Pervasive white and colored noise removing from magnetotelluric time series

Abstract

Magnetotellurics is an exploration method which is based on measurement of natural electric and magnetic fields of the Earth and is increasingly used in geological applications, petroleum industry, geothermal sources detection and crust and lithosphere studies. In this work, discrete wavelet transform of magnetotelluric signals was performed. Discrete wavelet transform decomposes signals into coefficients in multi-scales. Noise and signal portions are separable in multi-scale mode. Therefore, noise can be discarded in each scale; a threshold value is constructed dependent to coefficients of the scale then, the noise coefficients are discarded by thresholding the coefficients with the proper values. Proportional threshold values can be used to remove white and 1/f noise from time series. After that, a new signal is constructed using clean coefficients. This method is widely used in various fields of sciences from image processing to seismic studies. This work tried to show the effectiveness of this technique in decreasing pervasive noise from magnetotelluric signals. The results emphasized the advantageous effect of wavelet techniques in magnetotelluric data noise removing process.