Izvestiya, Physics of the Solid Earth

The influence of near-surface inhomogeneities is a major challenge in the magnetotelluric sounding (MTS) method, as it hampers the retrieval of information on deep structures. Local, thin near-surface inhomogeneities cause static shifts of MT amplitude curves while leaving the phase curves unchanged. A more complicated case arises from significant variations in the total longitudinal conductance S of the upper layer, which lead to changes in the shape of the amplitude curves. Variations in S also affect the phases of the impedance tensor components [Z] and the telluric tensor [T], as well as the Wiese-Parkinson matrix [W] and the magnetic tensor [M]. E.B. Fainberg was the first to propose a dynamic correction of MT amplitude curves, which reduces to multiplying them by a frequency-dependent factor. M.N. Berdichevsky generalized this idea to the impedance tensor and the electric distortion matrix. V.A. Kuznetsov proposed an algorithm that also accounts for magnetic distortions. These approaches are based on three-dimensional modeling of the MT field and require specifying the distribution of S in the upper layer and a background 1-D layered model. We implemented dynamic-correction algorithms and assessed their performance using synthetic (model) MTS data. The data were computed for a three-dimensional model of the tectonosphere containing large-scale inhomogeneities at three structural levels: the sedimentary cover, the consolidated crust, and the upper mantle. Numerical experiments showed that dynamic correction effectively suppresses the influence of sedimentary-cover structures. In the different MTS data components, only the effects of deep crustal and mantle structures remained, enabling their confident localization and potentially simplifying subsequent inversion. Clearly, the success of dynamic correction depends on the accuracy with which the conductivity of the upper layer and the background 1-D model are specified, as well as on observation errors. Future work will evaluate the effectiveness of dynamic correction under realistic conditions.

The article presents a systematic review of modern methods for solving inverse problems in geoelectrics. Three main paradigms are considered: deterministic - based on Tikhonov regularization, probabilistic - based on the Bayesian approach, and approximation - based on the neural network (NN) method. Particular attention is paid to the fundamental aspects of the inverse problem related to solution non-uniqueness and its dependence on the depth of investigation and parameterization detail. The evolution of NN methods is analyzed, from classical ones using the simplest perceptron to multi-layer physics-informed neural networks (PINNs) employing deep machine learning (DML) methods. Numerical results are presented.

As part of the deep-water monitoring of the vertical electric field component E_z, long-term observations of magnetic field variations on land in the southern basin of Lake Baikal were also conducted. These observations were used to calculate gradient functions between the reference point (Patrona Observatory) and existing magnetic measurement points over periods ranging from 5.6 minutes to 24 hours. These data were used to compare with model calculations when considering various models of the geoelectric structure of the Baikal Rift Zone (BRZ). The characteristic features of the mantle diapir structure are most pronounced in the southernmost part of the BRZ, with a boundary along the Main Sayan Fault. In general, the diapir structure is sharply asymmetric, with the supply channel located under the northwestern periphery of the mantle body. The position of the southeastern boundary of the body requires refinement by expanding the monitoring network.

Construction of the crystalline basement surface model for the Lena-Tunguska petroleum province was performed based on the results of integrated interpretation of electrical survey data from near-field transient electromagnetic (TEM) soundings and magnetotelluric (MT) soundings, magnetic survey, and gravity survey. Each method is characterized by its own complicating physical-geological factors and ranges of ambiguity in the solutions of inverse problems. It is shown that the integrated interpretation of gravitational and magnetic fields allows reducing the ambiguity caused by the lack of a priori information, the variety of structural constructions from previous years, and complex geoelectrical conditions at the regional stage. To delineate 3D heterogeneities based on density and magnetic properties, cascade inversion technology was used. Formation of a generalized model of volumetric distribution of density and magnetic properties was carried out based on structural-material complexes, identified as a result of classification by the classical machine learning method — K-means. The integrated approach increased the reliability of geological interpretation in zones of data complications for TEM and MT. The resulting models of the medium provided a refined understanding of the presumed basement surface structure.

