Revealing Hidden Architecture: UAV LIDAR and Ground Classification at the Ancient City of Kadyanda

This study reports on a 2023 UAV-borne full-waveform LIDAR campaign carried out at the ancient mountain city of Kadyanda (near Yesiluzumlu, Fethiye-Türkiye), one of the most important settlements of northwest Lycia. Persistent dense Mediterranean forest and rugged topography have historically impeded comprehensive topographic mapping of the roughly 10 hectare fortified site, leaving legacy plans from the 1980s incomplete. A DJI Matrice 200 platform carrying a Surestar R-Fans-16 LIDAR sensor was utilized to acquire high-density 3D point clouds. Kinematic GNSS post-processing using RINEX baseline data yielded a total of 80,555,945 points. Robust ground classification was performed in the R statistical environment using Cloth Simulation Filtering (CSF; Zhang et al. 2016) that was calibrated across adaptive height and window-size parameters to separate vegetation from bare-earth returns. The resultant digital terrain model (DTM) achieved centimetre-level resolution and exposed previously unrecorded architectural remains under the forest canopy. When it was integrated into a GIS, the LIDAR derivatives enabled critical revision of earlier city plans, facilitated precise georeferencing of newly documented inscriptions, and supported ongoing systematic surface surveying. These results demonstrate that dense full-waveform UAV LIDAR data, coupled with advanced ground-classification algorithms, is a powerful solution for producing accurate digital baselines and revealing sub-vegetation archaeology in the complex mountainous landscapes of the Mediterranean region.