The nonlinear prediction problem in the study of the runoff cycle

The general theory of nonlinear synthesis and analysis in hydrology is discussed, with particular reference to the problems of predictive uncertainty. These problems are associated with incomplete descriptions of the hydrologic systems, limited model-prototype equivalences, system variability in time, and system nonlinearity. Solutions to the nonlinear analysis problem of hydrologic systems under the assumption of approximate time invariance, based on complex cascade network approximations, power series and polynomial expansions, and variable response function approximations are presented. The relations between present theoretical knowledge and its practical application are summarized.