GPS and other geospatial data recorded and used during the WHPS (including topographic data and high-resolution satellite imagery) were input into a Geographic Information System (GIS). GIS is rather broadly understood as a group of computer-based applications that not only allow input data to be spatially fixed according to their geographic coordinates (as in traditional paper mapping), but more importantly allow that data to be described, queried, manipulated, and visualized in a multitude of ways. A GIS stores input data not as a single digital map, but rather as discrete layers of information that can be viewed separately or in combination, depending on what spatial relationships between and among features the user is interested in discerning. And for every piece of data input into a GIS, the user can create any number of variables (or “attributes”) to describe the data.
Using the basic query and analysis tools of the WHPS’s GIS, Glenn Corbett’s 2010 dissertation aimed to provide a better understanding of how Hismaic inscriptions and rock drawings are distributed within the landscape. The study focused on identifying distribution patterns among the carvings recorded from the Ḥafīr’s three main tributaries: Wādī Khāyneh, Tel‘at Rashid, and Wādī aṭ-Ṭfeif, as well as the cascade pools of Muqawwar. In all, the locations of 740 inscriptions and 95 signed rock drawings (recorded from 200 individual stones and rock faces) were mapped and analyzed. Quadrat analysis was used to detect areas of site and inscription clustering, hydrological modeling was used to better understand the runoff and drainage patterns of the tributaries relative to those clusters, while attribute analysis was used to study spatial patterning among the varied contents of the inscriptions and drawings.
Since the location of the inscriptions was recorded as GPS point data within the WHPS’s GIS, a useful means of studying their spatial distribution was quadrat analysis. In quadrat analysis, a grid of uniform quadrats (or cells) is draped over the extent of the survey area and the points that fall within each cell are counted. These cell counts provide a straight forward means of both measuring the density of points within a defined area and also comparing that density with other cell counts in the survey area. A GIS can graphically summarize these variations in point density and thereby provide a general impression of where inscriptions tend to cluster or concentrate. Quadrat analysis also provides a statistical measure of the degree of clustering, randomness, or dispersion within a particular point pattern.
Quadrat analyses of the study areas revealed that even if the overall distribution of inscriptions within a particular wadi is more or less dispersed, there do tend to be specific locations that show far more inscription and drawing activity per unit area (see figure). The quadrat analysis for Tel‘at Rashid, for example, shows that nearly two thirds of the recorded inscriptions were found near the confluence of the wadi’s south and southeast branches, across an area measuring less than six hectares. Likewise, in Wādī Ḫāyneh, the quadrat map reveals that just over a third of the recorded inscriptions were found along a narrow three-hectare stretch located at the far end of the wadi’s southern branch. In Wādī aṭ-Ṭfeif, just less than half of the wadi’s recorded inscriptions were found among seven identified clusters of various sizes (from 0.40 hectares to 1.22 hectares), nearly all occurring at bends along the wadi’s sinuous path. And perhaps most telling, more than a third of the inscriptions recorded in aṭ-Ṭfeif were found in and around the cascade pools of Muqawwar, across an area measuring less than a single hectare.
These distributions clearly indicate the Ḥafīr’s authors and artists were actively seeking out certain locations within these tributaries. But why? Hydrological modeling of the wadi’s drainage catchments provides the answer.
Figure: Quadrat clusters identified for each study area (darker quadrats = peak inscription densities)
Digital Elevation Models (or DEMs) of the Ḥafīr’s three major tributaries were created by digitizing the contour lines, spot elevations, and drainage paths of each tributary into the WHPS’s GIS. To generate a DEM, a GIS uses various interpolation algorithms to convert these discrete lines and points of elevation data into a continuous grid of elevation values spread across an area. A hydrologically-correct DEM (meaning a DEM where every grid cell has at least one neighboring cell with a lower elevation) allows for the creation of a number of mathematically derived grids that simulate the unimpeded flow of water across a terrain according to its path of steepest descent. These grids then allow one to model drainage paths and flow direction across the modeled terrain, and from there, the accumulated and subdivided catchments or watersheds for each tributary.
