What defines a “site” in geotechnical engineering?: A comparative study between local and global big indirect databases

Abstract

Geotechnical engineering continuously faces the challenge of accurately assessing a wide array of physical properties from limited soil data. To solve this problem using a data-driven approach, this study introduced an innovative local big indirect database (Local-BID) based on extensive sampling of Tokyo seabed clay and compared it with a comprehensive global big indirect database (Global-BID) compiling data from numerous international sites. The comparison revealed that data variability in Local-BID and Global-BID was almost identical, thereby highlighting fundamental issues such as the definition of “site” and the concept of parameter variability in geotechnical engineering. Through meticulous analysis, we explored the influences of database selection, quantity of observation points, and estimation indicators on the outcomes of hierarchical Bayesian estimation. Interestingly, while the medians of our estimates remained stable across varying conditions, the variance was strongly affected by these factors. Moreover, we critically reassessed the “site” definition used in hierarchical Bayesian models, which typically groups sites based on planar proximity without considering vertical stratification. Although this definition can fundamentally function, our findings suggested that incorporating stratigraphic classifications to define subsites could enhance site characteristic assessments and aid in understanding site similarity through a detailed analysis of soil composition variations.