Revealing lichen-covered rock art at stonehenge
: where terahertz imaging and photogrammetry meet machine learning, lichen simulation, and image analysis algorithms

Student thesis: Doctoral Thesis


Stonehenge is a World Heritage Site that features the largest panels of Early Bronze Age axe-head carvings in the world. Archaeologists use these carvings to better understand the significance of Stonehenge for people over its history. 23% of the stone surfaces at Stonehenge however are obscured by Ramalina siliquosa, a shrubby fruticose lichen, which is a symbiosis between algae and fungi. This renders almost a quarter of the stone surfaces of Stonehenge a mystery. Given the potential for further rock carving discoveries, this thesis presents the first detailed study on methods for revealing lichen-covered stone surfaces. We approached this problem from two different directions, the first was via surface imaging methods and the second was via subsurface imaging methods.

For the surface imaging approach, five different carving enhancement methods for revealing lichen-obscured carvings were firstly tasked with correctly labelling non-lichen covered carvings. The best-performing method was the machine learning-based MeshNet, which achieved 91.0% accuracy on labelling samples of surfaces at Stonehenge with and without carvings. A new method for simulating the growth of Ramalina siliquosa was then developed and applied to the same samples of surfaces at Stonehenge, on which MeshNet achieved 73.4% accuracy. With further development, this method can be used by archaeologists to gauge the likelihood of carvings beneath lichen-covered surfaces at Stonehenge.

A subsurface imaging approach can then be used to verify the presence of carvings in high likelihood areas identified by MeshNet. Terahertz time-domain spectroscopy (THz-TDS) was identified via literature review as the most promising non-destructive subsurface imaging modality. However, in the absence of any previous applications of THz-TDS on lichen, we firstly characterised the limits on THz-TDS for lichen. Through laboratory tests, 18% water content of the dry lichen weight and 2.6 mm lichen thickness were identified as the maximum bounds within which the topography of the lichen substrate could be revealed. Small regions of Ramalina siliquosa-covered surfaces of Stonehenge were then imaged with THz-TDS with partial success in revealing the underlying rock surface. For areas of the rock surface with even elevation and short lichen THz-TDS was successful; for uneven surfaces and tall lichen, the limits of our shallow focal length lens in our imaging setup prevented successful imaging of the substrate. With improvements to the imaging setup, archaeologists can implement the surface imaging workflow for narrowing down carving candidate sites and then subsurface imaging for verifying any prehistoric carvings hidden by lichen.
Date of AwardJul 2024
Original languageEnglish
Awarding Institution
  • University of Brighton
SupervisorMatthew Brolly (Supervisor), David Nash (Supervisor) & Chris Carey (Supervisor)

Cite this