Empirical Mode Decomposition for Seismic Applications

• Author / Creator

  Han, Jiajun

• Empirical mode decomposition (EMD) is a powerful signal analysis technique to analyze non-stationary signal systems, like seismic data. Through the sifting process, EMD splits the non-stationary features of the input signal into individual decomposition modes, which are called intrinsic mode functions (IMFs). Each IMF has a symmetric, narrow-band waveform, which ensures that their instantaneous frequency of them is smooth and positive. However some negative features encumber its direct application namely mode mixing and splitting, aliasing and endpoint artifacts. Two variants, ensemble EMD (EEMD) and complete ensemble EMD (CEEMD) have been recently introduced to overcome some of the negative features associated with EMD. Furthermore, two EMD-like methods are also introduced: first one is the synchrosqueezing transform (SST), which decomposes the input signal into SST modes, and these modes manifest similar features to IMFs; another one is the 2D extension of EMD, bidimensional empirical mode decomposition (BEMD), which can aid image analysis. This thesis focuses on testing the suitability of EMD methods for seismic processing and interpretation, and we present 4 new techniques. The first method is CEEMD combined with instantaneous spectra for seismic spectral decomposition. After CEEMD, the instantaneous frequency spectra manifests visibly higher time-frequency resolution than short time Fourier and wavelet transforms on both synthetic and field data examples. The second method is EEMD thresholding. It is effective for suppressing random noise in each trace, which is highly attractive for microseismic processing. Furthermore, the proposed EEMD thresholding can be extended into the f-x domain as f-x EEMD thresholding, which aims to reduce dip- ping coherent and random noise. The third application is SST for seismic signal time-frequency analysis. It shows comparable results to CEEMD combined with instantaneous spectra; therefore it is highly suitable for high resolution seismic interpretation. The last proposed method is BEMD thresholding, which aims to reduce random noise of 2D seismic images. Utilizing the particular features of IMFs or SST modes, the presented methods manifest excellent performance on seismic spectral decomposition and seismic denoising. The synthetic and real data examples illustrate that EMD methods are highly promising for seismic processing and interpretation.

• Subjects / Keywords

  • Empirical mode decomposition
  • Denoising
  • Time-frequency analysis

• Graduation date

  Fall 2014

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/R3JQ0T35R

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.