This paper will present recent advances in a remote sensing workflow developed by the Endangered Archaeology in the Middle East and North Africa (EAMENA) project using machine learning algorithms for automating the detection of threats to archaeological sites. The automated change detection (ACD) system uses free satellite imagery and high-performance computing power available via Google Earth Engine to compare multiple Sentinel-2 images to highlight areas of change and identify where, when, and what types of change have occurred within the proximity of known archaeological sites. Previous versions of EAMENA’s ACD methodology produce a binary classification of change/no change. The updated workflow uses machine learning classifiers (Random Forest) to carry out supervised classification of land cover in the study area and analyses time series of satellite images to classify changes in land use and categorise the type of change which has occurred. By analysing sequential classified images, we can identify where archaeological sites are within proximity to significant land cover changes and at risk of being affected by these changes. This workflow can improve efficiency in monitoring archaeological sites by rapidly identifying which sites are most at risk and helping heritage professionals to target resources where they are needed most urgently. Case studies undertaken in North Africa will be presented which demonstrate the workflow and outputs, as well as the advantages and challenges that this methodology presents.

Join online: https://zoom.us/j/94826858436?pwd=N1JCeU9Id3dUdkJ4ZDVyUjhtQ0FMQT09