Improvement of S/N of Magnetotelluric Time

The magnetotelluric (MT) method is an efficient technique for mapping the deep subsurface electrical conductivity structure. The method assumes a linear relationship between the horizontal natural magnetic and electric fields at the earth's surface, over a broad frequency range. These electric and magnetic fields are commonly stable in time at specific areas and show smooth trend in recorded time series, so suddenly changes in amplitude can be an attribute of near surface noise. One way to suppressing noise is dividing the time series into segments and stacking together, but if the noise amplitude be high abnormally, then stacking progress can not canceling the noise properly. In this paper we introduce a new method for canceling the noise and improvement S/N of time series. This method is based on periodic component of MT signals which have 3 steps: 1. Selection the high s/n parts of time series (it is important to note that this selection needed high experience and expertise) and correspondingly removing the noisy parts. 2. Decomposition of time series into principal components which has a narrow band frequency spectrum and well defined characteristics to be estimated. 3. Each principal component (PC) is modeled by AR model.