Abstract:

As one of the main noise interferences of the magnetotelluric sounding system (MT), like-impulse electromagnetic noise not only affects the count of subsequent apparent resistivity and impedance, but also badly influences the target information extraction. In this paper，we briefly introduced the basic principles and algorithms of the empirical mode decomposition method (EMD), and based on the analysis of actual data, we discussed its application in magnetotelluric signal processing and noise suppression. First, we can get intrinsic mode function (IMF) after EMD decomposition from magnetotelluric signal; then, select a suitable threshold for each of the IMF and truncated the excess part; last, did EMD reconstruction. Test results of field data show that corrected signal is of high correlation coefficient and less energy loss compared with raw signal, and is a high accuracy approximation of raw signal. The like-impulse noise of field data can be restrained efficiently.

Key words: empirical mode decomposition, magnetotelluric, like-impulse electromagnetic noise