Application Of Time-frequency Analysis Technology In Thin Layer Inversion

The Mesozoic and Cenozoic continental oil-bearing and coal-bearing basins in the eastern and western parts of China are mainly composed of thin sand and mudstone deposits,mixed with a small amount of shale,carbonate rocks and gypsum rocks.The lithologic changes and even the lateral changes of thickness of the above strata are relatively large,and the thickness is far lower than the vertical resolution of current seismic technology,and exists in the form of thin layers or thin interbeds.Therefore,prediction and inversion of thin reservoirs have become one of the main means to realize fine exploration of lithologic reservoirs.Conventional seismic inversion technique is based on Zoeppritz equation with single impedance difference interface hypothesis and its approximation,without considering the wave field aliasing effect of thin layer top and bottom reflection,interlayer multiple and converted wave,and directly applying it to thin reservoir inversion will cause noticeable errors.The reflection of thin layer is not only related to the impedance difference between the top and bottom interfaces of thin layer,but also related to the thickness of thin layer and the frequency of incident wave.The frequency dependence of seismic response of thin layer makes the study of thin layer more complicated.In addition,there are many inversion parameters in a single thin layer,so it is necessary to obtain nine elastic parameters(P-wave velocity,S-wave velocity and density)and thickness parameters,and the multiplicity of solutions has become a major problem hindering the inversion of thin reservoirs.Time-frequency analysis method is one of the important means of reservoir prediction,which can accurately extract time-frequency information from seismic signals and improve reservoir prediction accuracy.This paper discusses the advantages and disadvantages of several common time-frequency analysis methods and applies them to thin layer thickness prediction.Through the test and analysis of theoretical models,the generalized S transform is selected as the time-frequency analysis method for thin layer thickness calculation in this paper,and the effect of generalized S transform under different parameters is discussed.Using the inflection point of time-frequency curve to calculate the thin layer thickness,the recognition accuracy is improved to one tenth of the dominant wavelength.Based on the thin layer thickness obtained by generalized S transform,the least square inversion method is used to realize the multi-wave joint inversion of thin layer elastic parameters.After testing four typical thin-layer models,the method of combining time-frequency analysis with least square method to obtain elastic parameters and thickness of thin-layer has obtained better inversion results.By comparing different initial models,iterative error limit of inversion system,PP and PS wave weighting coefficients,the optimal solution of inversion effect under different models is obtained Inversion results under different thickness conditions show that the method is not suitable for extremely thin layers,and the error is large.The anti-noise test shows that the anti-noise of the program still needs to be strengthened,and the accuracy of wavelet extraction is required.