Investigations of non-visible features in archaeological sites: testing aerial remote sensing with UAV in Pompeii

The study described below aims to confirm the potential of UAV-based multispectral imagery as a flexible and cost-effective tool to detect possible buried archaeological structures, expanding upon previous approaches based on satellite or traditional airborne data. In parallel, the authors investigate the role of such imagery within a conjectured workflow that incorporates multispectral analysis as a preliminary, extensive, and non-invasive step in archaeological prospection strategies. The study evaluates the performance of a commercial sensor and analyses spectral signatures by generating index maps within the significant context of Iulia Felix Praedia in Pompeii (Italy). A significant opportunity was the possibility of acquiring multispectral data in the hortus area, previously investigated through non-invasive geophysical surveys and archaeological excavations. The UAV photogrammetric flight, as well as the subsequent analyses, focused on the visual interpretation and geolocated examination of vegetation and soil index maps, accurately selected among those available, considering the UAV-acquired band dataset. This approach enhanced the features of the complex hortus environment, where natural elements alternate with numerous man-made structures. These analyses led to the detection of anomalies consistent with those previously identified by the aforementioned investigations, alongside additional anomalies distributed across the study area. The detected anomalies were further analysed and synthesised; this involved generating a confidence map based on the frequency of anomaly occurrence across the analysed index maps. The consistency between detected anomalies and previous investigations’ results underlines the potential for continued research on processing multispectral data captured by UAVs. Thanks to such data, a valuable alternative to satellite imagery was provided due to their much higher spatial resolution, enabling rapid and cost-effective campaigns to plan more targeted geophysical and archaeological investigations. The findings also validate the hypothesised workflow involving the use of multispectral imagery as a preliminary, extensive, and non-invasive tool to define excavation areas’ perimeters and, subsequently, guide targeted analyses.