Estimating Anisotropic Soil Properties Using Bayesian Kriging

Site investigation is a prior for the preliminary design of most geotechnical structures. Estimation of soil characteristics must rely on regional knowledge from the site investigation. The geotechnical investigation in offshore area is often expensive and conducted at a limited number of locations. The soil properties, with uncertainties in soil layering and soil strength, cannot be well characterized by the limited number of site investigations. Moreover, the current practice takes no account of geotechnical knowledge from nearby sites which is capable to reduce the uncertainties in the soil properties. This paper provides a theoretical framework using Bayesian Kriging to allow predicting the anisotropic soil properties at unsampled locations by combining the limited site investigations with the existing information at nearby locations. The spatial variability of the soil property is considered in this method. An example is presented to demonstrate the procedure of the method and verify the accuracy of the results.