Empirical Mode Decomposition Method Applied to Atmospheric Variables
Abstract
The majority of atmospheric data are non-stationary and are the result of nonlinear processes, yet, the techniques nearly always used to analyze these data (i.e. Fourier analysis, Wavelet analysis) are inherently inappropriate for this data type. This study investigates a relatively new data processing technique called the Empirical Mode Decomposition (EMD) method, which is specifically designed for evaluation of nonlinear and non-stationary processes. The method essentially decomposes time series data into a complete and finite intrinsic set of nearly orthogonal functions. The method is discussed and the characteristics of the intrinsic functions are analyzed with respect to the partitioning of energy, variance and covariance with frequency. Implications of the method are then compared and contrasted with current theory and methods based upon Fourier analysis. Also, a precise definition of 'Trend' is described using the EMD method, which is advantageously adaptive and intrinsically data-based as opposed to current fitting methods.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2008
- Bibcode:
- 2008AGUFM.H51B0804B
- Keywords:
-
- 0394 Instruments and techniques;
- 1839 Hydrologic scaling;
- 1878 Water/energy interactions (0495);
- 4568 Turbulence;
- diffusion;
- and mixing processes (4490)