【摘要】 根据地震剖面上不同的频段和时段有不同的信噪比，以及相邻地震道反射波有效成分（除去噪声的有用信号）在波形和能量上具有较强的相关性的特点，本文提出了应用小波包变换和多项式拟合理论消除噪声和提高信号分辨率的方法。该方法的内容是：先将每道地震记录进行小波包分解，形成地震记录分时分频小波包剖面；然后利用共深度点反射波相关的性质用多项式拟合进行信噪分离；再将各个小波包剖面重构为经去噪后的地震剖面。理论记录计算和实际地震剖面的处理结果表明：利用纵向地震道的小波包分解和横向多项式拟合的分时分频处理方法，能够有效地提高剖面的信噪比和分辨率。更多还原

【Abstract】 In the light of the fact that different signal/noise ratios occur in different fre-quency bands and different time intervals of a seismic section,and the fact that there are quite good correlations between the wave forms or energies of the effective waves of neighbouring seismic traces whose noises have been removed,we advance the method for improving signal/noise ratio and resolution by using wavelet pack-age trans form and polynomial fitting. The method includes the following steps: 1)decomposing the wavelet package of every seismic trace to form time interval and frequency-band wavelet package sections ; 2) separating signals from noises by using polynomial fitting and CDP reflection correlation .and 3)combining wavelet packagesections into one seismic section whose noises have been suppressed. Theoretical and real data processing results show that time-interval and frequency-band data processing method involving vertical wavelet package decomposition and lateralpolynomial fitting can improve both signal/noise ratio and resolution of seismic sec- tion.更多还原