大地电磁测深资料处理中的奇异值分解法

Singular value decomposition in magnetotelluric sounding data processing

本文详细介绍了一种大地电磁测深资料处理方法——奇异值分解法,并导出了它的实数运算形式。该方法的特点是能把数据矩阵分解为信号矩阵和噪声矩阵进行分析,它不仅能给出对响应函数的最小二乘估计,而且能给出对数据噪声的定量分析、计算出信噪比,并给出对响应函数估计的方差。通过对理论合成数据的计算表明,该方法在抑制噪声、作出对响应函数的估计,特别是在作出噪声及方差分析方面是一可靠而有效的方法。

Singular value decomposition method,a method forrnagnetotelluric sounding data processing,is reeommended minutely;and its real number operarion process is derived.For analysis,this method decomposes data matrix into signal matrix and noise matrix,It may give least squares estimation of response function,performs quantitative analysis of signal and noise to calculate S/N ratio,and offers the estimated variance of the response function.Theoretical calculation shows that this method is reliable and effective in suppressing noise,estimating response function,and analysing noise and variance.