Spatial variability characterization of soil properties in offshore wind farm based on the Bayesian theory and conditional co-simulation method

Spatial variability characterization of soil properties in offshore wind farm is essential for offshore engineering. Multi-source data fusion can reduce the uncertainty of characterization. However, existing methods cannot simulate geotechnical properties based on non-co-located multi-source data, and do not consider the effects of statistical uncertainty. To overcome these challenges, a conditional co-simulation method based on Bayesian theory is proposed. The Bayesian theory is first used to estimate the cross-variogram model based on non-co-located multi-source data. Then, the conditional co-simulation is used to generate realizations of spatial varied soil properties, which can characterize the spatial variability with consideration of statistical uncertainty. The proposed method is applied to an offshore wind farm to build the spatial variability model of unconfined compression strength (qu) by integrating data on qu and standard penetration test (SPT) N value. The results show that the proposed method can characterize the spatial variability of qu from non-co-located data on qu and N value, and statistical uncertainty is properly taken into account. In addition, it is shown that the uncertainties of the variogram models and conditional co-simulation results can be reduced when the prior distribution with more information and/or SPT data is used.