Thin Interbed Analysis Based On The Characteristics Of Time-Frequency Spectral

How to describe the subsurface geologic structure much more truly and meticulous by seismic exploration and look for underground reservoir is the insistent struggling goal of geophysicist. Now it is a hotspot to research thin-layer or thin interbed. We has analyzed several methods of improving the resolution of seismic exploration and apply them to seismic data.Firstly, we has analyzed the characteristics in time domain and frequency domain of the reflection coefficient of thin-layer in theory, and designed some reflection coefficient models, then obtained frequency spectrum by doing FT to these models. When there are few reflection coefficients, the characteristics of frequency are very clear. Its amplitude spectrum is a period spectrum with multi-extreme value and we can estimate the thickness accurately by notches-in-thin-bed. This study has proved that the amplitude of thin-layer reflection wave is determined by the difference of their properties. The form of amplitude spectrum and frequency-extreme relate with the polarity of reflection coefficient. The frequency spectrum of reflection wave is determined by product of frequency spectrum of incident wave and reflection coefficient. When the more reflection coefficients and the more complex combination, the larger change about frequency spectrum form.Time-Frequency analysis is a very effective way to analyze non-stationary signal, it can obtain high frequency resolution in low-frequency and high time resolution in high-frequency. It can transform one-dimension time signal to two-dimension time-frequency spectrum. This paper has selected Generalized S Transform (GST) as time-frequency analysis tool, for it has higher resolution and better adaptability time window, and its length and size can adjust. For the different seismic data, we can obtain their time-frequency spectrum only through changing the parameters of time window. So the two-dimension time-frequency domain filter is much more flexible than normal frequency domain filter. In seismic cross-section because the frequency of seismic data is lower, after we do time-frequency decomposition for them, we can extract higher single-frequency section, the thin-layer can display more clearly and the complex structure can depict more fine.Spectral-decomposition technique was used widely for its capability of accurate thin-layer thickness estimate. This paper has referenced the theory which Ethan. J. Nowak etc. (2008) put forward and combined the P and G section which contain s-wave and p-wave information from the spectral response of AVO to carry on cross-spectral decomposition. For the wedge model, the number of periods of cross-spectral has increased obviously, its frequency-extrema and frequency-notch are more close to the frequency spectrum of reflection coefficient, which becomes more helpful to inversion the thickness of thin-layer. The time resolution of thin-layer has been improved largely for the cross-spectral contained the information of p-wave and s-wave and it was nearly double times than auto-spectral decomposition's resolution. When we process cross-spectral decomposition using real seismic data, it is the first step to inversion the P-section and G-section from the AVO.This article has combined the study of thin-layer reflection coefficient with GST time-frequency analysis, reconstructed the high resolution seismic cross-section. At first, analyze the frequency spectrum of the seismic cross-section; secondly, do GST to this seismic data and choose suitable single-frequency sections; thirdly, compensate frequency to these single-frequency sections, that is, derivate these sections in time domain(equal to multiply (i w )2), in this way, it can keep the continuity of low-frequency and highlight the resolution of high-frequency; The last, do inverse Fourier transform to these frequency spectrum and then we can obtain the high resolution reconstruction section. The effect of this method in high resolution processing is clear after tested by artificial model and real seismic data.This paper has studied thin-layer and thin interbed. We get some good effect on improving the resolution of seismic and estimating the thickness of thin-layer. However, single method is not more real and suitable for processing seismic data, only by combining these advantages of several methods, can we reconstruct high resolution seismic cross-section and get satisfactory result.