Rescaled range analysis of paleoclimatic proxies

Two geophysical measurements have been widely used as paleoclimatic indicators: oxygen-isotope ratios from deep-sea cores and magnetic susceptibility of loess sediments. Both types of record have been shown to possess climatic information, as they share the same dominant frequencies of the Earth's orbital movements, which, according to Milankovitch's theory, drive the climatic system. But these two physical quantities may respond to climatic variations in different ways. Consequently, they may possess different long-term characteristics, which can be detected by the technique of rescaled range (R/S) analysis. Here we report results from four sets of data: two oxygen-isotope records and two loess susceptibility profiles. Analysis indicates that Hurst exponents of about 0.8 are typical for deep-sea oxygen data, whereas values of about 0.9 characterize loess susceptibility sequences. If Mandelbrot's explanation is taken, these distinct values imply that the loess data are more self-correlated and possess stronger persistence compared with the deep-sea records. This may reflect differences in the ways in which these proxy recorders respond to climatic variations, or differences between continental and oceanic climates.