Spatial Variability of Cohesive Soil Parameters Based on Cone Penetration Test and Geotechnical Laboratory Tests

Spatial autocorrelation distance is a key parameter for analyzing the spatial variability of geotechnical parameters. This study utilizes three methods, including the average distance method, recursive space method, and the curve limit method, to calculate the autocorrelation distance based on side friction (f_s), tip resistance (q_c), and soil parameters. These methods were applied to 246 cone penetration tests (CPT) and 18 soil sampling points obtained from a construction site of cohesive soil in Zhejiang Province. The effectiveness of different methods in computing the vertical and horizontal autocorrelation distances was compared. It’s found that the recursive spatial method is most effective in calculating the autocorrelation distances, with fewer outliers and smaller variation coefficients. The autocorrelation distances based on f_s and q_c are comparable, with vertical and horizontal autocorrelation distances of approximately 2 m and 19 m, respectively. Nonetheless, the outcomes vary somewhat depending on the soil parameters utilized. The typical values of autocorrelation distance for cohesive soil were determined, which provides a foundation for analyzing the spatial variability of soil parameters. This research has important value for laying of sampling points, establishment of random fields, and subsequent reliability analysis.