Parameter Estimation and Uncertainty Analysis for Rainfall Infiltration in Unsaturated Soils Using a Bayesian Approach

In this study, a probabilistic method is proposed for estimating soil property and model uncertainty of a rainfall infiltration model in unsaturated soils using field observation of pore-water pressures. The proposed method is formulated based on the Bayesian framework and solved using the Markov Chain Monte Carlo simulation method. A case study of a field test of rainfall-induced slope failure is presented to illustrate the proposed method. It is found that all the parameters except q₂ are well identified within a small region. The posterior distributions of k_s and α show a normal shape and two parameters are positively correlated. Before the rain period, the 95% total prediction uncertainty bounds indicate a relatively good model performance. During the rain period, the results are less accurate.