2025 Volume 81 Issue 15 Article ID: 24-15018
Geotechnical surveys are conducted to understand the spatial distribution of soil properties. However, due to cost and time constraints, these surveys are often limited and may not provide sufficient data. Therefore, it is preferable to effectively utilize indirect data, such as construction data, to reasonably estimate the spatial distribution of soil properties. In this paper, we propose a Gaussian process regression method that considers the cross-correlation with indirect data. The direct data used in numerical examples are SPT values from the Standard Penetration Test (SPT), and the indirect data are torque values obtained during pile construction. The proposed method is applied to both simulated data and actual measurement data. The estimation accuracy of the proposed method is validated through cross-validation, demonstrating that a more accurate spatial distribution can be estimated by utilizing indirect data.
