Work Package 1

The southwestern parts of the North Sea are the last remains of a vast landmass that connected modern continental Europe with the British Isles, making it an important gateway for animals and humans alike.

Climatic and sea level change at the beginning of the Holocene cut off this connection around 8500 BP, landscape change with potentially dramatic repercussions. A complex interaction between developments in ground water rise, coastal erosion, animal habitats (terrestrial and marine), and human ecodynamics has resulted in multiple narratives and unresolved hypotheses regarding the short- and long-term impacts of sea level rise on the inhabitants of that landscape.

The archaeology of this submerged landscape is hampered by deep water, thick modern sediment deposition and lack of visibility of the terrestrial landscape. A wealth of archaeological evidence has been recovered from this environment, but none of it provides a contextual setting that could enable us to locate a focus of archaeological interest. At present we are only able to define the potential for archaeological exploration, not an expectation that we will find it.

By utilizing artificial intelligence derived from WP5 this work package will set out to explore the outputs of autonomous geophysical interpretation that can classify archaeological indicators, identify archaeological surfaces, and assess their accessibility to investigation.

This WP will look to identify organic deposits, such as peat. These are known from existing shallow-water studies to be an important proxy for the presence of archaeological material and represent a highly efficient focus for coring, dredging and robotic survey.

The work package will also utilise data from the Flevoland Polder (FP) project to model and ground truth marine survey design (36). The FP project has produced a large archive of geostratified sediment-cores containing palaeoenvironmental and archaeological indicators (artefactual, palaeobiological and geochemical) in peat and clay deposits of a deeply-buried early Holocene landscape on the Dutch coast that was progressively inundated and waterlogged by land subsidence and relative sea-level rise before modern drainage and land reclamation We will use these data as training sets for machine learning, to characterise and model archaeological indicators and landscape attributes of cultural activity for use further offshore and in deeper water- We will test these models by ground-truthing them against comparable data on shallow-water palaeoenvironmental conditions and known underwater archaeological sites in WPs 3 and 4. This combined approach will facilitate experimentation in the design of deep-water surveys, and the identification of specific targets for directed coring, dredging, robotic survey and the recovery of comparable proxies of cultural activity.

WP1 will provide a new, high-resolution, 3D framework of the submerged landscape to support future exploration, facilitated through behavioural modelling and AI/Machine Learning processes. Replicable and ground-truthed, the process can be implemented across the areas of the European coastal shelf which have previously defied systematic exploration and in comparable landscapes globally.