By: Gary L. Christopherson & D. Phillip Guertin
The University of Arizona
Presented At:
The Annual Meeting of the
American Schools of Oriental Research
New Orleans, Louisiana -- Marriot Hotel
November 23-26, 1996
Geographic information systems (GIS) are proven tools for discerning the relationship between ancient humans and their environment. This has led to an emphasis in GIS circles on environmental models for explaining ancient settlement strategies. Less common are GIS approaches that include socio-cultural factors in settlement strategies. This study applies one such approach, visibility analysis, to data from the Umayri regional survey. The results suggest that visible communication played an important role in local settlement strategies throughout antiquity.
Geographic information systems (GIS) are one of the most recent technologies appropriated by archaeologists. More common in New World archaeology, GIS has recently been making inroads in Near Eastern archaeology, and there are now a number of projects that have incorporated GIS into their research designs. With a couple of exceptions, the GIS presentations at past ASOR meetings have focused on the abilities of GIS to model the environment, and thereby make connections between ancient humans and the natural world. For example, the Madaba Plains Project has been using a GIS to create environmental probability models for hinterland sites in the Umayri and Hesban regions. The predictive ability of these models has routinely provided an improvement over chance of 20 to 60 percent, allowing archaeologists to focus their efforts on relatively small areas, with the expectation that they will still find the majority of sites within the region (Christopherson, 1994; Christopherson & Dabrowski, forthcoming; Christopherson, Guertin, & Borstad, 1996).
While these environmental models have proven beneficial to survey archaeologists, they have also created some problems. If we assume that ancient humans lived and worked within the constraints of their environment, but were not determined by it, we should expect that a variety of socio-cultural factors would also have been important considerations in site location strategies. Unfortunately, these factors can be difficult to define, and almost impossible to quantify. Since GIS are essentially quantitative, there has been a tendency to avoid the problem by ignoring cultural factors and concentrating on a more determinist ecological systems approach to settlement patterns.
This is changing. GIS are powerful tools whose potential for archaeological research should not be restricted to environmental models. A number of recent GIS based studies that included socio-cultural factors in settlement strategies have recently been published (Boaz & Uleberg, 1995; Gaffney & Leusen, 1995; Gaffney & Stancic, 1991; Gaffney, Stancic, & Watson, 1995; Stead, 1995; Wheatley, 1992; Wheatley, 1995) . Most notable are approaches that utilize logistic trend surface, cost surface, and visibility, or viewshed analysis. Discussion in this paper is limited to visibility analysis, and addresses two questions raised by the Umayri regional survey of the Madaba Plains Project. The first of these concerns the limited visibility at Tall al-Umayri and uses a GIS to reconstruct a system of watchtowers that would have functioned to expand Umayri's visual control of the surrounding landscape. The second examines the issue of site intervisibility as a factor in the site selection process.
The Madaba Plains Project has been involved in Jordan for almost 30 years, with excavations and regional surveys at Tall Hesban, Tall al-Umayri, and Tall Jalul. The Umayri regional survey, was an intensive investigation of archaeological remains within five kilometers of Tall al-Umayri. Within this region, 133 archaeological sites were documented, ranging in size from large tells to small agricultural installations, and in age from Palaeolithic to late Islamic (Boling, 1989; Christopherson, forthcoming, 1997; Younker, 1991). The survey collected a variety of environmental, geographical, archaeological feature, and ceramic data from these sites. These data reside in a FoxPro database (Microsoft Corporation, 1993), allowing the researcher to query the sites based on feature types, ceramic periods, etc. in order to create point maps based on a variety of data types.
Additionally, an extensive GIS has been developed for the region. If the reader is not familiar with these spatial database systems, discussions can be found in many sources, including Burrough, 1986; Christopherson, 1996; Kvamme, 1989; and Kvamme, 1992. A short description can also be found by folowing this link to a GIS Primer. For the Umayri GIS, data for hypsography, hydrography, physical, and cultural features were captured from the Jordan 1:25,000 series of topographic maps. Elevation, roads, and wadi channels were taken from the following sheets:Amman, Sheet 225/145 (D. Survey, 1958a), and Naur, Sheet 225/135 (D. Survey,1958b). Additionally, geology and soil maps were created in 1992 by the MPP geologist, Douglas W. Schnurrenberger. Based on 1:10,000 aerial photographs and field samples, these maps were created specifically for the Umayri GIS.
