A decomposition approach to cyclostratigraphic signal processing

Sedimentary rocks can record signals produced by highly complex processes. These are generated a progressive deposition of sediments which be affected, mainly through the climate system, regular astronomical cycles (i.e. Milankovitch cycles), and irregular oscillations like El Niño-Southern Oscillation. Also, usually biological, chemical and/or physical post-depositional processes, sedimentary records affected pattern-creating heterogeneous The noise in further complicates records, rate (or sedimentation rate) fluctuate, greatly reduces effectiveness classical stationary time-series analysis methods commonly used cyclostratigraphy study found records). Faced with this multiplicity common approach is to reduce each signal more manageable sub-signals, either over given range frequencies (e.g., filtering), or considering continuum constant using transforms). This makes it possible focus on features interest, cycles. However, working sub-signals not trivial. Firstly, have certain amount cross-cancellation when they summed back reconstruct initial signal. means that filters transforms, wiggles present appear sub-signals. Secondly, considered often cannot signal: there processes affecting remain unstudied. It take into account consider entire content dividing decomposition: set added together original We discuss here how reframe techniques context decomposition, cross-cancellation, adequate for comprehending whole signals. also show decomposition carried out non-stationary methods, minimise now reached sufficient maturity tackle novel tools adapt cyclostratigraphic purposes, based concepts Empirical Mode Decomposition (EMD) Instantaneous Frequency (IF), mainly: (1) fast Ensemble (EEMD) algorithm, (2) quality metrics (3) plots visualise instantaneous frequency, amplitude frequency ratio. illustrate use these applying them greyscale from site 926 Ocean Drilling Program, at Ceara Rise (western equatorial Atlantic), especially identify characterise expression main goal minimising we apply real what call wiggle-in-signal approach: making representative expression, wiggle wiggle, all cycles). finally argue could as practical standard output interpretation