用四阶累积量子函数改进剩余静校正量的计算

Improvement of calculating residual statics by fourth-order cumulant subfunction

(1.西南石油大学资源与环境学院,四川成都610500;2.成都理工大学四川省数学地质重点实验室,四川成都610059)

School of Resources and Environment,Southwest Petroleum University,Chengdu 610500,China

针对复杂山地地震勘探中常规剩余静校正方法在处理低信噪比和大时移量的转换波资料时出现的“周波跳跃”现象,引入了四阶累积量子函数进行时间延迟估计。利用常规最大能量法对包含高斯噪声的低信噪比、大时移量模型地震数据进行了试验,对比分析了应用互相关函数和四阶累积量子函数进行时间延迟估计的效果。模型试验发现,互相关函数对模型地震资料处理效果较差,得到的时间延迟出现整体漂移,构造成像出现严重的“周波跳跃”现象,与模型构造不符;而四阶累积量子函数处理效果较好,得到的时间延迟与期望输出几乎完全吻合,符合模型构造。为了进一步说明问题,对低信噪比、大时移量的转换波实际数据进行了试验,对采用上述两种函数估算的结果进行了对比分析,发现两种方法得到的叠加剖面成像效果都较好,进一步对中层和深层进行比较发现,利用互相关函数估算的结果出现了串相位现象,而利用四阶累积量子函数估算的结果更符合实际地质构造。上述试验证明了四阶累积量子函数能很好地压制高斯噪声,特殊资料(含有高斯噪声)和一般资料的应用均能取得较好的效果。

During seismic exploration in complex mountain area,“cycle jump” appears when using conventional residual statics method for processing converted wave data with low S/N and large time-lapse.In order to solve the problem,fourth-order cumulant subfunction was introduced to estimate time delay.Conventional maximum energy method was used to process seismic data with low S/N and large time-lapse including Gauss noise.The result of time delay estimation by applying cross-correlation function and fourth-order cumulant subfunction was compared and analyzed.Model testing shows that cross-correlation function processing has bad result displaying as the obtained time delay appears overall shift and structure imaging result shows serious “cycle jump”,which does not coincide with model structures.However,the fourth-order cumulant subfunction processing has good results;the obtained time delay and expected output are almost the same,which coincides with model structures.Furthermore,the above two methods were applied on real converted wave data with low S/N and large time-lapse,the obtained stack sections have good imaging quality.But the comparison at middle layer and deep layer indicates the result from cross-correlation function exists cross event phenomena,and the one from fourth-order cumulant subfunction is more accord with actual geologic structures.The test proves fourth-order cumulant subfunction can well suppress Gauss noise,application on conventional and special data (with Gauss noise) both can achieve good results.