Interferometric SAR and Machine Learning: Using Open Source Data to Detect Archaeological Looting and Destruction

Archaeological heritage in the Near East is under an ever increasing threat from multiple vectors such as looting and systematic destruction, militarization, and uncontrolled urban expansion in the absence of governmental control among others. Physically monitoring endangered sites proves to be infeasible due to the dangerous ground conditions on the one hand, and the vast area of land on which they are dispersed. In recent years, the abundant availability of Very High Resolution (VHR) imaging satellites with short revisit times meant that it was possible to monitor a large portion of these sites from space. However, such images are relatively expensive and beyond the means of many researchers and concerned local authorities. In this paper, I present an approach that uses open source data from two of the European Space Agency’s (ESA) Copernicus Constellation, Sentinel-1 and Sentinel-2 in order to generate disturbance patches, from which looting and destruction areas are classified using Machine Learning. Such an approach opens the door towards sustainable monitoring over large swaths of land over long periods of time.