Novel lichen simulation and laser scan modelling to reveal lichen-covered carvings at Stonehenge

• 23% of surfaces at Stonehenge are lichen-covered, potentially obscuring carvings. • Lichen thickness distribution of Ramalina siliquosa is obtained via laser scanning. • Scans inform a lichen simulation for testing visibility of carvings through lichen. • MeshNet achieves 73.3% accuracy in lichen-covered carving recognition. • Carvings can be found beneath shrubby lichen without resorting to lichen removal. Stonehenge boasts the world’s largest collection of early Bronze Age axe-head carvings, offering vital clues to the significance of the monument. However, 23% of the stone surfaces are concealed by fruticose lichen, primarily Ramalina siliquosa , potentially obscuring undiscovered carvings. To address this, a novel lichen growth simulation (RDLA), informed by laser scans of R. siliquosa , was developed to model lichen growth on early Bronze Age carvings. Denoising and visualisation revealed that the presence of R. siliquosa did not hinder the identification of carvings. Subsequently, a machine learning method for 3-D shape classification (MeshNet) was trained on these simulated lichen-covered carvings. Despite a reduced accuracy (73.3%) compared to non-lichen-covered carvings (90.7%) on Stone 53, MeshNet demonstrates the feasibility of semi-automatic identification of carvings through lichen coverage. These findings offer the prospect of uncovering additional carvings at Stonehenge and other prehistoric sites without resorting to invasive lichen removal or subsurface imaging.