Analysis of Uncertainty in River Water Quality Modelling

The case is given for attention to the evaluation of uncertainty in water quality modelling, in the contexts of new demands for assessment of risk to water quality status, and typical paucity of supporting data. A framework for the modelling of water quality is outlined and presented as a potentially valuable component of broader risk assessment methodologies, and potentially useful methods of numerical uncertainty analysis are reviewed and demonstrated. A selective library of dynamic models and numerical tools for model solving and uncertainty analysis are compiled into novel software for model uncertainty analysis and risk-based decision-support. This software is applied to a series of case studies in an exploration of the underlying numerical problems and their relevance to modelling and management objectives using relatively sparse data sets. Issues examined in some detail are the importance of reconciling numerical solution tolerances with overall model precision; relative effects of numerical approximations, data and model structural biases on optimal design of field experiments and on prediction reliability; and the value and limitations of extending established methods of uncertaint