The coordinates for all environmental and archaeological data were based on the Palestine Grid and Transverse Mercator projection of the Jordan 1:25,000 series maps. The unit of measure for this coordinate system is the meter, and the point of origin is 035° 12' 43.490" E. longitude, 31° 44' 02.749" N. latitude. The digitized area was contained within the following coordinates: Minimum X = 228000 meters East, Maximum X = 240000 meters East, Minimum Y = 136000 meters North, Maximum Y = 148000 meters North, representing an area of 144 square kilometers. This area was gridded, with a cell size of 20 X 20 meters, created a matrix of 600 rows and 600 columns, or 360,000 cells. Of the cells in the grid, 195,356 were contained within the five kilometer radius of the Umayri survey region. Although the viewsheds made in this study were not confined to the survey area, all analysis was.
Derived from the captured data, a variety of environmental data themes were developed in this GIS. Most important for this study was the digital elevation model (DEM). The DEM for the Umayri region was interpolated in ARC/INFO using the TOPOGRID command. This command executes a program module based on algorithms developed by M. F. Hutchinson allowing for the interpolation of a hydrologically correct DEM based on the captured elevation contours and hydrology vectors (ESRI, 1995; Hutchinson, 1989).
In antiquity, visibility would have been an important aspect of communication, and therefore critical for site location strategies. A number of ancient sources mention the importance of visible communication between sites, including the Bible, which mentions both signal fires and flags as part of the defensive system of Iron Age cities in Palestine (Isaiah 30:17), and the Lachish letters, which discuss the signal fires at Lachish and Azekah (Pritchard, 1969: 322). Checking line of sight visibility between archaeological sites has most often meant physically visiting sites to determine if neighboring sites could be seen. More difficult is calculating the viewshed for a site. In the same way that watersheds go beyond individual drainage channels, viewsheds go beyond individual points to include everything that can be seen from a given location. Viewshed maps can be created by careful interpretation of topographical maps and extensive field testing, but their construction is very time consuming and few projects have been interested in pursuing this line of investigation.
This reluctance is likely to change. GIS technology has significantly reduced the time necessary to create viewshed maps. In a GIS, creating a viewshed map is a relatively simple process involving a digital elevation model (DEM) and a map containing the point location, or locations from which the viewshed will be calculated. Using the DEM, the GIS draws a straight line from the viewpoint cell/s to each cell in the raster. It then checks the elevation of each cell that this line passes through in order to determine if any of the elevations are great enough to block the line of site between the viewpoint cell and the target cell (ESRI, 1995). If the target cell is visible the GIS places a "1" in that cell in the viewshed map, if it is not visible the GIS assigns a "0" to the cell. The result is a map of zeros and ones, with the ones designating the viewshed of the viewpoint/s. In addition, factors such as vegetation or the introduction of towers can be accounted for in order to more accurately model the real world (Figure 1).
One of the most obvious applications of visibility analysis to Near Eastern archaeological material is the establishment of viewsheds for fortified sites. Fortified against attack, these sites had a special interest in maintaining visible communcation within their immediate vicinity. Viewshed maps allow researchers to identify problem areas in the region enabling them to focus on likely solutions. As an example of this type of analysis, a viewshed map was created for Tall al-Umayri, and a system of wathctowers, calculated to enlarge the effective viewshed of the site, were proposed.
