Acoustic Impedance Inversion Based On Wavelet Edge Analysis And Modeling

At present, oil and gas exploration is an international problem. How to use theseismic exploration to describe the subsurface geologic structure more real anddetailed, and to search for the underground oil and gas reservoir, has always been ourgeophysical workers’ goal. This paper analyzes a method for improving the resolutionof seismic exploration, and applied to the actual seismic data.The technique based on the model of wave impedance is the main stream ofseismic inversion technology. The error doesn’t accumulate with depth. Inversionresults relatively with high precision and wide frequency band. This technique hasgradually replaced the inversion method based on the deconvolution. However, theconventional wave impedance inversion method based on model often appear theapplication problem of the following:①When the deviation between initial modeland the actual geological model is big, the inversion results are often not convergent,and can’t get the right results;②Accuracy and resolution of the inversion resultscan’t meet the requirements of reservoir prediction;③The inversion results presentthe instability and multiplicity. At the same time, because the seismic data can onlyprovide the medium frequency wave impedance parameters, the low frequency andhigh frequency components mainly acquired from logging, geological data. To solvethe above problem is the key to improve the effect of wave impedance inversion.Therefore, this paper proposes wave impedance inversion technology based onwavelet edge detection and modeling, the acoustic impedance and the characteristicsof seismic data acquired directly from seismic data itself as a constraint condition, andparticipate in the establishment of the initial model, to avoid the negative influencebrought by inaccurate initial model. So this method can solve above problem well.The traditional edge detection method is based on the space operation, with thehelp of the airspace differential operator using convolution, mainly have the effect ofthe high-pass filter, such as Sobel edge operator, Roberts edge operator, gauss Laplaceedge operator, etc. These operator due to only use the change rule of the first or thesecond order directional derivative near the edge, extremely sensitive to noise, and theability of removing noise is poor, and will attract the noise the when detect edge. Thewavelet operator using the multi-scale edge detection, and it’s scale function is gaussequation. When carries on the edge detection, use the derivative of gauss equation tofilter (convolution) of the original image, in the process of convolution to take a sidelength N for the square window, the derivative of gauss equation with infinite longtrailing to avoid excessive cut to the tail, the window can’t be too small. When the window’s radius increase, the convolution template will increase exponentially, resultto computation of convolution is too large, the position is not accurate enough,influence the practicality of the algorithm. The scale parameter decreases gradually,wavelet transform can analyze the local structure of the signal, has the ability to detectlocal mutation, is a good tool for edge detection. Wavelet transform based multi-scaleedge detection method, namely: using large scale of image filtering, restrain the noiseeffectively, to maintain reliable edge; filter using small scale on image filtering, makeup the large scale filter caused by edge loss, and improve the positioning accuracy ofedge detection. Using matching or edge connection technology in the different scalesedge image, from large scales to the small scales of edge focusing, obtain the clearedge.Due to the effect of absorption and attenuation of the earth, the seismic waveletis usually contains only the low frequency part, as a result of seismic geophone design,it usually can only detect the data part of more than8Hz, and it is the purpose of waveimpedance inversion to show full frequency range of the formation. So we needcombined with well logging data and lithologic data to guide inversion, and fill thewhole frequency band, to achieve high resolution of reservoir prediction.In this paper, the wavelet transform modulus maxima edge detection technologyas the theoretical basis, for the mutation point of seismic record, to modify the initialmodel and participate in the model perturbation gain the characteristic parameters ofseismic data. Seismic characteristic parameters are extracted from seismic data itselfreflects the local variation of lithology, the feature parameters can not directly foundwith the naked eye in the seismic and real hidden character. These seismiccharacteristic parameters for modeling and participate in the iterative inversion, andthese points are optimized. Finally, calculation of wave impedance inversion, theresult is more close to the actual geological condition.Results of the calculation model and real seismic data show that this method haspractical application value. The inversion results of wavelet edge detection modelingof wave impedance inversion technology can improve the seismic data, reducingambiguity.