Research On Seismic Data Spectral Inversion Method

The oil and natural gas are important strategic resources for national development. With the deepening of oil and gas exploration, the thin reservoir prediction has become increasingly important. To predict the thin reservoir whose thickness is less than the native resolution of the seismic data and improve the accuracy of the prediction of thin reservoir has become the focus of people’s attention. Aiming to improve the thin reservoir prediction accuracy problems, the paper studies the spectral inversion method for seismic data in detail. This method is an inversion method based on the technology of seismic signal spectral decomposition that developed in recent years. Through establish the objective function between the actual reflection coefficient spectrum and the calculated spectrum, the method can inverse the value and location of the reflection coefficient. The result of spectral inversion can recognize the thin reservoirs whose thickness are less than tuning thickness because of the time resolution of reflection coefficient profile is far higher than seismic data profile.Under the condition of the spectrum of seismic data and wavelet being known, we derivate the objective function of spectral inversion in detail according the odd and even decomposition principle with the seismic sampling interval as the time thickness of thin layer and two adjacent sampling points as the top and bottom interface of a thin layer. We chose the more stable and accurate LSQR method to inverse the objective Matrix equation after comparison and analysis to get the reflection coefficient value of every sampling point finally. The spectral inversion method will get the size and polarity of reflection coefficient at the point when the sampling point is a real reflection interface or the result will be close to zero. Finally we apply the spectral inversion method in the process of post-stack seismic data successfully and get the reflection coefficient of underground reservoir accurately to achieve the purpose of improving the resolution of seismic data.The seismic wavelet must be known if we use this spectral inversion method to process the seismic data. But the wavelet will change with the time and space in real situations. Existing conventional wavelet extraction methods are mostly assume a constant phase seismic wavelet which result in the inaccuracy of estimated seismic wavelet spectrally the phase. The higher-order spectrum of seismic data include the information of both amplitude and phase of wavelet and can suppress the additive Gauss noise to highlight the real data at the same time. The paper estimates higher precision seismic wavelet using the Higher-order spectrum of seismic data for the accuracy of spectral inversion results to study the Bispectrum and Bicepstrum wavelet estimation methods respectively. The results of estimated wavelets by simulation and real seismic data show that the Bicepstrum method which can get a more accuracy wavelet.To verify the effectiveness of the method in paper, we have tested the synthetic seismic data of different simulation models and analyze the factors such as the frequency band of spectral inversion, SNR and so on which can impact the spectral inversion effect at the same time. The results show that the spectral inversion method can get fine output result within a certain range of SNR and a suitable frequency range. Finally, we processed the seismic data of a sea where Natural Gas Hydrate occurs using the spectral inversion method of the paper to get the reflection coefficient profile of Seabed strata and determined the top and bottom of the Natural Gas Hydrate reservoir with the help of BSR. The delineated Natural Gas Hydrate reservoir area is consistent with the super-speed and super resistance area of logging information which indicates the good effect of this method in the delineation of reservoirs.