Located in the hills above the Madaba Plain, Tall al-Umayri is nearly surrounded by peaks and ridges. This is most notable to the west, where a long ridge, now the home of the Amman National Park, runs north and south. This ridge effectively blocks visibility from Umayri to the south, west, and north. In addition, there are a number of hills and ridges to the east, leaving the site with a limited view of its territory (Figure 2). Establishing the limits of visiblility at the site involved the calculation of a viewshed for Tall al-Umayri. In creating the viewshed map, it was estimated that the combined height of the city walls and an observer would be at least six meters and the elevation was set accordingly in the GIS. The Umayri viewshed can be seen in Figure 3, where the yellow represents what could be seen from the Tell, and the blue what could not be seen. Even knowing that the viewshed would be limited, it was surprising to find that only 8.5% of the Umayri survey region was visible from the Tell. Compare this with the other sites excavated by the Madaba Plains Project and this percentage looks even smaller. The viewshed for Tall Hesban contained 27% of its survey area (Figure 4), and Tall Jalul 44% of its region (Figure 5).
This restricted visibility would have been a serious problem for the inhabitants of Umayri. Since the site persisted in spite of this limitation, it must be assumed that the inhabitants were able to overcome the constraints placed on them by local topography. The most obvious strategy being a system of watchtowers, whose combined viewshed would allow Umayri to visibly control a greater percentage of the land surrounding the site.
Identifying the most likely candidates in a system of watchtowers is a task well suited to GIS analysis. The first step was to identify the main sites in the region. The size and strength of the fortifications at Tall Jawa and Umayri Survey Site 126 indicate that, along with Tall Umayri, they would have played important roles in any regional system. The second step was to identify likely watchtowers that could have been used to increase the viewing area of the three larger sites. The Umayri survey recorded 42 sites with rectilinear or circular structures. Although most of these sites were designated farmsteads by the survey, the structures could have functioned as watchtowers. Overlaying these 42 sites on the Umayri viewshed identified three sites (Sites 2, 40, and 60) that also had visual communication with Tall Umayri. Finally, although not in the Umayri viewshed, Site 54 was included because its location on the ridge between Jawa to east and Site 126 to the west would have provided a communication link between these two sites. In Figure 6, the position of the proposed watchtowers can be seen in relation to each other and to the Umayri viewshed.
Of these proposed watchtowers, the most important to the inhabitants of Umayri would have been Site 2, the tower atop the ridge just west of the Tell. This site was within shouting distance of Umayri's walls, and provided direct visual communication with each of the tower sites in the proposed system (Figure 7). An estimated combined height of the walls and the observer was set at 6 meters for Tall al-Umayri and Tall Jawa, at 5 meters for Site 126, and at four meters for the watchtower sites. The change in viewsheds when these sites were added was dramatic, increasing from 8.5% of the survey region for Umayri alone to over 57% for the combined system of watchtowers (Figure 8). If employed, this system of watchtowers and fortified sites would have provided Umayri with complete visual access to activities in the eastern part of the region, as well as an increased presence to the west. It should be noted that there were probably more watchtowers on the ridges to the west of Umayri, but because they were not found by the survey they could not be included in this study. In future field seasons, a new examination of these ridges, using the viewshed of Site 2 as a guide, will be undertaken in an effort to locate towers. Their inclusion in this model would increase Umayri's visual control of the area to the west.
Less obvious than a system of defensive towers, intervisibility may have been important for small hinterland sites. It can be assumed that communication between hinterland sites in the region would have been important for a number of reasons, including economic concerns, safety concerns, and social interaction, but the visual component of this communication has seldom, if ever, been examined on a regional basis in the Near East. Again, the principal reason for this oversight can be found in the high cost, principally in time, of carrying out the research. Ground testing the importance of visual communication between small sites in a region like Umayri would be an extremely difficult and tedious task. To further complicate the procedure, statistical significance of any perceived relationship would necessitate the collection of a comparative set of data. Without this comparative data, the researcher would never be certain whether a perceived relationship was real, or simply a product of the local landscape.
GIS based visibility analysis provides a natural course to examine this problem. A GIS can rapidly determine how many archaeological sites can be seen from any point in the study region, as well as provide a comparative sample for statistical testing. Determining the importance of intervisibility in the Umayri region was accomplished in three steps: First, the creation of a cumulative viewshed; second, the querying of this viewshed by the sites used in its creation, and by a sample of random point locations; and third, a statistical test to determine the significance of any differences between samples.
