OBSERVED CLIMATIC VARIABILITY: TIME DEPENDENCE
JOHN M. WALLACE
University of Washington
Seattle, USA
Contents
1 Introduction
2 Climate time series
2.1 Periodic phenomena . . . . . . . .
2.2 Quasi-periodic phenomena . . . . .
2.3 Aperiodic and random variability .
2.4 Trends . . . . . . . . . . . . . . . .
2.5 Regime shifts and unprecedented events
3 The view from "phase-space"
1
3
3
7
10
17
19
3.1 Progressive oscillations in two-dimensional phase-space.
21
22
26
3.2 Random variability in phase space . . . . . . . . . . . .
4 Concluding Remarks
26
5 Acknowledgments
29
1 Introd uction
Preconceived notions concerning the nature and causes of climate variability determine the datasets that scientists examine, the analysis tools they
employ, and the questions they address in their research. Their choices, in
turn, define and limit the range of possible outcomes. If these notions are
wrong, the research is likely to get off on the wrong track. If they are lacking altogether, the course of the research may be determined by default,
through ad-hoc choices, as in the maxim: "If all you have is a hammer, all
you'll see is nai~s)) .
This chapter is designed to complement the abundant literature on statistical analysis techniques used in climate research (e.g., Barnett 1981,
Joliffe 1986; Richman 1986; Preisendorfer 1988; Vautard and Ghil 1989;
Bretherton et al. 1992; von Storch et a1.1995). Rather than illustrating
NATO AS! Series. Vol. I 44
Decadal Climate Variability
Dynamics and Predictability
Edited by David L. T.Anderson and Jtirgen Willebrand
© Springer· Verlag Berlin Heidelberg 1996