Efficient method for probabilistic estimation of spatially varied hydraulic properties in a soil slope based on field responses: A Bayesian approach

Abstract

An efficient probabilistic back estimation method for characterization of spatial variability is proposed by integration of the Karhunen–Loève (K-L) expansion method, the Polynomial Chaos Expansion (PCE) method and the Markov Chain Monte Carlo (MCMC) method. To reduce the dimension of back estimation, the spatially varied soil property is simulated using the K-L expansion method and the basic random variables of K-L terms are parameters to be estimated. To further reduce computation load, a PCE surrogate model is constructed to substitute the original model. The proposed method is applied on an example where a randomly heterogeneous soil slope is subject to surface infiltration. The pressure responses are used to estimate the spatial variability of the saturated coefficient permeability. The results show that the spatial variability can be satisfactorily estimated. The coefficient of variation of the estimation is less than 5%.