The hydrological analyses revealed that almost every major inscriptional cluster within each wadi was situated at or in close proximity to the drainage point of a significant watershed (see figure). In Tel‘at Rashid, for example, over half of the wadi’s 3 km2 total drainage flows through or past the four quadrats of highest inscription density. Likewise, the three quadrats located at the far end of Wādī Ḫāyneh’s south branch receive nearly a third of the wadi’s 10 km2 total catchment. The cascades pools of Muqawwar, where the highest number of inscriptions were found, receive direct runoff from a catchment area of more than 3.0 km2, with much of that drainage originating from the well-watered Rās an-Naqab escarpment. Finally, the outlet of the extremely narrow and sinuous Wādī aṭ-Ṭfeif receives runoff from a total catchment area of 13 km2, the largest catchment of all the Ḥafīr’s tributaries.
The combined results of the quadrat and hydrological analysis thus present a clear picture of why Hismaic inscriptions and drawings tend to cluster in certain areas. The area’s tribesmen were intimately acquainted with the Ḥafīr’s drainage patterns and knew precisely where waters from seasonal downpours would tend to drain, collect, and pool. They and their families found their way to these spots to fill their water skins, water and pasture their camels and flocks, stalk wild game, and, while there, carve their names and messages into stone.
Figure: Modeled drainages and catchments of each study area (with quadrats of peak inscription density shaded yellow)
In the WHPS’s GIS database, attribute fields were created to describe every inscription and/or drawing that was identified on a particular stone or rock face. The various attribute fields were chosen to reflect the full range of the content and structure of the inscriptions, as well as the various categories or types of drawings that could be discerned (see figure 1). Thus, basic descriptive fields were created for “Author” and any identified “Ancestors,” “Inscription Type” (i.e., simple authorship, drawing signature, prayer/curse, emotive) and “Drawing Type” (i.e., camel, hunt, horse), but also a host of other categories that could be used, as necessary, to further subdivide the content of the inscriptions and drawings (i.e., “Prayer Type,” “Emotive Type,” “Hunt Type”). Once entered, this information was then queried and sorted both thematically and spatially to look for possible relationships among the inscriptions and drawings from specific locations. Particular attention was given to analyzing and mapping the distribution of inscription types, drawing types, author names, tribal names, as well as specific members of identified families.
One of the most significant conclusions to be drawn from the content analysis was that few if any locales show obvious signs of having been exclusively reserved for use by a single family or tribe. At every level of spatial analysis, the inscriptions show an incredible diversity of both personal and tribal names. Furthermore, the signatures and drawings of members of identified families, with a handful of possible exceptions, are generally not confined to particular locales but rather are found in almost every part of the study area (see figure 2). The available epigraphic evidence would thus suggest that areas of concentrated drainage, wadi confluences, and naturally-occurring water sources like Muqawwar were not reserved for individual families but rather were common resources available to all who camped, pastured, and hunted in the area.
Perhaps equally significant is that neither prayer nor emotive texts seem to have been carved in specific locales; rather, they appear just as broadly and as widely as other attested expressions. The scattered, non-clustered distribution of these texts shows that authors did not restrict the carving of these often quite personal messages to specific places within the landscape. More important, the fact that these texts do not consistently cluster around specific locales, such as natural pools or prominent boulders, shows that prayers and emotives were rarely used to visibly demarcate a specific place in the landscape as sacred or socially significant.
Finally there is little evidence that location had any bearing on whether artists chose to draw camels, hunted animals, or something else entirely. Although some noticeable spatial patterning does exist in select areas, the overall distribution pattern of each drawing type shows that both motifs regularly occur within the same wadi and within even the same quadrat cluster. Simply put, there is no clear and consistent distinction between where camel drawings are found and where hunt scenes are found. This overall heterogeneous distribution, therefore, suggests that neither the camel nor the hunt motif was used by carvers to set off particular locales as functionally, ritually, or symbolically unique or different from other areas.
Figure 1: Numerous attribute fields describe each inscription Figure 2: Distribution of members of identified families by study
within the GIS area