schema allows for complex mathematical calculations, for example, Boolean operations or map algebra, to calculate topographic prominence of individual features (or classes of features) and percentage of intervisibility among features.
Cons: “Viewsheds depicted in a GIS map bear little resemblance to what people experience on the ground” (Conolly and Lake, 2006: 233). This limitation occurs because viewshed data are 2.5D. In other words, viewsheds store heights and elevation, but they are not actually 3D models (Figure 2). For digital humanists, these flat maps lack a sense of mass, scale, and aesthetics integral to human perception and experience, and the numerical outputs fail to differentiate visibility of a building’s façade versus its sides or back—essential for close reading interpretation. Technically, data resolution (i.e., ratio of pixel size to earth’s surface) can dramatically affect viewshed results—low spatial resolution often masking variation and too high a spatial resolution underestimating visibility (King et al., 2015).
3D technologies offer an alternative to GIS. 3D data acquisition (e.g., airborne LiDAR, terrestrial laser scanning, and photogrammetry), 3D modeling (e.g., SketchUp, 3D StudioMax, Agisoft), and interactive 3D visualization (e.g., Unity, Oculus Rift) are transforming archaeological practice. But, what impact are such 3D technologies having on visibility analysis across ancient landscapes? Airborne LiDAR, for example, rapidly collects 3D data for archaeological sites across vast areas (Thompson and Prufer, 2015). Most LiDAR data are of unexcavated mounds requiring subsequent 3D modeling of architecture and proper alignment within terrains in order to perform visibility analysis—traditionally time-consuming tasks (Richards-Rissetto, 2013).
While most visibility analyses of archaeological landscapes use traditional 2.5D GIS, recently archaeologists have been exploring the potential of 3D approaches for visibility analysis in archaeology. Paliou (2014) developed a computational visibility approach to analyze the visual range of paintings first using 3D modeling programs (3DStudioMax and AutoCAD) and then converting the results into raster maps to be analyzed in a GIS. Dell’ Unto and colleagues (2015) bring georeferenced 3D architectural models (using laser scanning and photogrammetry) into a GIS to calculate visibility of building interiors at Pompeii. While Saldana and Johanson (2015) also use 3DGIS, they employ procedural modeling to rapidly generate alternative 3D building reconstructions based on a set of architectural rules and attributes stored in a GIS to explore visibility in Ancient Rome (Saldana, 2015).
Building on this scholarship, I employ an iterative 3DGIS approach to explore the role of visibility at the ancient Maya site of Copan—today a UNESCO World Heritage Site in Honduras. The approach is twofold: computational and experiential. In the computational approach, I employ traditional 2.5D viewshed analysis in GIS to establish a baseline for comparative analysis with viewshed results in 3DGIS.
First, I use ArcGIS 10.3 (standard GIS software) to assign known building heights and interpolate building heights of unexcavated mounds and run viewsheds to calculate topographic prominence and percent visibility in relation to settlement of major temples and classes of architecture (Richards-Rissetto, 2013). Recent acquisition of airborne LiDAR data has generated a 1m resolution terrain allowing for greater accuracy than earlier visibility analyses (Richards-Rissetto, 2010; von Schwerin et al., 2016). Second, I employ CityEngine—a procedural modeling program that convert GIS data to 3D models—to generate 3D models for Copan’s 3,000+ buildings with the LiDAR terrain using the GIS data and a set of architectural rules as well as laser scanned and photogrammetric models of some standing monuments at Copan (Figure 3) (Muller et al., 2006; Richards-Rissetto and Plessing, 2015; von Schwerin et al., 2013). These procedurally-generated 3D models are then returned to ArcScene (a 3D viewer for ArcGIS) and the viewshed analysis is rerun for comparative analysis of 2.5DGIS vs. 3DGIS of visibility at ancient Copan.
In the experiential approach, I export the 3D models and terrain from CityEngine into Unity 5—a gaming engine—to interactively explore the 3D models. In this model, vegetation is added to the landscape and avatars proceed along set paths generated from a combined cost surface and visibility analysis (Figure 4) (Richards-Rissetto, 2013; Richards-Rissetto and Landau, 2014). Oculus Rift—a head-mounted virtual reality display—is employed to create an immersive experience for ancient Copan as a means to more intuitively interact with archaeological data (Bartolo et al., 2000; Frisher and Dakouri-Hild, 2008).
Strongly embedded in the Digital Humanities, this 3DGIS iterative approach tacks back and forth between 2.5D and 3D data to compare results and potentially derive new methods and interpretations for visibility analysis of ancient landscapes—analyses that would not be possible without taking advantage of the digital to cross-cut the computational and experiential.
The Layman Award, University of Nebraska-Lincoln provided a seed grant to carry out initial procedural modeling tests. This research would not be possible without permission and assistance from the Honduran Institute of Anthropology and History (IHAH). The MayaArch3D Project has generously providing the airborne LiDAR data and the laser scanned and photogrammetric models for this research. I want to thank UNL students Zachary Day, Stephanie Sterling, and Rachel Plessing for their important work on the visualizations.