Spatial Variability of Soil Properties in Saudi Arabia: Estimation of Correlation Length

Spatial variability of soil properties plays an important role in geotechnical engineering, influencing the design and performance of geo-structures. This study investigates the characterisation of this spatial variability by examining in-situ soil data obtained from 74 Cone Penetration Tests (CPTs) conducted in a reasonably homogeneous granular fill of a large construction project in Saudi Arabia. The primary focus of this work is to assess, from the available in-situ CPT data, the spatial variability of this sand fills in terms of the vertical correlation length (θv). The autocorrelation function (ACF) method is used to estimate θv, and the geo-statistical patterns of all estimated θvs are analysed through the values of its mean (µ), standard deviation (σ), coefficient of variation (CoV), and fitted probability density function (pd f). An assessment of the accuracy in the estimated θvs is then carried out from the statistical analysis of 1,000 one-dimensional (1−D) random fields with spatial statistical information similar to the in-situ fill. The study shows that as much as 25% error can be expected for the specific depth (D = 10m) and sampling interval (dx = 0.01m) of the 74 CPTs considered.