variety of models is needed-one mathematical model cannot provide answers to all questions. These models are analogous to the emissions-to-air-quality models used in developing air pollution control programs.

To be successful, there must be good predictive capability for the dominant factors that determine the engineering choices for satisfying the standards. These factors are determined by sensitivity analyses as well as the experience of the modeler. Thus for engineering purposes it is not necessary to understand every process if more knowledge would have no effect on the choice of control strategy. For example, it is not necessary to understand the behavior of a certain pollutant at a location where the input is far below any possible threshold value of concern. Or another example, if the range of uncertainty of a biological effect is a factor of 10, the effect may be of no importance if the upper end of the range is fully acceptable.

The preceding discussion relates to the wastewater disposal system for a municipality including source control programs, a treatment plant, and outfall. The three kinds of information required apply equally well to all other types of pollutant sources and the approach to devising an engineering system. In the case of CSOs, for example, it is necessary to have a set of water quality objectives; some knowledge of the environment; and information on the amount, quality, frequency, and distribution of existing CSOs.