The Method Of Time-frequency Analysis Applied In Thin Interbedded Reservoir

Currently, Oil exploration mainly aims at thin interbedded reservoirs, it has become an urgent problem to determine the regularities of the spatial distribution of thin interbedded reservoirs and their characters. In eastern China, the vast majority of Mesozoic and Cenozoic continental oil basins are dominated by thin-layer sand and shale depositing, filling with a few thin-layer carbonate rock, shale and plaster layers. The lithology and thickness of the stratums varies largely in the transverse direction, and the thicknesses of these stratums are far less than the vertical resolution of conventional seismic exploration.In order to predict thin interbedded reservoir effectively, geophysicists had developed many new methods and technologies recently, and one of which was spectrum decomposition based on time-frequency analysis theory. Spectrum decomposition was independent of the phase of wavelet, hence more stable than wave peak-trough method presented by Widess.This paper analyzed the advantage and disadvantage of common time-frequency methods such as STFT, CWT, S transformation, WVD etc, and analyzed the characteristic of spectral of reflection coefficient series and its influence factors, emphasizing on the influence of time thickness of single layer, number of thin interbedded layers, magnitude of reflectance, and polarity of reflectance on spectrum, and we draw such conclusions:(1)The number of thin interbedded layers, magnitude of reflectance and polarity of reflectance could only affect the amplitude of the spectrum of reflectance, and the period of the notch was determined by the time thickness of single layer.(2)Through the testing of synthetic seismic signal and that of wedge-shaped model, the results shows that although WVD was highest in time-frequency resolution, it presents badly cross-term; to analyze thin interbedded layers, the window of STFT should be short, that is at the expense of the frequency resolution hence the window could not be too short, and there would be a strong direct current component when the frequency is zero; conventional Morlet wavelet was unsuitable to analyze thin layers, and an improved Morlet wavelet was adopted in this paper, by introducing an item to control the width of the frequency band, it avoided the difficulty in selecting parameters in three-parameters wavelet application, the results of the model testing indicated that improved Morlet wavelet could better analyze thin layers; the time-frequency resolution of S transformation was between that of STFT and CWT. Hence, in this paper, improved Morlet wavelet was applied as the tool for time-frequency analyzing, to save CPU time, a fast CWT was carried out in frequency domain.(3)The analysis of Marmosi2 model and actual data from DQ indicates: the CWT based on improved Morlet wavelet could not only better descript the information of thin layers, but also be used to detect the difference of lithology. For thin layer analyzing, the example shows that, it could optimally distinguish the thin layers when the dominant frequency of seismic signal matches with the scale of CWT, the higher the dominant frequency and the wider the frequency band, the higher the time resolution of decomposed frequency profiles; for the anomaly low frequency indication of oil or gas reservoirs, since oil or gas reservoirs could absorb high frequency, causing the seismic wave to attenuate, which certainly will present low frequency anomaly. By selecting appropriate scales, the low frequency anomaly could be highlighted on decomposed frequency profiles. However, since there are various factors determining the low frequency anomaly, we could take low frequency anomaly as one condition indicating the existing of oil or gas reservoir, but not the only condition.