In this letter, two machine learning algorithms are improved, including a sample-compressed neural network algorithm for magnetotelluric (MT) inversion and an adaptive-clustering analysis algorithm for boundary demarcation. MT is widely used in deep geological structure exploration; however, data processing and interpretation still need to be further improved. Inverting the underground electrical structure model from the surface electromagnetic response is a highly nonlinear optimization problem. Common quasi-linear algorithms rely on the initial model and are easy to converge to a local minimum. In addition, demarcating the boundary and attributes of the abnormal bodies according to the inversion results is often manual, inefficient, and haphazard. The validity of the above two machine learning methods is proved by using the simulated data and the actual data. The new algorithms can improve the efficiency and automation of MT data inversion imaging.

In audio magnetotelluric (AMT) sounding data processing, the absence of sferic signals in some time ranges results in a lack of energy in the AMT dead band, causing unreliable resistivity estimations. To address this issue, we develop a deep convolutional neural network (CNN) to automatically recognize sferic signals from redundantly recorded data over a long time range and use these sferic signals to accurately estimate resistivity. The CNN is trained using field time-series data with different signal-to-noise ratios (S/Ns) acquired from different regions of mainland China. To solve the potential overfitting due to the limited number of sferic labels, we develop a training strategy that randomly generates training samples with random data augmentations while optimizing the CNN model parameters. The training process and data generation are stopped when the training loss converges. In addition, we use a weighted binary cross-entropy loss function to solve the sample imbalance problem to optimize the network better, use multiple reasonable metrics to evaluate the network performance, and perform ablation experiments to optimize the model hyperparameters. Extensive field data applications indicate that our trained CNN can robustly recognize sferic signals from noisy time series for subsequent impedance estimation. The results indicate that our method can significantly improve the S/Ns and effectively solve the lack of energy in the dead band. Compared with the traditional processing method, our method can generate smoother and more reasonable apparent resistivity-phase curves and depolarized phase tensors, correct the estimation error of the sudden drop in high-frequency apparent resistivity and abnormal behavior of phase reversal, and better estimate the real shallow resistivity structure.

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.

In the audio magnetotellurics (AMT) sounding data processing, the absence of sferic signals in some time ranges typically results in a lack of energy in the AMT dead band, which may cause unreliable resistivity estimate. We propose a deep convolutional neural network (CNN) to automatically recognize sferic signals from redundantly recorded data in a long time range and use them to compensate for the resistivity estimation. We train the CNN by using field time series data with different signal-to-noise rations that were acquired from different regions in mainland China. To solve the potential overfitting problem due to the limited number of sferic labels, we propose a training strategy that randomly generates training samples (with random data augmentations) while optimizing the CNN model parameters. We stop the training process and data generation until the training loss converges. In addition, we use a weighted binary cross-entropy loss function to solve the sample imbalance problem to better optimize the network, use multiple reasonable metrics to evaluate network performance, and carry out ablation experiments to optimally choose the model hyperparameters. Extensive field data applications show that our trained CNN can robustly recognize sferic signals from noisy time series for subsequent impedance estimation. The subsequent processing results show that our method can significantly improve S/N and effectively solve the problem of lack of energy in the dead band. Compared to the traditional processing method without sferic compensation, our method can generate a smoother and more reasonable apparent resistivity-phase curves and depolarized phase tensor, correct the estimation error of sudden drop of high-frequency apparent resistivity and abnormal behavior of phase reversal, and finally better restore the real shallow subsurface resistivity structure.