Step One: As their name suggests, cumulative viewsheds are constructed from a number of individual viewsheds (ESRI, 1995; Wheatley, 1995) . The first task in creating cumulative viewsheds for the Umayri region was to create viewshed maps for each Early Bronze Age, Iron Age I, Iron Age II, Byzantine, and Umayyad site in the Umayri region. The only difference between the creation of these viewshed maps and those discussed earlier was the decision to leave the height of towers and walls out of the viewshed calculation. Viewpoints were calculated at 1.5 meters above the surface, approximately eye-level for someone walking the fields around their house or tent.
Summing all the viewsheds from each period created cumulative viewsheds for the Early Bronze, Iron Age I, Iron Age II, Byzantine, and Umayyad sites. Since viewshed maps are matrices of zeros and ones, summing a series of geo-referenced viewsheds on a cell-by-cell basis has the effect of increasing the value, by one, for each cell, whenever that particular cell is in a viewshed (Figure 9). For example, if the 41st cell in each of eight viewshed maps was in a viewshed 5 times, then the value in the 41st cell of the cumulative viewshed would be 5. Using this simple equation to construct a cumulative viewshed, the value in each cell would correspond to the number of sites that could be seen from that cell. An example of the maps produced can be seen in Figure 10, where each color represents a different number of sites visible during Iron Age II. (Note that although each step in the process of creating a cumulative viewshed is described separately here, the sofware used in the analysis, ARC/INFO 7.0.4, completed them as a single procedure that was transparent to the user (ESRI, 1995)).
Step Two: Site intervisibility was revealed by querying the cumulative viewshed map for the values contained in the cells which corresponded to the location of the sites. This produced an ASCII file containing the number of sites visible from each site during a given archaeological period. A comparative sample was obtained by having the GIS create a map of 10,000 randomly located points. These points were all within the Umayri survey region, representing a 5% sample of this area. The cumulative viewsheds for the different periods were queried by the random sample, providing the comparative data necessary for testing the statistical significance of site intervisibility in the Umayri region.
Step Three: Each pair of queried samples were subjected to a
Kolmogorov-Smirnov test to determine if the differences between them
were significant. The Kolmogorov-Smirnov is a non-parametric
statistical test that measures the difference between the cumulative
proportions of two samples (Kvamme, 1990). In
Figure 11, cumulative proportions of
intervisibility for both EB site sample and the random sample are
displayed. The Kolmogorov-Smirnov test locates the point at which the
two samples are farthest apart. In this particular case that occurs
at zero. It then subtracts the lesser proportion from the greater
proportion and returns this number as the test statistic.
Significance at the five percent level is reached if the difference
between the samples is greater than , where n is the site sample. Because
the square root of n serves as the denominator in this
equation, sample size greatly affects the point at which the
difference between samples becomes significant. A large sample size
lessens the difference necessary for significance to be reached,
while a small sample increases it. With a small sample size of 18 for
sites with Early Bronze Age pottery (Figure
11), the difference must be greater than 0.3206 for significance
at the 5% level. In fact the difference between these samples was
0.3684, indicating that intervisibility was a significant factor in
site location strategies during the Early Bronze Age. The graph also
provides details about the type of intervisibility preferred by the
Early Bronze inhabitants of the Umayri region. Because the cumulative
proportion of sites is to the right of the random sample, it
indicates that they preferred values higher than would be expected if
the site selection process were random. That is, they wanted to see
more of their neighbors and were locating their sites accordingly.
Sites from each period were subjected to this test, yielding the graphs in Figure 12. In each case the site sample lies to the right of the random sample, making it clear that inhabitants from all periods preferred a higher level of intervisibility than was found in the random sample. Significance at the 5% level was achieved by the Iron II, Byzantine, and Umayyad samples, with only the Iron I sample failing to achieve this level of significance (Table 1). This failure is most likely a function of a small sample size. With only 16 sites, this was the smallest sample used in the analysis and therefore required the largest difference between proportions to reach significance. Taken together, these results indicate that visual communication was a significant factor for ancient humans when deciding where to locate their sites.
