Bootstrap-enhanced BUS method for probabilistic characterisation of correlated geotechnical parameters with sparse data

Accurately inferring the joint distributions of geotechnical parameters is essential for geometric structure modelling and reliability assessment of geotechnical structures. In engineering practice, only sparse data can be acquired for a specific site, posing a challenge in modelling the joint distributions of geotechnical parameters. To address this challenge, this study proposes a Bootstrap-enhanced Bayesian Updating with Structural reliability (BBUS) method for the probabilistic characterisation of correlated geotechnical parameters with sparse data. The proposed method employs the parametric bootstrap technique to reconstruct the likelihood function, enabling the efficient generation of robust posterior samples. These posterior samples are subsequently used to infer the posterior predictive distributions of correlated geotechnical parameters through random sampling. The effectiveness of the proposed method is validated through two case studies. The results demonstrate that the proposed method can accurately model the joint distributions of correlated geotechnical parameters with sparse data, yielding posterior predictive distributions that closely match the observed data. Notably, the proposed method is versatile enough to be extended to correlated non-normal parameters and high-dimensional parameter scenarios. These findings also confirm the potential of the proposed BBUS method for the probabilistic characterisation of correlated non-normal geotechnical parameters with sparse data.