Reliability approach to slope stability analysis with spatially correlated soilproperties

A reliability approach to a slope stability analysis, considering spatially correlated soil properties, is used to systematically evaluate the various sources of uncertainty that arise in these types of problems. Measurement errors and bias that often occur during soil investigations are now incorporated into the probabilistic model of soil properties in addition to the spatial variability and the effects of spatial averaging. The results of the analysis show that the unconditional approach, which considers spatial variability, but does not account for measurement-related uncertainties, can yield unconservative results because of an unrealistically large variance reduction. In contrast, the results obtained from the unconditional approach, which does consider spatial variability and measurement-related uncertainties, as well as those from the conditional approach, lead to significantly lower probabilities of failure that are much less sensitive to the scales of fluctuation.