Table 1: Kolmogorov/Smirnov Test Statistics for Umayri
Region Intervisibility
Period |
Sample Size |
Test Statistic |
Probability |
---|---|---|---|
Early Bronze Age |
18 |
0.3684 |
0.008 |
Iron Age I |
16 |
0.1937 |
0.479 |
Iron Age II |
101 |
0.2557 |
0.000 |
Byzantine |
110 |
0.2456 |
0.000 |
Umayyad |
43 |
0.2617 |
0.003 |
In summary, there are a number of interesting points raised by this study. First, GIS makes viewshed analysis realistically possible. Although it is always better to carry out research in the real world, the difficulties of creating and field checking intersite visibility make the virtual world of digital viewsheds an attractive alternative. Interesting phenomena or special problems raised by the GIS analysis can always be field checked after the fact. Second, the proposed system of watchtowers in the Umayri region suggests that the ancient inhabitants took an active role in modifying the circumstances of their environment. Although Tall al-Umayri suffered from a restricted viewshed, the judicious placement of watchtowers would have allowed the inhabitants of Umayri to increase the size of their viewshed and thereby their control of the region. Finally, intervisibility should be considered when looking for settlement strategies in hinterland regions. The data presented here indicate a clear preference for high intersite visibility during most archaeological periods. This preference indicates, once again, that the ancient inhabitants of Umayri were not passive elements in a static system, but took an active role in modifying the circumstances of their environment.
Boaz, J. S., & Uleberg, E. (1995). The Potential of GIS-Based Studies of Iron Age Cultural Landscapes in Eastern Norway. In G. Lock & Z. Stancic (Eds.), Archaeology and Geographical Information Systems: A European Perspective. (pp. 249-259). London: Taylor & Francis.
Boling, R. G. (1989). Site Survey in the el-Umeiri Region. In L. T. Geraty, L. G. Herr, Ø. S. LaBianca, & R. W. Younker (Eds.), Madaba Plains Project 1: The 1984 Season at Tell el-Umeiri and Vicinity and Subsequent Studies. (pp. 98-188). Berrien Springs, Michigan: Andrews University/Institute of Archaeology.
Burrough, P. A. (1986). Principles of Geographical Information Systems for Land Resource Assessment. Oxford: Clarendon Press.
Christopherson, G. L. (1994). In Pursuit of the Long Durée: Environmental Models and Cycles of Intensification and Abatement from the Umeiri Survey. Paper presented at the Annual Meetings of the American Schools of Oriental Research. Chicago.
Christopherson, G. L. (1996). Computer Mapping. In E. M. Meyers (Ed.), Encyclopedia of Near Eastern Archaeology. New York: Oxford University Press.
Christopherson, G. L. (forthcoming, 1997). Madaba Plains Project: Regional Site Survey. In L. T. Geraty, L. G. Herr, Ø. S. LaBianca, & R. W. Younker (Eds.), Madaba Plains Project 3: The 1989 Season at Tell el-Umeiri and Vicinity and Subsequent Studies. Berrien Springs, Michigan: Andrews University/Institute of Archaeology.
Christopherson, G. L., & Dabrowski, B. (forthcoming 1998). Using a Geographic Information System to Create Probability Models for Sites with Tombs from the Umeiri Regional Survey. Studies in the History and Archaeology of Jordan VI. Amman, Jordan: Department of Antiquities of Jordan.
Christopherson, G. L., Guertin, D. P., & Borstad, K. A. (1996). GIS and Archaeology: Using ARC/INFO to Increase our Understanding of Ancient Jordan. Paper presented at the Sixteenth Annual ESRI User Conference. Palm Springs, CA: Environmental Systems Research Institute. http://www.esri.com/resources/userconf/proc96/TO150/PAP119/P119
D. Survey (1958a). Amman. Amman, Jordan: War Office and Air Ministry.
D. Survey (1958b). Naur. Amman, Jordan: War Office and Air Ministry.
ESRI (1995). ARC/INFO. Redlands, CA: Environmental Systems Research Institute.
Gaffney, V., & van Leusen, M. (1995). Postscript-- GIS, Environmental Determinism and Archaeology: A Parallel Text. In G. Lock & Z. Stancic (Eds.), Archaeology and Geographical Information Systems: A European Perspective. (pp. 367-381). London: Taylor & Francis.
