Abstract
In recent years more and more long-term broadband data sets are collected in geosciences. Therefore there is an urgent need of algorithms which semi-automatically analyze and decompose these data into components which are associated with different processes. In the past, the standard tools for decomposing data sets were based on either the Fourier or the Wavelet Transform. In this paper we investigate the novel approaches Empirical Mode Decomposition and Sparse Decomposition as well as the Harmonic and Wavelet Decomposition for long-term sea floor pressure data analysis. In a comparative investigation conducted with Matlab ® these tools were applied to real and synthetic sea floor pressure data sets. Our results indicate that none of the methods is entirely suited for this objective, but Sparse Decomposition performs best except for computing efficiency.
| Original language | English |
|---|---|
| Pages (from-to) | 4-12 |
| Number of pages | 9 |
| Journal | Computers and Geosciences |
| Volume | 45 |
| Early online date | 5 Apr 2012 |
| DOIs | |
| Publication status | Published - 1 Aug 2012 |
Funding
We would like to thank Hans-Hermann Gennerich and Earl Davis for providing the sea floor pressure data and Peter Maass for supporting one of us (ME) during this study. All computations were done on the computer network provided by the Department of Mathematics and Computer Science of the University of Bremen. Their help and support is kindly acknowledged. Reviews of two anonymous reviewers helped to improve the article. Appendix A
Keywords
- Empirical Mode Decomposition
- Fourier Transform
- Sparse Decomposition
- Time series
- Wavelet Transform
ASJC Scopus subject areas
- Information Systems
- Computers in Earth Sciences