ABSTRACT
Soils are geological materials formed by weathering processes, transported by physical means to their present locations. They have been subject to various stresses, pore fluids, and physical and chemical changes. Thus, it is not surprising that the physical properties vary from place to place, even in one soil stratum. Researchers have identified and classified major components of uncertainties associated with the estimation of soil properties (Vanmarcke, 1977; Tang, 1984; Phoon and Kulhawy, 1999; Baecher and Christian, 2003; Zhang et al., 2003). Vanmarcke (1977) presented basic concepts and methods for describing the spatial variability of a soil stratum within random field framework. Three major sources of uncertainties were identified for stochastic modeling of soil profiles. The first is natural heterogeneity or in situ variability of the soil, which is caused by variation in mineral composition, varying depths of strata during soil formation and stress history. The second is attributed to limited availability of information about subsurface conditions. The third is measurement errors due to sample disturbance, test imperfections, human factors, and estimating soil properties through correlation with index properties. Kulhawy (1992) classified three primary sources of geotechnical uncertainties: inherent variability, measurement uncertainties, and transformation uncertainties (Figure 8.1), which are basically the same as those defined by Vanmarcke (1977). The uncertainties caused by equipment, operator, random test effects, and statistical uncertainties were all included in measurement errors. The transformation uncertainties meant the uncertainty introduced when field or laboratory measurements are transformed into design soil properties using empirical or other correlation models.
