重复中位数估计Robust算法在长周期大地电磁中的应用

摘要

With the broadband signal, weak signal energy and small magnitude in comparison with the tradition MT, the long-period magnetotelluric survey has a higher data processing requirement. In this paper, the Robust algorithm based on repeated median estimator was applied to long-period magnetotelluric impedance tensor estimates. Compared with the Robust maximum likelihood estimate ( M estimate) , the repeated median estimator has no need of reset parameters in data processing, and can increase the estimate of the breakdown point up to 50% and get an unbiased estimate of impedance tensor. An application example shows that the Robust algorithm based on repeated median estimator has a strong advantage in the solution of the impedance tensor.%将基于重复中位数估计的Robust算法应用于长周期大地电磁阻抗张量估算中.与传统Robust估计(M估计)相比,重复中位数估计求解过程不需要重设参量,并且提高了算法对时间域原始资料中噪声干扰的容限,使算法的理论崩溃值从30％提高到50％,也能得到准确的估算结果.龙门山地区实测资料的处理结果表明,基于重复中位数估计的Robust算法应用于阻抗张量的求解具有较强的优越性.