Applications to Proxy Data

Abstract

Normally, the proxy data are not studied in climatology as time series in time and frequency domains, and this chapter presents an attempt to begin filling up this gap. Its main goal is methodological. Eight proxy time series from ice core sites in Greenland and Antarctica are analyzed as scalar and bivariate autoregressive time series. The analysis includes estimation of PDFs, mean values, variances, correlation functions, spectral densities, coherence functions, and spectra. The time scales of analysis are from 40 years to a millennium (from 0.001 cpy to 0.025 cycles per year). Variations of ¹⁸O concentration are found to have significant spatial variability; they can be approximated with low-order autoregressive models having smooth spectral densities. The Greenland time series do not show much dependence upon their past, and their spectra have a small dynamic range. In Antarctica, the dependence upon the past is strong, and the dynamic range of their spectra is large showing that processes of climate generation in Greenland and Antarctica are significantly different. The bivariate versions of time series from different ice cores show that they are not related to each other even at short distances thus showing the lack of a common climate signal.