Home – Home – Jornals – Russian Geology and Geophysics 2018 number Неопубликованное

A significant problem of the magnetotelluric sounding (MTS) method is the influence of local near-surface inhomogeneities, which distort MTS data across the entire frequency range. These distortions complicate the analysis and interpretation of the data and, ultimately, the retrieval of information about the studied deep structures. A widely used method for suppressing near-surface distortions is the normalization of MTS curves using spatial low-frequency filtering. The aim of the article is to evaluate the effectiveness of this approach, including through the use of weighting characteristics proposed by the authors. A simple geoelectric model of the earth's crust has been compiled, containing a conductive sedimentary cover, a high-resistivity basement, and a three-dimensional conductive depression in the basement. Two model variants were considered: with a homogeneous upper part of the section and with multiple local near-surface inhomogeneities. Using three-dimensional modeling, synthetic magnetotelluric sounding (MTS) data has been calculated for a system of profiles. The data for the second model variant exhibit the effect of near-surface distortions, which leads to a shift in the MTS amplitude curves by level. To suppress the distorting effect from near-surface inhomogeneities, we used normalization of the MTS curves with smoothing filters in a sliding window with different radiuses. Various weight characteristics of the filters were used, namely: for the distance of a point from the center of the sliding window radius, for the difference in amplitude at a given point from the average in the window, and for the difference in the principal directions determined from the polar diagrams of the modulus of the principal component of the impedance tensor and the principal component of the phase tensor. In general, the experimental results showed high efficiency of normalization of the principal components of the impedance tensor and telluric tensor and lower efficiency of normalization of additional components.