Transient responses recorded during electromagnetic soundings using the TEM (Transient Electromagnetic) method are generated and formed in the studied geological environment by several physical processes. The induction process (IN) is generated by eddy currents dependent on the electrical conductivity distribution in the probed volume. The superparamagnetic (SPM) effect arises from the remagnetization of single-domain magnetite grains in rock. The nature of polarization processes (PP) is related to electric current flow through rock exhibiting frequency dispersion of electrical conductivity. Inductively induced polarization (IIP) is generated by eddy currents in a dispersed electrically conductive medium. The induced polarization (IP) process can be generated by displacement currents in antenna loops with distributed capacitance and resistance. IN and IIP are manifested in the early stages of transient responses, while IN, IP, and SPM are manifested in the later stages. The relaxation rates of these processes differ: IN and IIP are "fast", whereas IP and SPM are "slow". IN and SPM decay in-phase over time, while IP and IIP decay in antiphase to IN. In TEM, information about the geoelectric structure of the medium is derived from IN, whereas the IIP, IP, and SPM processes are typically considered interference. However, identifying the presence or absence of polarization and magnetic effects in recorded responses is only possible in cases where they dominate, significantly distorting the shape of the IN induction responses. If IIP, IP, and SPM are small, their presence in the signal becomes apparent only during the inversion and interpretation of TEM data. Monitoring of transient responses has established that the IIP, IP, and SPM processes are formed in the soil layer and vary over time. Variations in IIP and IP are modulated by a diurnal cycle, with maxima during midday hours and minima at night. It has been found that time-multiplied SPM responses in the TEM magnetic sensor can be represented as a convolution of the particle volume distribution function with a narrowband filter. SPM effects depend on soil moisture and the orientation of the exciting field relative to the Earth's magnetic field. Due to internal biological processes in the soil colloid-gel system, the contained SPM particles coagulate, forming clusters. This effect alters the particle size distribution and, consequently, the decay pattern of SPM responses. The processes of cluster formation and decay are in continuous alternation, generating anomalies in the recorded TEM signals. IP and SPM processes generally accompany each other; however, the balance of the antiphase SPM-IP effects is highly unstable over time, and the polarity-changing variations of TEM signals are several times greater than the induction responses of the medium (IN). The paper presents results of IIP, IP, and SPM effect measurements obtained during a three-year monitoring period and investigates previously unknown SPM properties of soils.

A method for the statistical detection of weak ultra-low frequency (ULF) electromagnetic anomalies of lithospheric origin under conditions of intense anthropogenic noise is proposed. The method is based on the hypothesis of the temporal stability in the statistical distribution of background electromagnetic noise amplitudes in the ULF range, where the appearance of an additional signal leads to a change in this distribution. A numerical experiment was conducted by superimposing an artificial noise signal onto magnetic data obtained at the Tashkent Geodynamic Polygon over a 72-day observation period from December 2022 to February 2023. The results of the numerical experiment show that a stable shift in the mode of the probability distribution of the root-mean-square deviations of magnetic disturbance amplitudes toward larger values can serve as an indicator of the emergence of an electromagnetic disturbance from a remote source. The method has been tested and its applicability is demonstrated for the M6.4 earthquake.

The results of experimental studies on monitoring variations in apparent resistivity, as a precursor to earthquakes, by audiomagnetotelluric soundings at the Goryachinsk test site (Baikal region) are considered. A comparison of different types of groundings of receiving electrical lines (brass, lead, chlorine-lead electrodes) has been performed. Chlorine-lead electrodes are characterized by the most stable data. When selecting stress-strain sensitive zones for the installation of monitoring equipment, the geological features of the territory were analyzed, field experimental studies were carried out to study pre-selected sites at fault zones, and test-monitoring sessions were conducted to assess the response to tidal deformations of the earth. The test monitoring data were used to select the frequency range for monitoring variations in apparent resistivity. The obtained apparent resistivity curves are characterized by stable behavior and small spread of values in the frequency range of 7-300 Hz. Therefore, it is advisable to perform standard magnetotelluric processing of monitoring data to obtain apparent resistivity and impedance phase curves in the specified frequency range, and analyze variations in the magnetic and electrotelluric fields at frequencies below 7 Hz. According to the results of the test monitoring, it was found that tidal effects are observed in the values of apparent resistivity for the E-polarized field. For the H-polarized field, there are no noticeable fluctuations. The values of variations relative to the average level of 35-36 Ω·m are small, 0.3-0.4 Ω·m (about 1%). The results of the apparent resistivity monitoring showed high measurement accuracy, about 0.3%. The selected stress-strain sensitive zone can be used for long-term monitoring.

In the southwestern part of Lake Baikal, a long-term experiment has been conducted on continuous high-precision measurements of the vertical component of the electric field E_z and the signals of macroscopic nonlocal correlation detectors with surface-to-bottom spacing. The sources of the field are located both in the hydrosphere itself and outside it, which makes it possible to monitor the source processes in the conjugated geospheres. For periods up to 10 days, the field induced by flows dominates. This made it possible to obtain information about variations in the total transport of synoptic flow. For longer periods, the current field of the global electrical circuit dominates. These variations turned out to be related to variations in the solar X-ray radiation. Signals from detectors of macroscopic nonlocal correlations reveal an advanced response to some regional and global dissipative processes with a large random component. This provides a unique opportunity to forecast such processes, in particular, hydrothermodynamic activity in the active layer of Lake Baikal.

Space weather is the set of processes occurring in near-Earth space under the influence of solar activity. This review presents the main components of space weather, the causes of their formation, and their impacts on the Earth. One of the most significant space weather factors is geomagnetically induced currents in technological conducting grounded systems, caused by rapid variations of the geomagnetic field. Extended conducting systems such as power transmission lines, pipelines, and railways are particularly vulnerable. This paper presents the main results of analysis of the effects of geomagnetic disturbances on the operation of power systems, main trunk pipelines, and railway automation. The results confirm the necessity of developing adaptive protection measures. For an adequate assessment of the impact of geomagnetic disturbances, it is necessary to further develop short-term and long-term space weather forecasting systems, local models of the Earth's crustal conductivity, and their integration with real network configurations.