A high-precision time-varying wavelet inversion method

Abstract

The invention discloses a high-precision time-varying wavelet inversion method, including: 1. Taking the second-order cumulant autocorrelation of the seismic signal as the initial value of the inversion initial wavelet; 2. Selecting within the time range of the seismic signal The wavelet inversion time window length that is at least four times the wavelet length is divided into time windows; 3. Within each time window, use the well reflection coefficient in the time window to construct a reflection coefficient matrix; 4. Use the inversion of step 1 The initial wavelet is the initial value of the inversion iteration, and the error-decreasing iterative algorithm is used for wavelet inversion; 5. Follow steps 3 and 4 to traverse all time windows in order to complete the high-precision inversion of time-varying wavelets. The present invention does not require any assumptions on the phase of the wavelet, and creates a reflection coefficient matrix based on the exchange rate of the convolution, taking the best approximation between the forward modeling record and the actual earthquake as the objective function, and can invert any phase wavelet in each time window , and finally a time-varying wavelet is obtained.

Description

technical field [0001] The invention relates to the technical field of oil and gas exploration, in particular to a high-precision time-varying wavelet inversion method. Background technique [0002] Seismic wavelet extraction is an important part of seismic high-resolution inversion. At present, wavelet extraction is mainly divided into two types: well-seismic combined wavelet extraction and seismic high-order cumulative statistical wavelet extraction. Among them, the well-seismic combined wavelet extraction method uses the subtraction of the seismic complex spectrum and the reflection coefficient complex spectrum to obtain the wavelet complex spectrum, and uses the polynomial spectrum simulation method to simulate the wavelet amplitude spectrum for the wavelet complex spectrum, and the phase spectrum is usually It is artificially set; the seismic high-order cumulant statistical wavelet is a global statistical wavelet, usually the second-order statistical wavelet can only co...

AI Technical Summary

Problems solved by technology

Among them, the well-seismic combined wavelet extraction method uses the subtraction of the seismic complex spectrum and the reflection coefficient complex spectrum to obtain the wavelet complex spectrum, and uses the polynomial spectrum simulation method to simulate the wavelet amplitude spectrum for the wavelet complex spectrum, and the phase spectrum is usually It is artificially set; the seismic high-order cumulant statistical wavelet is a global statistical wavelet, usually the second-order statistical wavelet can only count the zero-phase wavelet, and the complex cepstrum wavelet estimation of the third-order and fourth-order statistics can be simultaneously Estimate the amplitude spectrum and phase spectrum of the wavelet. In the process of wavelet estimation, it is necessary to estimate and invert the coefficients of each phase harmonic wavelet, and use the harmonic wavelet to construct a mixed phase statistical wavelet, but the algorithm is sensitive to the length of the earthquake. It is necessary to solve the overdetermined ill-conditioned equations with the harmonic wavelet as the target, and it is difficult to solve the objective equation

[0003] Affected by stratum heterogeneity and viscoelasticity, the actual seismic wavelet has time-varying characteristics. The traditional seismic wavelet extraction is mostly based on the assumption of time invariance, which often cannot meet the target requirements for actual exploration. The quantum wave extraction method of order statistics requires a large time range for calculation, and it is difficult to ensure the stability of large-scale seismic data; while the spectral simulation method assumes that the simulation is similar to the smooth unimodal curve of the Reker wavelet, and cannot accurately estimate the wavelet phase spectrum

Images