Uncertainty Assessment in Environmental Risk through Bayesian Networks

Assessing environmental risks on large dams is a challenging task. This paper describes a study on a novel and comprehensive application of Bayesian Networks (BNs) on the Abolabbas dam in Iran. Bayesian networks are based on probability theory and provide a powerful tool for structuring conceptualizations of the interactions between variables with uncertainties. Firstly, the interaction-based structure of variables is developed using the graphical model. Then, the Bayesian Network input variables, which affect the risk in two categories ("hazards index" and "consequences index"), are determined and the relations between different variables are modeled. The probability values for the risk levels are derived from a novel fuzzy set analysis. The results show that the environmental risk of the Abolabbas dam is considered at a high level with 12.8 percent probability. Also, the sensitivity analysis is used to find out the most effective variables on the environmental risk of the dam site. Finally certain important action plans are suggested to reduce and control the risk which represents a novel way for the risk reduction.