Probabilistic analysis of soil-water characteristic curve with Bayesian approach and its application on slope stability under rainfall via a difference equations approach

The soil-water characteristic curve (SWCC) is an important tool in the interpretation of the properties of unsaturated soil, and is widely applied in Geotechnical and Geo-environmental engineering. The SWCC uses several fitting parameters to express the best fit equations, and the fitting parameters depend on the experimental data. The reliability and predictability of SWCC are always being questioned. In this paper, Fredlund and Xing model is used to fit the experimental data of loamy sand and sandy clay loam in the UNSODA database, the fitting parameters values are discussed. Bayesian framework and the Markov Chain Monte Carlo simulation method are applied to estimate the uncertainty of SWCC from the errors and variance of the fitting parameters, and SWCC of different levels of confidence is obtained. Applications of the confidence on slope stability analysis are presented by a simplified finite element program.