Introduction

As mentioned in Chapter 1, generating a spectrum is often the first step in the analysis of cyclostratigraphic time series. Unfortunately, researchers sometimes use one of the widely available computer programs without regard to the theory or methodology being implemented. A working understanding of the main issues will reduce the chances of making mistakes and enable the investigator to make the most of the data. A primary use of the spectrum is for the detection of periodic or quasi-periodic components in a time series, which are indicated by peaks that can be distinguished from background noise (i.e. significant spectral peaks). Note that the periodic or quasi-periodic components are referred to here as ‘regular cycles’ rather than just ‘cycles’ for reasons explained in Section 1.3.

This chapter is divided into four sections that consider: (a) processing time series prior to spectral analysis; (b) the background to the generation of power spectra; (c) the main methods used to generate spectra; and (d) some of the issues concerning distinguishing spectral peaks from background noise. The practical interpretation of power spectra is also addressed in Section 5.4.2.

It should be kept in mind that it is impossible to use cyclostratigraphic time series to obtain the theoretical or ‘true’ spectrum. The reasons for this will become clear in this chapter. Different spectral techniques employ different methods to obtain a ‘best guess’ as to the true spectrum.