Time-Frequency Domain Deconvolution Processing And The Research Of Thin Reservoir Recognition

Deconvolution processing is an important method to improve the quality of seismic data.Conventional based on the fourier transform to realize frequency domain deconvolution does not take the absorption and attenuation effects of seismic waves.In addition,the research on thin reservoirs has become a hot issue for scholars domestic and international,and how to improve the recognition accuracy of thin layers is the key to the problem.Therefore,this thesis focuses on the resolution of seismic data and the identification of thin interbedded reservoirs.Based on the non-stationary seismic record convolution model,the theory of dynamic deconvolution in time-frequency domain is established based on the high-precision time-frequency analysis method.A set of effective high-resolution processing and high-precision inversion seismic data processing method system has been formed.Based on the time-frequency domain analysis method,this paper compares the time-frequency focusing of time-frequency analysis methods such as STFT transform,wavelet transform and S-transformation,and adds the frequency adjustment factor to the S-transform to obtain two-parameter improved generalized S-transformation.S-transformation improves the time-frequency focusing and processing is more flexibility.Preferably,the improved generalized S transform not only can clearly depict the trend of signal amplitude with time and frequency,but also obtain multi-resolution time-frequency analysis results according to actual needs.By making use of the advantages of multi-resolution of improved generalized S transform,the time-varying wavelet extraction method for spectral simulation,time-varying wavelet extraction from quadratic Spectrum,And the method of extracting wavelet in cepstrum time-frequency domain are studied based on result of the time-frequency analysis.According to the wavelet extraction method,deduced the time-frequency domain spectral simulation deconvolution method based on the improved generalized S transform,the quadratic time-frequency spectrum deconvolution method based on the improved generalized S transform and the cepstrum smoothing time-frequency domain deconvolution method based on the improved generalized S transform.The one-dimensional seismic record model and the two-dimensional seismic record model were established to compare the deconvolution results of the three time-frequency domain deconvolution methods.Through the comparison of one-dimensional models,it indicates that the three deconvolution methods can recover the energy of the seismic wave attenuation to some extent without considering the influence of the Q value,at the same time,compressing the seismic wavelet to improve the resolution.Time-frequency domain deconvolution based on the improved generalized S-transformation can obtain more accurately deconvolution results and show a stronger thin layer recognition ability.Based on the model testing,optimizd the cepstrum smoothing time-frequency domain deconvolution to improve resolution of seismic data.Combining high resolution processing method with inversion technology to form high resolution inversion system.The c actual data processing and well-seismic analysis shows that the improved generalized S-transformation cepstrum smoothing time-frequency domain deconvolution method can improve the resolution of seismic data better.Impedance inversion performed on the basis of high-resolution seismic data can obtain a more accurate wave impedance inversion data,which can improve recognition accuracy of thin reservoirs and accurate reservoir prediction.