The Signal-noise Ratio Weighted Stack By Wavelet Division Frequency

Processing of seismic data is base on stack. When there is better stack section, the all kinds of follow-up raising processing of resolution and signal-noise ratio has the meaning, if the signal-noise ratio of the poststack section itself is lower and after the original seismic record is stacked, the fidelity of it become bad, the resolution lower, so raising resolution and signal-noise ratio to such stack section in the follow-up processing is very difficult, even what is raised is false resolution. At present the stack exist a lot of disadvantage, it is very hard satisfied with the precision requirement of high resolution processing, and studying a kind of more perfect stack is very necessary.Doing the further research on the division frequency stack on the foundation of study of predecessor, at first the residual moveout correction is done to the seismic record, make the phase axle regularity, doing foundation for the same phase stack; then carried on division scale processing by wavelet transform; spectrum whitening is done to each scale, the high and low frequency band need to do prolongs, the middle frequency bands only increase the value of the frequency spectrum; several kinds of methods that estimate the value of signal-noise ratio has been studied further, and summarize their advantage and disadvantage as well as the scope of application; the seismic record after spectrum whitening is stacked by weighting with the value of the signal-noise ratio; then estimate the value of signal-noise ratio which is each scale section after stacking, the scale that the signal-noise ratio is big is assigned big weighting, otherwise, the scale that signal-noise ratio value is small is assigned small weighting, and carried on weighted reconstruct to each scale section. This can increase efficiently the signal-noise ratio of seismic record, under the general situation, the signal-noise ratio of big scale is small, because low frequency surface wave in big scale is more serious, the correlate of signal is smaller, so weighting stack can press efficiently low frequency surface wave, comparative promote the percentage of the effectual signal in constructing the signal.In this paper, wavelet analysis is applied in stack at the first time, instead of Fourier transform, wavelet transform is applied in frequency division processing of seismic record, no matter in high frequency or low frequency window, the waveform of wavelet function is unchanged, and only change its amplitude and width, the waveform of shorten Fourier function is changing in different frequency band. For seismic wavelet, wavelet function is more approximate, and the time-frequency location ability of wavelet transform is far higher than that of Fourier transform. To this paper we were carried tentatively on spectrum whitening in prestack, and each section after division scale is processed with spectrum whitening, particularly the composition of high frequency is raised apparently, the signal-noise ratio of high frequency raise. The resolution of whole section is better than stack.