PREDICTING CHANGE IN NON-LINEAR SYSTEMS

Complex systems are characterizedby surprising switches to new behaviours. Evaluating and predicting these changes demands anunderstanding of the behaviour of the whole system. The combined ecosystem-climate system shows chaoticor pseudorandom behaviour, stochastic or trulyrandom behaviour, as well as simple bifurcation andsemi-stability. Semistability involves the suddenchange from a destabilized attractor to a newstable attractor which may occur after an apparentlyunpredictable time delay. We present some recentresults for analyzing time series data and for usingsimulations of non-linear models to predict these changes.