Abstract: The relationship of superposed formations in Es4 in the west of Dongying Sag is complex,the physical property varies acutely in both vertical and lateral directions,and the divergence of reflection feature is small,thus posing difficulties in discriminating sand body,especially its borders.In view of the foregoing problem and based on the theories and techniques relating to wavelet multi-scale analysis and higher-order statistics for seismic signals,the paper carries on the transformation of the seismic records using higher-order cumulants first,then the results are decomposed to multi-scales.To insure the reliability of the correlation analysis,proper multiscale seismic envelop results have been incorporated in the correlation analysis.The algorithm has been verified by theoretical modeling and actual data,especially in the aspect of capability of practical problem solving,for example,the originally obscure or weak seismic signals can be distinguished clearly after the processing using this method.As a result,the originally unrecognizable sand body and its borders become explicit.

Key words: wavelet multi-scale analysis, higher order cumulant, correlation analysis, prediction of sand body

摘要： 东营凹陷西部沙四段地层叠置关系复杂,物性纵横向变化剧烈,地震反射特征差异小,给砂体识别尤其是砂体边界的识别带来了困难。针对以上问题,文中提出了基于地震信号多尺度高阶累积量的相关分析技术。在高阶统计量分析理论和小波分析理论的基础上,首先对地震记录进行高阶统计量变换计算,然后对该变换结果进行多尺度分解。为了使相关分析结果更加可靠,在对上述结果进行相关分析计算时,适当地加入了计算多尺度地震记录的包络结果。对算法从理论模型和实际资料上给出了论证,特别是在解决实际问题方面,经技术处理后,使原来较模糊甚至很弱的地震记录有了清楚的显示。在平面上,原来的地震资料中几乎不能分辨的砂体及边界得到了清晰的反映。

关键词: 小波多尺度分析, 高阶累积量, 相关分析, 砂体预测