Quantifying nonlinearity and geometry in time series of climate

Statistical methods provide the opportunity to quantify what are typically qualitative descriptions of climate change. A review of linear and nonlinear processes as applied to time series of climate change is presented. Methods to quantify 'nonlinear' climatic events are given, using the late Pleistocene record of glacial-interglacial oscillations as an example. Methods to discriminate climate cycles driven by linear processes from nonlinear processes and to determine the strength of nonlinear interactions are described. The adoption of a quantitative approach to descriptions of climate variability allows direct comparison of geologic and paleoclimatic data to results of models that seek to describe modes of climate oscillations. Copyright (C) 1996 Elsevier Science Ltd