子空间增强在大地电磁信噪分离中的应用研究

Application research of subspace enhancement for magnetotelluric signal to noise separation

为了提高强干扰地区大地电磁测深数据质量、保留大地电磁信号低频段的有用信息,压制大尺度强干扰已刻不容缓。针对矿集区典型的强干扰类型,引入子空间增强方法对大地电磁信号与强干扰进行信噪分离研究。通过计算机模拟常见的大尺度方波和充放电三角波干扰,讨论了子空间增强算法中帧长与特征值判别阈值的最优选取范围,给出了算法流程,并对矿集区实测大地电磁数据进行了信噪分离处理。实验结果表明:该方法是切实可行的,能有效剔除时间域序列中的大尺度强干扰,近源效应得到了有效压制;经处理后,重构的大地电磁信号中包含了更为丰富的细节成分,视电阻率曲线更为光滑、连续,低频段的大地电磁数据质量得到了明显改善。

In order to improve the quality of magnetotelluric sounding data and retain the useful information of low frequen-cy band for magnetotelluric, suppressing the large -scale strong interference is imperative. Aimed at typical strong interference types of ore concentration area, we introduced the subspace enhancement method to study signal to noise separation of magne-totelluric data and strong interference. Through simulated large -scale square wave and charge and discharge triangle wave by computer, we discussed the optimal range of frame and criterion threshold of eigenvalue in subspace enhancement, and the con-crete steps of the algorithm are presented. Then, the measured magnetotelluric data of ore concentration area were processed by signal to noise separation. Experiment results show that the method is feasible, which can effectively eliminate the large -scale strong noise interference in the time domain and the near source interference of ore concentration area has been effectively sup-pressed. At the same time, the apparent resistivity curve is more smoothly and continuously. Moreover, the quality of magne-totelluric sounding data is improved significantly in the low frequency band.