Magnetotelluric (MT) sounding data can easily be damaged by various types of noise, especially in industrial areas, where the quality of measured data is poor. Most traditional de-noising methods are ineffective to the low signal-to-noise ratio of data. To solve the above problem, we propose the use of a de-noising method for the detection of noise in MT data based on discrete wavelet transform and singular value decomposition (SVD), with multiscale dispersion entropy and phase space reconstruction carried out for pretreatment. No “over processing” takes place in the proposed method. Compared with wavelet transform and SVD decomposition in synthetic tests, the proposed method removes the profile of noise more completely, including large-scale noise and impulse noise. For high levels or low levels of noise, the proposed method can increase the signal-to-noise ratio of data more obviously. Moreover, application to the field MT data can prove the performance of the proposed method. The proposed method is a feasible method for the elimination of various noise types and can improve MT data with high noise levels, obtaining a recovery in the response. It can improve abrupt points and distortion in MT response curves more effectively than the robust method can.

Magnetotelluric (MT) sounding data are easily contaminated by various noise sources, especially noise with a long duration (even full-time noise), which makes it difficult to obtain accurate values when calculating the weight factors of response function, resulting in distortion of the response results. Based on blind source separation theory, fast independent component analysis (FastICA) can separate this kind of noise. However, this method is challenged by the number of separated field sources and the unequal signal amplitude before and after processing. We have developed a novel signal noise separation method, improving the traditional FastICA, where a correlation coefficient is used to compute the number of field sources, and the fast iterative shrinkage threshold algorithm (FISTA) is used to adjust the signal amplitude problem before and after FastICA decomposition. Compared with common field source division methods and the traditional FastICA, the experimental results indicate that our method can separate and remove noise with a long duration, increase the signal-to-noise ratio (SNR) of the data, and improve the MT response curves. Meanwhile, case studies of measured data illustrate that our method obtains a more robust MT response than the conventional robust method and FastICA.

The cross-gradient joint inversion of gravity and magnetic data is a commonly used way, which requires dividing the subsurface into closely arranged cells with structured or unstructured grids to solve the physical properties of the discrete subsurface. Unstructured grids can more effectively fit undulating terrain and irregular geological bodies than structured grids. However, the existing cross-gradient joint inversion method with structured grids cannot be used to complete the inversion with unstructured grids. We proved that the second-order Taylor formula is more suitable for calculating the cross-gradients of unstructured grids; therefore, we first proposed and implemented the second-order Taylor formula cross-gradient (2-TCG) method for the cross-gradient joint inversion of gravity and magnetic data with unstructured grids, which can calculate the cross-gradient item more easily than other ways because the first few coefficients of Taylor formula are gradient items. The model tests showed that the 2-TCG joint inversion method can accurately achieve the 3-D joint inversion of gravity and magnetic data with unstructured grids, and this method can better describe the detailed information of irregular geological bodies in complex terrain. To explore the minerals and analyze the metallogenic model in the Huanggang–Ganzhuermiao metallogenic belt of the southern Greater Khingan Range, Inner Mongolia, we measured gravity and magnetic data on a scale of 1:10000. We estimated the mineral distributions with 150–280 m depth by 2-TCG joint inversion. We analyzed the relationship between ore bodies and faults, and we established a mixed metallogenic model of the skarns and hydrothermal veins in the survey area.

Magnetotelluric (MT) impedance estimation requires a high signal-to-noise ratio (SNR). When low-SNR data are processed, it is difficult to obtain a robust MT response. In this paper, based on the spectral characteristics of noise sequences, the influence of the scale and waveform of noise sequences on impedance estimates is studied, and a multiscale denoising method for MT signals is proposed. This method applies the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) to decompose the multiscale noise into different components, and then the influence of noise on the spectrum is evaluated through the spectrum obtained by the short-time Fourier transform (STFT) of each component. This ICEEMDAN- and STFT-based MT denoising method (ICMT) can thus filter out the small-scale abrupt noise that has a great impact on the MT response and retain the large-scale smooth noise that has a small impact to suppress noise and reduce the loss of the effective signal at the same time. Various noise is added to a pure MT signal to test the performance of ICMT. It is demonstrated that ICMT has a small loss of effective signals and can obtain MT response results with small recovery errors, even when the SNR of the signal is as low as -20 dB. Finally, ICMT is applied to process the heavy noisy MT data in an ore concentration area. The results suggest that ICMT can effectively suppress noise, and robust impedance estimates can be obtained.

When magnetotelluric (MT) sounding data are measured in mining areas and urban areas, the useful signals are buried under the surrounding interference sources in the whole period, which completely covers up the useful signals, resulting in jump points and distortion of the response curves. Sparse representation uses atoms in a dictionary to process noisy signals. Based on the arbitrariness of the length of these dictionary atoms, they can effectively suppress noise even if the noise fills the entire observation period. We propose an improved sparse representation based on an adaptive dictionary, which can construct a dictionary according to the characteristics of the data itself (extracting the noise and the useful signals separately) and then automatically filter the noise atoms (which are regard as noise) in the dictionary to suppress long-duration noise (noise lasts for a long time). For the synthetic data and the measured data, in a comparison with the common signal noise separation methods, the results indicate that the proposed method can more completely extract the profiles of the long-duration noise and greatly increase the signal-to-noise ratio. Moreover, for various noise sources, the proposed method indicates better improvement performance, obtaining smoother and more reliable response results.