Tell spotting - surveying near eastern settlement mounds from space

Digital elevation models (DEMs) have become a recognized tool in the visualization of his torical sites. Applications range through various scales from 3D presentations of the settings of individual buildings up to detailed GIS analysis of the hinterlands of archaeological sites. However, the development of space-borne instruments has seen the rise of high resolution DEMs even on the global scale. In the present study we demonstrate the use of such data sets, namely from the Shuttle Radar Topography Mission (SRTM), for a virtual survey on archaeological sites -the detection and localization of ancient settlement mounds in the Near East. These so-called "tells" (the Arabic word for a settlement-mound) are the result of millennia of settlement activity and visible landmarks of the first human urbanism in Mesopotamia. Though a large number of the sites are very well studied, the best current listings of them on a region-wide basis are neither comprehensive (circa 1400 major sites) nor accurate (to 10km grid-square). In order to obtain such a catalogue and to map these cultural monuments for the purpose of conservation and landscape archaeology, we set up a tool to process the SRTM data of this wide geographical region to be as automated as possible. Archaeological sites from a north Syrian plain provided the training and testing data for the design of a machine learning decision rule, which is able to show robustness against background variations and to control the number of false detection events . Landsat ETM+ image data is used to confirm sites proposed by this classifier and to allow a manual positioning of tell sites well below the DEM resolution. In a first application of the classification algorithm to a region with known settlement mound positions --identified in a prior study on high resolution CORONA satellite images -we were able to detect nearly all settlement higher than 5-6m which are of characteristic shape (85 out of 133), at a reasonable number of false positives. Present applications of the proposed method are the detection of tells in areas not yet studied, and the mapping of regions with a direct threat of these historical monuments by agricultural land-use. Archaelogical Gis, SRTM, DEM, Digital Survey, Archaeological Remote Sensing