【摘要】 表面多次波压制是海洋地震数据处理的关键环节,基于波动方程的多次波压制方法可由多次波模型预测和自适应相减两步实现.因此,除了采用有效的算法实现高精度的多次波预测外,多次波的自适应相减也是改善多次波压制效果的重要手段.文中采用多道自适应滤波器完成表面多次波的自适应相减,利用数据驱动的基于波动方程预测多次波模型的高频重建道、Hilbert变换道、以及它们相应的上下延拓道改善预测多次波模型的振幅、相位、旅行时以及频带信息,使得修正后多次波模型与实际地震数据中的多次波更好地匹配.实际数据测试表明,文中所述方法可有效地完成表面多次波的自适应相减.通过讨论分析交叠时窗和滤波器的长度对自适应相减效果的影响,表明合理地利用时窗以及选择滤波器长度将有效地改善自适应相减的效果.更多还原

【Abstract】 Multiple suppression is a key link of marine seismic data processing.Multiple suppression approach based on wave equation can be accomplished by multiple model prediction and adaptive subtraction.To compensate the difference between predictive multiples and real multiples,usually there is need to do adaptive subtraction.While considering to eliminate the energy of multiples,the amplitude of primary also need to be protected,when energy of multiples and primary are intersect,direct subtraction can not protect primary effectively.Comparing the predictive multiples predicted by SRME(Surface-related Multiples Elimination) approach with real multiples,at least convolve one more wavelet,leading that the frequency band energy of multiple wave field does not match with total wave field.So,except using effective algorithm to accomplish multiple prediction,multiple adaptive subtraction is also a very important way to improve multiple suppression effect.In this paper,we introduce the information of predictive multiple modeling trace,its high-frequency reconstruction trace,and Hilbert transform trace,also their corresponding up and down extension trace,then use multi-trace filter to realize surface-related multiple adaptive subtraction.Hilbert transform trace can improve the phase information of predicted multiple,also high-frequency reconstruction trace extend the length of predicted multiple frequency and band compensate high-frequency information,so that the redundance wavelet effect can be eliminated.By multi-trace adaptive subtraction,we improve information of amplitude,phase,travel time and frequency band of prediction multiple model,so that the modified multiple model can match multiple with real seismic data better.In this paper,we also discuss and analysis the effect of overlapping window and filter length on adaptive subtraction.During multiple adaptive subtraction,the length of filter is a vital parameter,the purpose of adaptive subtraction based on L2 norm is to make the residual data has least energy after suppressing multiples.Long filter can match with original data better,so more residual energy can be eliminated during subtraction.In the theory,when the length of filter is infinite,the residual energy will be zero,but if the length of the filter is too long,it will be difficult to protect the portion of primary energy contained in the residual energy.And when using long filter,if the events of primary and multiples are so close that maybe multiple model can match with primary.Raw data tests show,the approach discussed in this paper can realize surface-related multiple adaptive subtraction effectively,use window control and choose the length of filter reasonably can improve the effect of adaptive subtraction.更多还原