Morrison, Foster, "Predictability and Forecasting", 1985
ua435
Forecasting for complex nonlinear systems has proven to be elusive. Investigators have assumed the causes to be too little data and overly-simplified models. Recent studies in climatology reveal that nonlinear systems behave in ways quite different from the linear or static systems of traditional science and engineering. The behavior of nonlinear systems can be cyclical or essentially stochastic and usually is a mixture of both. New techniques, such as "attractors," are being devised to facilitate analysis. Methodologies must be applied with due consideration to the structure of the system under investigation.