Characterization of Millennial-Scale Climate Variability

Abstract

The oxygen isotope time series from ice cores in central Greenland [the Greenland Ice Sheet Project 2 (GISP2) and the Greenland Ice Core Project (GRIP)] and West Antarctica (Byrd) provide a basis for evaluating the behavior of the climate system on millennial time scales. These time series have been invoked as evidence for mechanisms such as an interhemispheric climate seesaw or a stochastic resonance process. Statistical analyses are used to evaluate the extent to which these mechanisms characterize the observed time series. Simple models in which the Antarctic record reflects the Greenland record or its integral are statistically superior to a model in which the two time series are unrelated. However, these statistics depend primarily on the large events in the earlier parts of the record (between 80 and 50 kyr BP). For the shorter, millennial-scale (Dansgaard–Oeschger) events between 50 and 20 kyr BP, a first-order autoregressive [AR(1)] stochastic climate model with a physical time scale of τ = 600 ± 300 yr is a self-consistent explanation for the Antarctic record. For Greenland, AR(1) with τ = 400 ± 200 yr, plus a simple threshold rule, provides a statistically comparable characterization to stochastic resonance (though it cannot account for the strong 1500-yr spectral peak). The similarity of the physical time scales underlying the millennial-scale variability provides sufficient explanation for the similar appearance of the Greenland and Antarctic records during the 50–20-kyr BP interval. However, it cannot be ruled out that improved cross dating for these records may strengthen the case for an interhemispheric linkage on these shorter time scales. Additionally, the characteristic time scales for the records are significantly shorter during the most recent 10 kyr. Overall, these results suggest that millennial-scale variability is determined largely by regional processes that change significantly between glacial and interglacial climate regimes, with little influence between the Southern and Northern Hemispheres except during those largest events that involve major reorganizations of the ocean thermohaline circulation.