同步压缩小波变换分频谱蓝化在薄层识别中的应用

Application of frequency-division spectral blueing via synchrosqueezing wavelet transform in thin layer recognition

传统谱蓝化拓频技术由于谱蓝化算子的单一性,蓝化拓频后频谱分布异常,从而导致地震资料高频段存在不合理的现象;同时,蓝化拓频后频谱低频段容易出现震荡,导致拓频效果不佳。为此,提出基于同步压缩小波变换的分频谱蓝化拓频技术。该技术首先利用同步压缩小波变换对频率的分辨率优势,将地震数据精确地分成多个分频数据体;然后,在分频段计算谱蓝化算子过程中,考虑不同频段振幅谱特征,提出一种基于不同频带能量差异加权的改进谱蓝化算子;再依据不同分频段振幅谱占比确定其对应的改进谱蓝化算子权重;最后,将改进谱蓝化算子与地震反射系数褶积,得到优化后的拓频数据。模型试算及实际地震资料测试结果表明:相较于传统谱蓝化技术,联合同步压缩小波变换分频及谱蓝化算子优化进行地震拓频,较大程度提高了地震资料的高频信息,拓频后地震数据分辨率更高,在保证薄层精细刻画下,其低频段信息也得到有效保留,视觉假分辨率得到一定程度的压制。

Owing to the unique operator of the traditional spectral blueing frequency extension technique,the blueing spectrum distribution is often abnormal and the high-frequency portion is unsuitable.Additionally,because it is easy to cause oscillation in the low-frequency region,this results in a poor frequency extension effect.Therefore,a new frequency-division spectral blueing technique,based on the synchrosqueezing wavelet transform,was proposed.First,seismic data are accurately divided into several frequency dividers using the synchrosqueezing wavelet transform,which is adept at frequency processing.Meanwhile,considering the spectral energy discrepancy of different frequency dividers and the aim of improving the accuracy of the blueing operator,an optimized blueing operator based on the weight of different spectral energies was proposed.In this model,the weight factor is determined by the amplitude spectrum proportion of different frequency dividers,which is convoluted with the seismic reflection coefficient to obtain optimized seismic data.The model trial and actual data analysis show that,in comparison with the traditional spectral blueing method,by combining the synchrosqueezing wavelet transform with blueing operator optimization,the middle-high frequency information and seismic resolution can be notably improved.Furthermore,under the guarantee of fine characterization of thin layers,the low-frequency portion is effectively preserved,and the visual false resolution is suppressed to a certain extent.