Abstract

Abstract 14 Paleoclimate records can be considered low-dimensional projections of the climate sys-15 tem that generated them. Understanding what these projections tell us about past cli-16 mates, and changes in their dynamics, is a main goal of time series analysis on such records. 17 Laplacian Eigenmaps of Recurrence Matrices (LERM) is a novel technique using uni-18 variate paleoclimate time series data to indicate when notable shifts in dynamics have 19 occurred. LERM leverages time delay embedding to construct a manifold that is map-20 pable to the attractor of the climate system; this manifold can then be analyzed for sig-21 nificant dynamical transitions. Through numerical experiments with observed and syn-22 thetic data, LERM is applied to detect both gradual and abrupt regime transitions. Our 23 paragon for gradual transitions is the Mid-Pleistocene Transition (MPT). We show that 24 LERM can robustly detect gradual MPT-like transitions for sufficiently high signal-to-25 noise ratios, though with a time lag related to the embedding process. Our paragon of 26 abrupt transitions is the “8.2 ka” event; we find that LERM is generally robust at detect-27 ing 8.2 ka-like transitions for sufficiently high signal-to-noise ratios, though edge effects 28 become more influential. We conclude that LERM can usefully detect dynamical tran-29 sitions in paleogeoscientific time series, with the caveat that false positive rates are high 30 when dynamical transitions are not present, suggesting the importance of using multi-31 ple records to confirm the robustness of transitions. We share an open source Python 32 package to facilate the use of LERM in …

Date

June 8, 2023

Authors

Alexander James, Julien Emile-Geay, Nishant Malik, Deborah Khider

Journal

Authorea Preprints

Publisher

Authorea