ASSESSING THE ABRUPTNESS OF CLIMATE CHANGE VIA STATISTICAL AND SPECTRAL STUDY OF ICE CORE ISOTOPIC RECORDS

Over the last decade, advances in the acquisition and interpretation of high-resolution proxies of climate change have lead to a greater appreciation of the linked dynamics of Earth systems. These records suggest significantly greater instability in the global climate system during the late Pleistocene than has been exhibited during the Holocene. Specifically, much attention has been given to the apparently rapid rate of change in Pleistocene climate (abrupt jumps).

Data from the GISP2 core has been broken into four subsets: annual dD values over approximately the last 700 years; d¹⁸O values over the last 9000 years, d¹⁸O values over the interval 25ky-50ky; and finally d¹⁸O values over the interval 50ky-110ky. These four subsets were selected due to their differences in temporal resolution and their significance to the notion of abrupt climate change. For each data set, time-rate-of-change in isotopic composition has been compared to a Poisson random model of compositional change. This technique allows for recognition of those changes in composition that are indeed unexpectedly rapid. For the Holocene records, less than 1% of the observed rates of change were greater than predicted by a Poisson model; while for the late Pleistocene, approximately 5% of the changes were larger than predicted. Additionally, relationships between age of sample and rate of change of composition indicate that while the magnitudes of Pleistocene changes are larger than those recorded during the Holocene, the rate of change (o/oo per year) in composition is some two to three orders of magnitude slower. Importantly, this change in rate is due more to an increase in temporal spacing of data than it is due to a decrease in the magnitude of change between samples. Finally, power-frequency relationships describe a pronounced shift to low-frequency dominated spectra as the age of the data set increases. That is P(f) µ f ⁰ for dD data; P(f) µ f ^-1/2 for Holocene d¹⁸O data, and P(f) µ f ^-1 for Pleistocence d¹⁸O data. Such shifts in the power-frequency relationship are consistent with previously described climatological records.