SPECTRAL-ANALYSIS OF CLIMATE DATA

The complexity of climate variability on all time scales requires the use of several refined tools to unravel its primary dynamics from obse rvations. Indeed, ideas from the theory of dynamical systems have prov ided new ways of interpreting the information contained in climatic ti me series. We review the properties of several modem time series analy sis methods. Those methods belong to four main classes: Fourier techni ques (Blackman-Tukey and Multi-Taper), Maximum Entropy technique, Sing ular-spectrum techniques and wavelet analysis. Their respective advant ages and limitations are illustrated by numerical experiments on synth etic time series. As climate data can be irregularly spaced in time, w e also compare three interpolating methods on those time series. Those tests are aimed at showing the pitfalls of the blind use of mathemati cal or statistical techniques on climate data. We apply those methods to 'real' climatic data from temperature variations over the last cent ury, and the Vostok ice core deuterium record over the last glacial cy cle. Then we show how interpretations on the dynamics of climate can b e derived on those time scales.