Application of empirical mode decomposition methods to ground penetrating radar data
M. Manataki, A. Vafidis and A. Sarris
Journal name: First Break
Issue: Vol 32, No 8, August 2014 pp. 67 - 71
Special topic: Near Surface Geoscience
Info: Article, PDF ( 2.12Mb )
Price: € 30
M. Manataki, A. Vafidis and A. Sarris compare empirical mode decomposition methods using both synthetic and real GPR data. Empirical Mode Decomposition (EMD) is a relatively new technique introduced by Huang et al., (1998) for analysing non-linear and non-stationary time series. The decomposition is based on a signal’s local extrema, which define different oscillation modes present in the signal. What EMD does is the separation of those different oscillatory modes into a finite and usually small number of stationary sub-signals called Instrinsic Mode Functions (IMFs). EMD suffers from mode mixing which limits the frequency separation among the different modes and makes the physical meaning of the IMFs unclear.