The synchrosqueezing transform to evaluate paleoclimate cyclicity

Abstract

Centennial and millennial-scale climate cycles are observed in paleoclimate proxies distributed globally. For example, periods of ∼2.3, ∼1.5, ∼1, and ≤0.5 thousand years (kyr) are reported in the Holocene, and numerous studies have related cycles with such periods to fluctuations in solar intensity received at the Earth's surface. For the 1.5 kyr cycle, however, there is still much uncertainty surrounding the exact period and presence of the oscillation. This uncertainty is primarily due to difficulty distinguishing cycles from one another; distinguishing this 1.5 kyr cycle from surrounding cycles and determining the onset and cessation of this cycle is crucial to the evaluation of the mechanism driving this cycle. Commonly methods for paleoclimate spectral analysis, such as the Continuous Wavelet Transform and Short-Time Fourier Transform cannot clearly reveal the onset or cessation of sporadic climatic cycles if neighboring spectral peaks are present within an octave due to spectral smearing and leakage. Here, we present a novel application of the Synchrosqueezing Transform in paleoclimate research to obtain improved separation of climate cycles and to more precisely determine the onset and cessation of cycles that are sporadic through the paleoclimate proxy time-series. The SST and its inverse provide a more exact estimate of the onset and cessation of which is useful in comparing to the spectra of other paleoclimate proxies to determine the drivers of specific climate oscillations. We demonstrate that application of the SST to a range of Holocene climate proxies provides consistent identification of the 1.5 kyr cycle.