This thesis presents a framework for stochastic analysis of infiltration and slope stability by considering uncertainty nature of soil-water retention characteristic in unsaturated soil. In the uncertainty analysis of infiltration, procedures were developed to quantify the statistical features of infiltration and the corresponding rainfall excess from a specified rainfall hyetograph. The Green-Ampt infiltration model was used in which soil-water retention parameters, such as saturated hydraulic conductivity and residual water content, were considered random. With the application of Monte Carlo simulation, the effect of random model input variables on the model outputs were examined. It is interesting to find that most of the variance of the Green-Ampt model output are contributed by the...[ Read more ]

This thesis presents a framework for stochastic analysis of infiltration and slope stability by considering uncertainty nature of soil-water retention characteristic in unsaturated soil. In the uncertainty analysis of infiltration, procedures were developed to quantify the statistical features of infiltration and the corresponding rainfall excess from a specified rainfall hyetograph. The Green-Ampt infiltration model was used in which soil-water retention parameters, such as saturated hydraulic conductivity and residual water content, were considered random. With the application of Monte Carlo simulation, the effect of random model input variables on the model outputs were examined. It is interesting to find that most of the variance of the Green-Ampt model output are contributed by the variance of saturated conductivity and followed by the inverse of residual water content and lamda. Also, the varying of clay content inside soil can cause effect to the statistical features of model output.

Furthermore, soil-water retention characteristics play important roles in stability of an unsaturated slope. Thus, the investigation of stochasticity in soil-water retention characteristics and shear strength parameters on slope stability was also conducted. Seepage and slope stability analyses were conducted by using computer software SEEP/W and SLOPE/W. Due to time-consuming nature of seepage analysis, a stratified random sampling method, called Latin hypercube sampling technique, was applied in the stochastic analysis to determine the statistical features of safety factor and to quantify variance contribution from the model input uncertainties. The statistical features for safety factor of the slip surface near the ground surface are quite different to other safety factors with slip surfaces located at the lower part of the slope. Also, the effect of uncertainty of shear strength parameters are very important to the uncertainty of model output from slope stability analysis compared with that of soil-water retention parameters.