Non-invasive geophysical exploration methods a play key role in the exploration of ore deposits. In the present study, the audio-frequency magnetotelluric (AMT) method was applied to metallic mineral exploration. The metallic mineral deposit targeted was the recently discovered super large lead–zinc deposit of the Zhugongtang mining area of Hezhang County in the northwestern Guizhou province in China. The main objectives of this study were to estimate the geoelectric strike and generate geoelectric models that estimate both the depth and distribution of resistivity structures across the deposit. To achieve the objectives, we deployed sixty-one (61) AMT survey sites with an interstation separation of 20 m on a 1280 m survey track perpendicular to the geological strike across the Zhugongtang deposit. We operated in fifty-three (53) frequencies in the range 1–10,400 Hz to record the resistivity distribution of subsurface to a depth of more than 1200 m. The results from the AMT data computations estimated the geoelectric strike that varies between NE285° and NE315°. This range of strikes suggested that structures across the deposit are oriented in the NW–SE direction. Obtained two-dimensional (2D) models elucidated a remarkably low resistivity body (0.50 km depth), thus extending to great depth and were interpreted as lead–zinc mineralization. Furthermore, low resistivity (1000 Ωm). Bioclastic limestone and dolomite limestone were inferred and characterized by moderate-high resistivity (>250 Ωm) and were not widely distributed. A unit of basalts was found with moderate resistivity (>63 Ωm). In addition, it was also found that regions with high number of faults tend to have low resistivity values compared to regions with a low fault number. In summary, this case study presents the results of applying an AMT approach to explore the conductivity characteristics of structures across the Zhugongtang deposit. The findings may contribute to the literature about this deposit.
Induced polarization (IP) is a widely used geophysical exploration technique. Continuous random noise is one of the most prevalent interferences that can seriously contaminate the IP signal and distort the apparent electrical characteristics. We develop a noise separation algorithm based on deep learning to overcome this issue. The standard IP signals are first produced by combining the Cole-Cole model and Fourier series decomposition, and then the mathematical simulation is used to generate various types of random noise interferences, which are subsequently added to the IP signals. Then, a denoising autoencoder deep neural network structure is built and trained by using noisy signals as input samples and pure signals as output samples. The resulting optimum network is capable of automatically reconstructing a clean IP signal from the noisy input. This network is tested using synthetic data sets. The trained neural network can perform the noise reduction of thousands of survey points in a matter of seconds and reduce signal distortion from approximately 25% to less than 5%. Deep learning-based denoising provides superior computation speed and precision compared with the wavelet denoising and smoothing filtering approach. The data for high-quality signals do not vary considerably before and after noise reduction. The noise interferences are successfully suppressed for low-quality signals. Based on the findings, the denoising autoencoder deep neural network has a promising future for suppressing random noise interferences, which can aid in improving the quality of IP data with high efficiency and precision.
Shallow conductive heterogeneity can lead to a static shift in the apparent resistivity sounding curve of controlled-source audio-frequency magnetotellurics (CSAMT). The static effect will shift the apparent resistivity curves along with log-log coordinates axial. Such an effect, if not properly processed, can distort the resistivity of the rock formation and the depth of the interface, and even make the geological structure unrecognizable. In this paper, we have discussed the reasons and characteristics of static shift and summarized the previous work on static shift correction. Then, we proposed the use of the Guided Image Filtering to suppress CSAMT static shift. First, we used the multi-window superposition method to superimpose 1D signals into a 2D matrix format-like image, and then, we conducted Guided Image Filtering on the matrix. In the synthetic model study and field examples, Guided Image Filtering algorithm effectively corrected and suppressed the static shift, and improved the accuracy and precision in data interpretation.