基于C-WNNM的地震随机噪声压制方法

Seismic random noise suppression based on C-WNNM method

针对低信噪比情况下地震信号同相轴不易识别的问题,提出用C-WNNM方法来压制地震勘探资料中的随机噪声.给出了基于WNNM的低秩逼近理论,利用地震信号在时间和空间上具有一定的相似性这一特点来构建近似低秩矩阵,并且由CEEMD分解得到的IMF1分量来近似估计局部噪声方差,从而获得更加精确的权值.经过迭代逼近得到最终去噪后的信号.对由雷克子波生成的模拟地震勘探资料进行C-WNNM滤波处理.结果表明,在地震数据存在强噪声的情况下,该方法能够有效压制随机噪声并且能够更好地保留有效信号,信噪比相比于原始WNNM算法提高了3 d B左右.

To solve the reorganization problem of seismic events with low signal to noise ratio,the weighted nuclear norm minimization（ C-WNNM） method was proposed to suppress the random noise in seismic data. The low rank approximation theory was presented based on WNNM. The low rank approximate matrices were constructed according to the similar characteristics of seismic signals in time and space domain. The IMF1 component obtained from CEEMD decomposition was used to estimate the local noise variance and lead to more accurate weights. The final denoising signal was obtained by iterative approximation. The C-WNNM method was used to process the simulated seismic exploration data generated by ricker wavelets. The results show that the proposed method can remove random noise effectively and preserve the effective signals well in the presence of strong noise in seismic data. The signal to noise ratio filtered by C-WNNM is promoted by about 3 d B than original WNNM.