The definition of the geotechnical model for slope design is based on the geological, structural, rock mass and hydrogeological models. Each model is described by different sets of information and parameters and is defined at a scale of interest for the purpose of the analysis of slope behaviour. However, no clear guidelines exist in terms of the appropriate statistical methods to manage this information. Probabilistic methods are traditionally used to account for the uncertainty in engineering design, however, the base assumptions of these methods are not always fully understood, resulting in misinterpretations of results. There are two main approaches of statistical analysis known as frequentist (classical) and Bayesian, which are based on different interpretations of probability. In the classical approach, probabilities are considered as frequencies in a series of similar trials, whereas in the Bayesian approach, probabilities correspond to degrees of belief. One of the main characteristics of the Bayesian approach is that makes use of both prior information on the hypothesis (or model) being examined and the likelihood of the available data, to provide a balanced answer to the probability of that hypothesis (or model). Another aspect of the uncertainty characterisation process is the understanding of the type of uncertainty present in the various components of the geotechnical model. At a broad level there are two main types of uncertainty in geotechnical engineering, one due to random variation of the aspect evaluated (aleatory) and the second due to lack of knowledge of the subject under analysis (epistemic). The uncertainty is considered epistemic if it can be reduced with the collection of additional data or by refining models, otherwise it is treated as natural variation. The majority of the uncertainty in the geotechnical model for slope design is epistemic, typically analysed with probabilistic methods. However, epistemic uncertainty has different aspects some of which (i.e. vagueness or non-specificity) can be represented more naturally with alternative approaches outside the field of probability (i.e. interval analysis, possibility and evidence theories). Simple examples will be included to illustrate the merits of treating uncertainty in the mine slope design process with unconventional methods such as Bayesian statistics and non-probabilistic based approaches.
