Seismic Wavelet Phase Influence On Reflection Coefficient Sequence Inversion And Its Correction Method

The accurate extraction of seismic wavelet has profound significance for deconvolution,wave impedance inversion and forward model,but the seismic wavelet phase is often difficult to estimate accurately in existing seismic wavelet extraction method.The conventional solution is that implementing wavelet phase correction for the seismic records after deconvolution in order to eliminate the influence of residual wavelet phase.But caused by a lack of understanding of correction object's properties,the limitations of correction methods,the results of traditional wavelet phase correction methods are not ideal.The thesis intensively researched the influence of seismic wavelet phase for reflection coefficient sequences inversion,wavelet phase correction method,and extended it to time-varying seismic record processing field,finished the following innovative work:To research the influence rule of seismic wavelet phase for reflection coefficient sequences inversion and determine the correction object characteristics,this thesis adopts Autoregressive Moving Average(ARMA)model to describe the seismic wavelet,and constructs the wavelets with the same amplitude spectrum and different phase spectrums by symmetrical mapping Pole-Zeros of ARMA model in z domain.The deconvolution was implemented by using constructed wavelets for synthetic seismogram,the influence rule of wavelet phase for inversion is analyzed and summed: a pure-phase filter was remained after reflection coefficient sequences inversion in the condition of inaccurate seismic wavelet phase estimation,and the phase spectrum of the phase-only filter was the phase spectrum difference between real wavelet and constructed wavelet.Simulation and actual seismic data processing results verified the rule of the wavelet phase influence on reflection coefficient sequences inversion,the real wavelet inversion results were identified by the evaluation criteria of Maximum variance norm and Variation.In order to solve the defect of traditional seismic wavelet phase correction evaluation criteria's sensitivity to Gaussian noise,a wavelet phase correction evaluation method based on high-order cumulants(HOC)zero slice was proposed.Based on comprehensive comparison of the common evaluation criterion's anti-noise ability,and setting up the noise deconvolution model,from the "blind" characteristic of HOC for Gaussian noise,the feasibility inference of adopting the HOC's zero slice of wavelet correction results as the wavelet phase correction evaluation criterion was given to enhance the anti-noise ability of wavelet phase correction method.Simulation and actual seismic data processing results showed that the criterion could evaluate the wavelet phase correction results effectively in seismic record with noise pollution.To solve the problem of low accuracy in traditional wavelet phase correction methods' results,a linear and nonlinear combination wavelet phase correction method was proposed.The method firstly adopts matrix equation method of HOC's tri-spectrum(linear method)to estimate the wavelet phase residual,then normalizes amplitude spectrum to construct phase-only filter which can traverse arbitrary causal and phase characteristics,and uses phase-only filter optimization method(nonlinear method)to accurately estimate the wavelet phase residual,eventually uses spectrum division to realize the wavelet phase correction.This method breaks through the constant phase assumption of traditional wavelet correction methods,it can be applied to any causal and any phase characteristics of wavelet phase residual correction,it can improve calculation efficiency under the premise of ensuring the correction precision.Simulation and actual seismic data processing results showed that the method could effectively correct the residual phase wavelet and had a good ability to resist noise.The linear and nonlinear combination wavelet phase correction method was used to correct time-varying wavelet phase residue in order to solve the problem of time-varying wavelet phase correction in non-stationary seismic record.Based on time-varying seismic wavelet model of seismic record,the time-varying wavelet phase influence rule of correction processing was researched,the defect of direct using the linear and nonlinear combination wavelet phase correction method was analyzed,and a piecewise time-varying wavelet phase correction method was proposed.The method divided the time-varying seismic record into several overlapping segments of isometric,the linear and nonlinear combination wavelet phase correction method was used to estimated the wavelet phase residual in each segment.The estimated phase residual is continuated to each time point by polynomial fitting method,original reflection coefficient sequence is recovered by phase deconvolution method based on linear time-frequency transformation.Simulation and actual seismic data processing results showed the validity and feasibility of the method.Through the above research content,the influence rule of the wavelet phase inversion results was determined,a phase correction evaluation criterion which has good resistance to Gaussian noise was provided,a linear and nonlinear combination wavelet phase correction method was proposed,a wavelet phase correction method in non-stationary seismic record was preliminarily discussed,and a new way for high resolution seismic records will be provided.