EVALUATION OF PROBABILISTIC FLOW IN TWO UNSATURATED SOILS

Abstract

A variably saturated flow model is coupled to a first-order reliability algorithm to simulate unsaturated flow in two soils. The unsaturated soil properties are considered as uncertain variables with means, standard deviations, and marginal probability distributions. Thus, each simulation constitutes an unsaturated probability flow event. Sensitivities of the uncertain variables are estimated for each event. The unsaturated hydraulic properties of a fine-textured soil and a coarse-textured soil are used. The properties are based on the van Genuchten model. The flow domain has a recharge surface, a seepage boundary along the bottom, and a no-flow boundary along the sides. The uncertain variables are saturated water content, residual water content, van Genuchten model parameters alpha (α) and n, and saturated hydraulic conductivity. The objective is to evaluate the significance of each uncertain variable to the probabilistic flow. Under wet conditions, saturated water content and residual water content are the most significant uncertain variables in the sand. For dry conditions in the sand, however, the van Genuchten model parameters α and n are the most significant. Model parameter n and saturated hydraulic conductivity are the most significant for the wet clay loam. Saturated water content is most significant for the dry clay loam.