LSTperiod software: spectral analysis of multiple irregularly sampled time series

George Caminha-Maciel, Marcia Ernesto

Abstract


Irregularly sampled time series are common in several different areas, such as astronomy, meteorology, biology, oceanography, cy- clostratigraphy, and others. The periodogram is a primary tool to extract meaningful information from irregularly spaced and noisy time series. It is an element of decision theory, meaning the periodogram usually transforms the data, and its ordinates are subsequently sub- mitted to a statistical test compared to a population originating from a known stochastic model (white Gaussian noise). If some ordinate f0 (usually a local maximum, a peak) fails in this test, we declare that it is a ‘periodicity’ at a frequency f0. Besides its full usage, this method until now suffer from numerous theoretical difficulties in adapting to real case situations and shows lack of usefulness for very poorly sam- pled and high noise cases. All of it implies low usefulness for applying in most sedimentary sequences at our disposal nowadays. The LSTperiod is an application, written in Matlab, conceived to perform spectral analysis of multiple irregularly sampled time series. It com- bines information from Lomb-Scargle periodogram estimates over different time series sampling the same phenomenon, enabling the re- covering of signals from very poorly sampled and noisy time series. The software comprises a set of four Graphical User Interfaces (GUIs) that allow the user to:

1) Have broad choices of the frequency-domain range and density for spectral estimation; 2)Select possible spectral features (i.e., pick “T”) for testing as a model [𝐴∗sin(2𝜋 𝑡–𝜃)] through the visualization of severalgoodness-of-fit statistics;
3) Visualize the fitting residuals in the time domain, for each time series, for the chosen sinusoidal model.

These tools help the user to identify and analyze any suspected feature in the estimated spectra through its related linear system responses. All estimated parameter can be saved on worksheets and the visualizations in several different figure formats. We illustrate the use of the software with a set of Ocean Drilling Program (ODP) data series that show long-period Milankovitch-related spectral features and demon- strate its performance using synthetic time series.


Keywords


Paleoceanography and paleoclimatology; Paleoclimate; Inverse methods; Statistical analysis; Methods - General or miscellaneous

Full Text:

PDF PDF

References


DOI: https://doi.org/10.4401/ag-7923
We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it (Read more).
Ok


Published by INGV, Istituto Nazionale di Geofisica e Vulcanologia - ISSN: 2037-416X