In the magnetotelluric (MT) method, long recording times are required to obtain usable readings at frequencies below 1Hz. However, due to the long recording times, signals may be contaminated by coherent noise such as signals generated by DC trains. If low frequency MT data are contaminated by such noise, it was difficult to remove or segregate the contaminated data from the usable data. In this paper, we show the validity of the new weighted stacking method proposed by Negi et al. (2010) using survey data. We used four magnetotelluric sites located between northern and southern regions of the Noto peninsula, central Japan, where MT signals are usually contaminated by coherent noise. In this region, DC train signals are dominant over long periods of time. However, it was easy to estimate the true signal spectrum because the DC trains do not travel late at night. We processed the data acquired at these sites by the weighted stacking method. As a result, we were able to decrease the influence of the low quality data acquired in the time slot dominated by coherent noise. In similar circumstances, it is expected that the weighted stacking method following the proposed algorithm can be used to edit the MT spectrum data efficiency and effectively.
View full abstract