Some limitations of water quality models for large lakes: A case study of Lake Ontario

Some of the problems encountered in the design of a predictive water quality model for a large lake are discussed with reference to Lake Ontario and within the framework of a simple dynamic phosphorus model which incorporates the essential mechanisms of complex plankton models. By fitting the model to a seasonal data base it is shown that equally satisfactory simulations are obtained with a variety of parameterizations and regardless of conditions of annual periodicity imposed on the solution. By comparison of model output with long‐term observations it is demonstrated that seasonal verification studies of dynamic models by themselves are not sufficient to diagnose the utility of such models for predicting long‐term trends. It is concluded that the uncertainty surrounding the formulation of sedimentation and nutrient regeneration, in conjunction with the sensitivity of models to assumptions regarding dynamic balance between lake concentrations and nutrient loadings, undermines the predictive capability of dynamic water quality models.