SUT’s Special Interest Groups – OSIG and ECOSIG – are joining forces with the Central Scotland Regional Group of the Geological Society for a hybrid technical event exploring the integration of geophysical and geotechnical data for offshore site characterisation. Featuring presentations from researchers at the University of Glasgow and the British Geological Survey, the event will highlight innovative data-driven approaches to soil classification, regional subsurface interpretation, and the use of legacy datasets to support offshore wind and infrastructure development in the North Sea.

PLEASE NOTE THIS IS HYBRID EVENT

Presentations:

A novel data-driven framework for probabilistic soil type classification using CPT data – Timo Zheng, University of Glasgow

Characterisation of the ground conditions at an offshore wind site is an inherently challenging and complex task. On top of this, there is a clear need for optimisation of site characterisation processes – the scale of the surveys mean inefficient scheduling can cost £Millions and/or delay the critical path, and imperfect characterisation can lead to overly conservative foundation sizing, which can derail a project’s business case.

One means of optimisation is using relatively cheaper in situ (e.g., CPT) data to predict engineering parameters which would typically be assessed using more expensive/time-consuming sample/laboratory tests. These empirical prediction models must be ‘learnt’ from data in the first instance, meaning their reliability is dependent on the quality, quantity, and applicability of the underlying dataset. Many such models are available in the literature; however, they are often associated with a number of key limitations, such as: (i) reliance on small and/or unrepresentative datasets; (ii) lack of transparency with regard to specification; (iii) providing deterministic outputs, which do not convey the uncertainties in prediction; and (iv) lack of a rigorous means for updating or recalibration.

­­This talk introduces a framework for leveraging huge quantities of existing data towards the development of improved CPT-based prediction models. A Bayesian soil type classification model is presented which aims to overcome traditional limitations by following a rigorous, data-driven philosophy and delivering probabilistic rather than deterministic predictions. Crucially, the model is designed to be adaptable and can be updated with respect to additional data, allowing for continual refinement and/or site-specific calibration. The general principles of the approach are discussed, considering how they can facilitate an industry-wide effort towards continual refinement of robust, data-driven, probabilistic correlations, leading to more confident predictions at new sites.

What Do 3,000 Geotechnical Logs Reveal About the Shallow Subsurface of the North Sea? A New Approach to Regional Interpretation – Duncan Stevens, The British Geological Survey

The shallow subsurface is critical to offshore infrastructure design, yet our understanding of its variability is often fragmented, with most geotechnical data used only at individual sites. At the same time, large volumes of legacy data now exist, offering an opportunity to rethink how we characterise ground conditions at regional scale. This talk explores a new, systematic approach to integrating over 2,900 geotechnical logs from the UK North Sea, including CPTs, boreholes, and cores. By standardising lithological descriptions and applying a sequence-based similarity framework, we quantify vertical variability within logs and group sites with comparable subsurface structure. The approach condenses complex, heterogeneous datasets into reproducible metrics and clusters that can be mapped and compared across the region. The results reveal coherent patterns in shallow subsurface conditions, demonstrating how legacy data can be repurposed into practical, screening-level tools to support offshore infrastructure planning, cable burial assessment, and early-stage decision-making.

Kindly supported by: