The Research Of Adaptive Spectral Modeling Deconvolution

With the development of lithologic exploration technique, precision and reliability that it calculated, dealing with to the seismic data have put forward higher request. The preserving relative amplitude, frequency, phase and waveform in seismic data handling are highlighted. Require that the amplitude not only reflect the correct position of reflection coefficient, but also reflect the relative variations of reflection amplitude, and expect the consistent corresponding relationship that true reservoir lithology information change and thin reservoir structural changes of reflection coefficient characteristics before and after processing is not damaged. The corresponding relationship between keeping degree directly affects the quantitative analysis of reservoir lithology and geological phenomenons explain accuracy. According to the deconvolution methods are concerned, Relaxing assumptions fundamentally and improving the adaptive of seismic deconvolution methods, which is one of the development direction to improve fidelity of deconvolution.Based on the analysis of spectral modeling deconvolution characteristics, researched parameter selection method of the traditional spectral modeling technology, summarized the optimization principle and with the form of numerical quantify to parameter selection provides guidance. Avoid a lot of lots of parameters experiment, to a certain extent guarantee the quality of the seismic deconvolution processing. Seismic data after energy compensation and stack processing ,at a certain extent changed amplitude spectrum shapes and broaden the frequency band of wavelet, after experiment find that the wavelet formula of traditional spectral modeling method can not good fit the variation of the real wavelet amplitude spectrum. Aimed for this limitation of the traditional spectral modeling method, combined the EMD method of Hilbert-Huang transformation developed for more than a decade .estimated residual tendency item extraction wavelet amplitude spectrum.Non-white noise component analysis of reflection coefficient and the normalized influence to estimate error of wavelet auto-correlation amplitude spectrum On the basis of formula derivation, We discuss the property of wavelet auto-correlation estimate in the condition of White noise assumption, which forms complementary strengths with Spectral simulation. combining the two method together, the technique of Adaptive spectral simulation can be generated. Whether white noise of reflection coefficient or not, the new method can estimate Wavelet amplitude spectrum Precisely, and improve adaptivity, stabilization and fidelity of Spectral simulation.