Gaffney, V., & Stancic, Z. (1991). GIS Approaches to Regional Analysis: A Case Study of the Island of Hvar. Ljubljana: Znanstveni institut Filozofske fakultete.
Gaffney, V., Stancic, Z., & Watson, H. (1995). The Impact of GIS on Archaeology: A Personal Perspective. In G. Lock & Z. Stancic (Eds.), Archaeology and Geographical Information Systems: A European Perspective. (pp. 211-229). London: Taylor & Francis.
Hutchinson, M. F. (1989). A New Procedure for Gridding Elevation and Stream Line Data with Automatic Removal of Spurious Pits. Journal of Hydrology, 106, 211-232.
Kvamme, K. L. (1989). Geographic Information Systems in Regional Archaeological Research and Data Management. In M. B. Schiffer (Ed.), Archaeological Method and Theory. (pp. 139-203). Tucson: University of Arizona Press.
Kvamme, K. L. (1990). One-Sample Tests in Regional Archaeological Analysis: New Possibilities Through Computer Technology. American Antiquity, 55(2), 367-381.
Kvamme, K. L. (1992). Geographic Information Systems and Archaeology. In G. Lock & J. Moffett (Eds.), Computer Applications and Quantitative Methods in Archaeology 1991. (pp. 77-84). Oxford: TEMPVS REPARATVM Archaeological and Historical Associates Limited.
Microsoft Corporation (1993). Microsoft FoxPro for Apple Macintosh. Redmond, WA: Microsoft Corporation.
Pritchard, J. B. (Ed.). (1969). Ancient Near Eastern Texts Relating to the Old Testament (Third ed.). Princeton, New Jersey: Princeton University Press.
Stead, S. (1995). Humans and PETS in Space. In G. Lock & Z. Stancic (Eds.), Archaeology and Geographical Information Systems: A European Perspective. (pp. 313-317). London: Taylor & Francis.
Wheatley, D. (1992). Going Over Old Ground: GIS Archaeological Theory and the Act of Perception. In J. Andresen, T. Madsen, & I. Scollar (Ed.), Computing the Past: Computer Applications and Quantitative Methods in Archaeology. (pp. 133-138). University of Aarhaus, Denmark: Aarhaus University Press.
Wheatley, D. (1995). Cumulative Viewshed Analysis: A GIS-Based Method for Investigating Intervisibility, and its Archaeological Application. In G. Lock & Z. Stancic (Eds.), Archaeology and Geographical Information Systems: A European Perspective. (pp. 171-185). London: Taylor & Francis.
Younker, R. W. (1991). The Judgment Survey. In L. T. Geraty, L. G. Herr, Ø. S. LaBianca, & R. W. Younker (Eds.), Madaba Plains Project 2: The 1987 Season at Tell el-Umeiri and Vicinity and Subsequent Studies. (pp. 269-334). Berrien Springs, Michigan: Andrews University/Institute of Archaeology.
The authors would like to thank the Madaba Plains Project for the archaeological data used in this paper, and for financial support of the field research, and the Advanced Resource Technology Program at the University of Arizona for access to ARC/INFO geographic information systems technology and expertise, and financial assistance in attending the meeting where the paper was presented.
Gary L. Christopherson is a Ph.D. candidate in the Near East Studies Department, and a Research Assistant in the Advanced Resource Technology Group at The University of Arizona. He has been associated with the Madaba Plains Project since 1987, where he is a Field Director in charge of archaeological survey. Correspondence may be sent to: ART Group (SRNR), 203 Biological Sciences East, The University of Arizona, Tucson, AZ, 85721, phone (520) 621-3045, or by email: Gary Christopherson -- garych@nexus.srnr.arizona.edu
D. Phillip Guertin is an Associate Professor within the School of Renewable Natural Resources, Advanced Resource Technology Group, The University of Arizona. Correspondence may be sent to: ART Group (SRNR), 203 Biological Sciences East, The University of Arizona, Tucson, AZ, 85721, phone (520) 621-1723, or by email: Dr. Phil Guertin -- phil@nexus.srnr.arizona.edu