This paper introduces the empirical mode decomposition (EMD) method into seismic facies analysis, to determine favorable attributes that reflect subsurface structures and characteristics of hydrocarbon saturation. The EMD method is the core of the Hilbert-Huang transformation which is a newly developed time-frequency analysis technique. The method uses the EMD to decompose seismic data into Intrinsic Mode Functions (IMF), and different IMFs have different frequency characteristics and indicate different geological information. This paper combines the EMD method with the Kohonen''s selforganizing Neural Network based seismic facies analysis. The reconstructed seismic data using only the characteristic IMF components can more clearly indicate fault distribution and favorable object areas. Therefore, the EMD-based method can be used to enhance the signal-tonoise ratio and the resolution of seismic attributes, and thus it is of great importance to structure interpretation and reservoir prediction.
The empirical mode decomposition (EMD) method developed by Huang et al. (1998) is a powerful method that can be used to detect nonlinear and non-stationary signals. Combined with the Hilbert transform, the EMD method forms the Hilbert-Huang transform (HHT). Compared with the traditional Fourier transform and wavelet transform, the Hilbert-Huang transform is not constrained by time windows; it decomposes signals according to characters of time scales, and thus has strong adaptability and good time-frequency localization. So it is more suitable for processing nonlinear and non-stationary signals. In addition, every IMF by EMD contains different frequency information, is able to better reflect the inherent physical properties of the input data, and so that brings much convenience to practical interpretation work. At present, this method is very popular in industry and has some application in seismic exploration, for example, denoising, seismic signal processing and channel identification. With the development of seismic technology, the seismic face analysis techniques also develop fast. Waveform analysis technique is one of them, which uses the character of seismic waveform to compare the seismic data trace by trace, finely characterizes the lateral variations of seismic signals, and then obtains the planar distribution of seismic anomalies. The waveform classification result is classified to attain the diagram of planar seismic facies. The seismic facies analysis technique based-on waveform, comprehensively uses all kinds of seismic information to attain the total variation and distribution of seismic signals, and has unique capability to solve problems. The real data application demonstrates that the diagram of waveform-based seismic facies analysis characterizes more detailed geological information and gives important help to identify the characters of fault distribution and forecasts the oil & gas distribution.
EMD decomposes a data series into a finite set of signals, called intrinsic mode functions (IMF). The IMFs represent different oscillations embedded in the data. They are constructed to satisfy two conditions: (1) the number of extrema and the number of zero-crossing must be equal to or differ at most by one; and (2) at any point the mean value of the envelope defined by the local maxima and the envelope defined by the local minima must be zero.