Technology Research Of Lithologic Oil And Gas Reservoir Thin Layer Identification Based On The Spectral Decomposition

Many lithologic oil&gas fields in China are thin reservoir oil and gas reservoirs. The seismic record of thin-bed below 10 meters have lower resolution, so how to improve the resolution of thin reservoir has become a research focus in the geological exploration. Seismic signal is non-stationary, time-frequency analysis is an important means to process non-stationary signal, which is used in the thesis to decompose time-frequency spectrum of seismic signals to research thin layer identification of reservoir.Firstly, the theoretical and geological models of thin layer are discussed in the thesis, four kinds of wedge models of reflection coefficient are constructed. The wedge model is studied and the reflection characteristics of thin layer seismic are summarized in time domain, frequency domain and time-frequency domain.The S transform is mainly studied, the relationships among the S transformation, the Fourier transformation (FT), the short time Fourier transform (STFT), the wavelet transform (WT) and their implementation algorithm are analyzed. The time-frequency resolution of the STFT, WT, ST transform are compared to obtain spectrum of wedge model, which can be used to distinguish thin-bed by extracting single frequency. The test results show that the thin-bed resolution ability of STFT is improved with frequency increases, but the thin layer resolution will decline if the frequency is too high. WT has the similar characteristic with STFT, but discontinuous will appear in single frequency profile. The S transform can adjust adaptively the time and frequency resolution with the change of signal frequency, so it has better thin-bed distinguish ability. Therefore, S transform method is mainly researched in the thesis.Aiming at shortcomings of the GST parameters selected by experience, a new parameter p in the S transformation is introduced to adjust the standard deviation of window function. Combined with time-frequency concentration measure criterion putted forward by LJ.Stankovic to determine the optimal value of p, adaptive generalized S transform is proposed, which adjusts adaptively parameters in generalized S transform according to the characters of signal energy and avoids the influence of artificial factors.The range of parameters p is discussed to verify the nondestructive reversibility and single frequency resolution. Compared with ST and GST, the test results show that the AOGST has good energy concentration performence. AOGST algorithm is applied to synthetic seismogram, wedge model and actual seismic signal and has better thin-bed distinguishment.Combining with the information entropy theory,a algorithm of time-frequency energy spectrum entropy is discussed combined with information theory in the thesis, which measures the total characteristic of spectrum by computing the energy entropy of time-frequency spectrum. The energy spectrum entropy od seismic signal is obtained by S transform and the noise tolerant ability is improved by extracting the peak point. The results demonstrate that the location of peak point isn’t influnced by noise when the SNR is high and the location is protruded. When theR is low, the location is declinational.The constrained algorithm is proposed. The location of the peak point of energy entropy can reflect the location of reflector, which can be used to detect the location of Seismic synclastic axle. The performance ofthree kinds of entropy algorithm based on S transform, GST and AOGST respectively are analyzed and compared, and used to analyze the synthetic record,wedge model and real seismic signal, the results demonstrate that the algorithm can detect the location of synclastic axle effectively and distinguish the thin-bed.Hilbert method is influenced seriouly by noise and negative frequency maybe included, the method of instaneous attributes extraction based on S transform is discussed. BecausetThe peak point of instaneous amplitude corresponds to the location of reflector, a method based on instaneous amplitude is put forward to distinguish thin-bed. The S transform of seismic signal is calculated and extract the peak point of instaneous amplitude to detect the location of synclastic axle. The three kinds of instaneous amplitude extraction methods based on S transform,GST and AOGST are compared and used to the synthetic records and real data, the detection effective is obvious.