利用信噪分离方法提高大地电磁张量阻抗估算质量

在强噪声环境下，由于相关噪声的影响，准确地估算大地电磁张量阻抗越来越困难.研究一种处理相关噪声问题的信噪分离方法，即采用最小二乘Robust方法，利用相关噪声相对较小的远参考站磁场信号求取分离张量，将本地磁场的观测信号分离为MT信号和噪声，估算大地电磁张量阻抗.分别对模拟数据与实测数据进行处理, 结果表明，信噪分离方法能够分离出较强的相关噪声，给出的视电阻率和相位曲线比最小二乘Robust方法和远参考方法获得的更加平滑，处理效果明显.研究表明，在处理受相关噪声影响的MT数据中，信噪分离方法具有明显的优势. 

A Signal-Noise Separation Method for Magnetotelluric Impedance Tensor Estimate Improvement

It is difficult to estimate precise magnetotelluric (MT) impedance tensor in high correlated noise areas. In order to approximate MT impedance tensor, we use remote sites which contain relatively lower correlated noise to compute separation tensor by using Robust least square method. The MT data in three sites which are contaminated by higher and lower correlated noises are processed using signal-noise separation method. The results show that it can separate higher correlated noise and get much smoother apparent resistivity and phase curves than those obtained by using Robust least square method and remote reference method. It is concluded that signal-noise separation method is efficient for much more precise MT impedance tensor.