Copy the page URI to the clipboard
Luby, Anthony; Kbaier, Dhouha and Brandon, Mark
(2024).
DOI: https://doi.org/10.1109/OCEANS51537.2024.10682375
Abstract
Evaluating components of environmental time series, such as diurnals, seasonal and trends of Arctic Sea water temperatures are key to providing an understanding of any pattern shifts in such a sensitive environment. This analysis requires a robust decomposition method, given the challenges posed by the presence of noisy, missing, non-stationary and non-linear signal characteristics. Traditional frequency analysis techniques, such as Fourier techniques, are compromised in the context of these data types. Two variants of the Empirical Mode Decomposition method and a Variational Mode Decomposition technique are comparatively assessed as prime candidates, given the data context. A baseline application of the three techniques to a synthesised signal is established, followed by direct application to an annual sea water temperature dataset from the Gascoyne Bay inlet in the Arctic.