Search

Cyclostratigraphic analyses rely on techniques that trace astronomical components in paleoclimate signals. These techniques have demonstrated their value but rely on certain assumptions on the presence and quality of the astronomical imprint. Here we explore a new conceptual approach to time series analysis. Specifically, we evaluate the potential of spectral moments to characterize the full spectral characteristics of a record and thus not only the frequency ranges of interpreted astronomical components. Mathematically speaking, moments are unique quantities describing a specific set of points. In the case of spectral moments, we apply the concept of moments on the distribution of spectral power in a signal's periodogram. We present four case studies that illustrate the advantages and disadvantages of the spectral moment approach in gaining insight in the (astronomical) features of a particular data record. We discuss the effects of outliers in a series, variable sedimentation rate, and changing climate dynamics on the spectral moments of a power spectrum. At the same time, we carry out a sedimentation rate reconstruction based on the spectral moment approach and compare that reconstruction to results obtained through classic approaches. Compared to classic approaches, the spectral moments approach is robust and requires less strict assumptions to obtain similar first‐order information. Yet such assumptions are still necessary to achieve more precise sedimentation rate reconstructions. In summary, we show that the spectral moments approach is suitable to obtain first‐order insights in variable components embedded in a depth series.

Geophysical research letters 44(13), 6914-6925(2017). doi:10.1002/2017GL072977, Published by Wiley, Hoboken, NJ