In the target strata of Guanjiabao area, developped many thin stages of sand-shale sequences in vertical orientation and these thin sandbodies are of great lateral variation. To predict the spatial distribution of the thin oil layers, a new method was applied: to extract attributes from prestack seismic data and to inverse constrained by well-logging data.The paper discusses the foundamental theory about seismic attribute and its application from chapter 1 to chapter 3.In chapter 4, as a new technique applied in Guanjiabao reservoir prediction, prestack inversion is introduced in detail.To broaden the frequency band of effective signals and improve the signal-to-noise ratio of prestack seismic data, amplitude-kept preprocessing and supergather sorting techniques are applied in T-X domain, which, to the most extent, keep the geophysical characteristics from lose.Catching more geophysical information about media underground,attributes derived from generalized Radon domain can depict the travelling of wave and indicate its dynamic characteristics.By selecting samples of high quality, the learning and extraplating technique of Support Vector Machine keeps the accuracy of training results between logging information and seismic attributes as well as enhances reliability of the extraplating result of log information.The well data standardized is reliable and would contribute to power balance of the inversing result.Attributes are extracted from the 3D prestack seismic data. Four geophysical parameters of the target strata: porosity, clay content, water saturation and electrical resistivity are inversed. The parameter value calculated has the same tendency withlogging data and matches it well. The contour maps of the four parameters are made out. The Guantao Formation near zh8 and Shahejie Formation between zh4xl and zh8 develop productive reservoirs. A conclusion could be drawn that prestack inversion is an effective method to sandy reservoir research in this area.
|