基于Gabor变换反褶积技术在渤海某工区的应用研究

Research on the application to seismic data in Bohai bay based on Gabor deconvolution

使用常规的Wiener反褶积必须假设震源子波在地层旅行过程中是平稳的即一成不变的,这个前提条件与实际野外地震资料采集差别较大,而基于Gabor变换反褶积技术考虑到地震能量的衰减、子波的形变等非平稳性特征.地震道在Gabor域可因式分解成三项即震源子波、衰减函数和反射系数,该技术设计POU窗函数,并利用此函数在Gabor域对地震信号进行局部时频分解.Gabor域反褶积算法在Gabor域通过除以衰减函数和震源子波的乘积来估算地层反射系数,然后再做Gabor反变换可求得时间域的地层反射系数.理论模型的测试和实际地震资料的应用均表明,与Wiener反褶积相比较,基于Gabor变换反褶积可补偿中深层的能量衰减并因此拓宽有效频带和提高时间分辨率.

Considering the fact that the source wavelet will change including its shape as the wave travels through the earth because of the stratigraphic effects such as anelastic attenuation. Conventional Wiener deconvolution, which assumes that the wavelet is stationary namely it does not evolve with traveltime and the reflected pulse travels back to the surface without distortion, is at a disadvantage during its application. But instead Gabor deconvolution which allows the seismic wavelet to evolve with time expresses nonstationary convolution as a generalization of stationary convolution by approximately factoring the nonstationary convolutional model into a product of three terms：source wavelet, attenuation function, and the Gabor transform of the reflectivity, POU function is defined during Gabor deconvolution, using POU can decompose recorded seismic signals into a time-frequency plane namely local Fourier transforms, Gabor deconvolution is achieved by direct division in the Gabor domain, the Gabor spectrum of the seismic trace is divided by the estimates of the attenuation function multiplying the source wavelet, and then the time-domain reflectivity is accomplished by an inverse Gabor transform. Tests on synthetic and real data show that this method works well, in comparison with Wiener deconvoltion, Gabor deconvolution is able to broaden the bandwidth and improve the time resolution due to the fact that it can adjust the amplitude within the deep stratum.