Dynamic Bayesian Networks (DBNs) are attractive techniques for modeling complex stochastic processes. DBNs are well suited to study the evolution of groundwater contamination. In this research, a framework is described that assesses causality or correlation between constituents of groundwater quality, based on data from Ontario. The probabilistic dependencies between these constituents are efficiently extracted. The models can be used to assess and to predict the impact of pollutants on watercourses to which the groundwater is discharging, and to assist decision-making by identifying control options to decrease future groundwater contamination.
DBNs are used to assess changes in temporal groundwater quality. The resulting network model is developed in two steps. In the first step, data resulting from the pre-processing model are examined, which was organized as yearly measurements covering data from Ontario. The first task was to identify dependencies between the variables of groundwater quality in order to detect useful information on process dynamics. In the second step, the constructed DBNs are used for predicting the values of the variables in the future. The resulting networks were investigated using WEBWEAVR-III Bayesian network inference tool for analyzing measurement data from three successive time periods. Use of DBNs to predict future values requires not only discovery of a dependency between the variables but also relates together, variables from successive time periods to embed temporal features into the model. In Bayesian reasoning, the marginal probability distribution of any node may be updated upon acquiring evidence for other nodes. In this work, the development and application of a prototype DBN is described that models groundwater quality in order to assess and predict the impact of contaminant on the ground water.