The delineation of subsurface stratigraphy is an indispensable task in geotechnical site characterization, whichbenefits subsequent geotechnical design and construction. In recent years, three-dimensional (3D) subsurfacegeological modelling has received increasing attention because it can realistically represent subsurfacestratigraphic variations at a specific site. Geotechnical site investigation data (e.g., boreholes) are necessary fordeveloping 3D subsurface geological models. Due to time, budget, or technical constraints, measurements arecommonly limited and sparse in engineering practice, leading to a long-lasting challenge in properly developing3D subsurface geological models. To tackle this challenge, a generative and data-driven machine learning methodcalled multi-scale generative adversarial network (MS-GAN) is adopted in this study. This method can properlylearn 3D stratigraphic information from a 3D training image representing prior geological knowledge forautomatically generating 3D subsurface geological models conditioned on limited boreholes in a data-drivenmanner. The performance of this method is illustrated using a simulated data example. The results indicate thatMS-GAN not only accurately and effectively delineates stratigraphic variations in 3D space, but also quantifiesstratigraphic uncertainty from multiple 3D realizations.
Lyu et al. (Wed,) studied this question.