Probabilistic Unsaturated Flow along the Textural Interface in Three Capillary Barrier Models

The probabilistic flow within capillary barrier models is evaluated by coupling a first-order reliability (probabilistic) model to a variably saturated flow model. The objective is to determine the most significant uncertain variable to probabilistic flow and the effect of different textural combinations. The model inputs include the mean and standard deviation of each uncertain variable. The van Genuchten curve-fitting model for unsaturated flow is used. The uncertain flow variables are saturated water content, residual water content, saturated hydraulic conductivity, and model parameters alpha $\left(\alpha \right)$ and $\mathbf{n}$ . A clay loam soil sample overlying a loamy sand sample, a clay loam sample overlying a sand sample, and a loamy sand sample overlying a sand sample are used to represent finer soils overlying coarser soil. Evaluations near the textural interface show that model parameters alpha $\left(\alpha \right)$ and $\mathbf{n}$ =most significant uncertain variables. These are related to the water entry pressure and the width of pore size distribution of the soils. Generally, soils that have the least uniform pore size distribution are the most effective capillary barriers. In this study, the clay loam overlying loamy sand satisfied this condition and performed better than the other soil combinations.