The uncertainty associated with predictions made by solute transport models in groundwater environments must be carefully evaluated to address spatial and time variability as well as data accuracy and precision. Such uncertainty influences the level of confidence in predictions used for decision making. Analytical and numerical solutions of the solute transport equation were first used to predict reference solute concentrations for average values of the following parameters: influent concentration, pore velocity, dispersion coefficient, and retardation factor. Then, estimates were made for conditions based on expected variations and combinations of the same parameters. Two techniques were used to study the extent of possible uncertainty: a) variations of one parameter at a time and b) a random Latin Hypercube Sampling Technique. Results show that different variables have various effects on predicted concentrations. Different techniques provide various ways of evaluating sensitivity. Simulations suggest the need for the careful measurement or estimation of variables in order to increase the confidence of predictions. At a minimum, predictions should be accompanied by a reasonable analysis of prediction sensitivity.
