Abstract: Due to historical variations in stress and sedimentation, geotechnical materials exhibit strong spatial variability. This means that the distribution and properties of geotechnical types vary across different sites and even within different locations of the same site. Existing random field models struggle to accurately capture this complex spatial variability and often lack verification with in-situ test data. To address these limitations, this paper focuses on studying the spatial variabilities of marine clay parameters in the three Nordic countries using the open database 304dB developed by the Engineering Practice of Risk Assessment and Management TC-304 in the International Society of Soil Mechanics and Geotechnical Engineering (ISSMGE). The paper analyzes and compares the differences and similarities in statistical characteristic parameters of marine clay along depth variations in the three Nordic countries. It also investigates the physical mechanisms of mutual influence between various parameters and clarifies the spatial variability laws of each parameter. Through trend analysis, fluctuation analysis, and correlation analysis, the study reveals that the mean value, standard deviation, and fluctuation scale of marine clay parameters in the three Nordic countries are non-constant along the depth. The mean value increases linearly with depth, while the standard deviation shows an approximate quadratic function variation trend along the depth. The correlation function is dependent on distance and location, but does not converge uniformly with increasing distance. Based on stratification correlation analysis of comprehensive samples, a stratification non-homogeneous random field model of marine clay parameters is established. In this model, the mean value of the random field follows a linear function along the depth, while the standard deviation is a spatially varying random field with a quadratic function trend. Parameters exhibit different mean values, initial standard deviation values, and fluctuation scales at different stratification locations. Each layer displays distinct characteristics of discontinuous variation. Compared to homogeneous random field models and non-homogeneous field models that only consider linear increases in mean value along the depth, the stratification non-homogeneous random field model comprehensively describes the complex spatial variability of mean value, standard deviation, and correlation of marine clay parameters along the depth. It also effectively captures the stratification characteristics of discontinuous variation.

Key words: marine clay, non-homogeneous random field, trend component, fluctuation component, stratification characteristic, database 304dB