小波包分析法压制井间地震资料噪声

Noise elimination for crosswell seismic data using wavelet packet transform

井间地震资料的随机干扰发育,由于随机干扰的存在,给后续的波场分离与反射成像带来很大困难,为此,必须进行噪声压制。运用coif3小波对两个相邻地震道进行同层次分解,在分解后的小波树中分别提取相同节点号的节点进行重构信号,并对此两重构信号做互相关估计,由于噪声的随机性,如果两重构信号相关系数趋近于零,返回该节点并对它的小波包分解系数进行阈值处理,以消除随机噪声成份。实际资料的处理试验表明,小波包分析法可以较好地压制井间地震资料的随机噪声。

Crosswell seismic data are rich in random noises, which lead to difficulties of wavefield separation and reflection imaging. In this paper, two adjacent seismic traces are decomposed by Coif3 wavelet of common levels, and then nodes of the same level in the resulted wavelet packettree are used to reconstruct the signal. Cross correlation is performed on the reconstructed signals. Considering the randomness of noises, the values of cross correlation should be close to zero. In such cases, we established a threshold for the wavelet packet coefficients of the corresponding nodes to eliminate random noises. The effectiveness of the method was demonstrated on real data.