Addressing the impact of near-field effects in the Controlled Source Electromagnetic Method(CSEM) has long been a focal point in the realm of geophysical exploration. Therefore, we propose a deep learning-based near-field correction method for controlled-source electromagnetic methods. Initially, diverse datasets for a layered geologic model are generated through forward simulation. Building upon the characteristics of near-field effects, a deep learning network utilizing LSTM-CNN is meticulously constructed. Multiple experiments are executed to scrutinize the network’s effectiveness in mitigating near-field effects and its resilience against noise. Following this, the proposed method is applied to actual CSEM data to validate its applicability in practice. The method is subsequently tested on measured CSEM data, confirming its practical efficacy. Results from experiments indicate that, for theoretical data, the LSTM-CNN network-trained data closely aligns with simulated data, showcasing a significant improvement. Moreover, when applied to measured data, the method eradicates false high-resistance anomalies at lower frequencies. In conclusion, this deep learning-based correction method proficiently eliminates the influence of near-field effects in the CSEM, delivering practical application benefits that more accurately reflect the authentic geologic structure.

Transient Electromagnetic (TEM) method is widely used in various geophysical applications, providing valuable insights into subsurface properties. However, time-domain TEM signals are often submerged in various types of noise. While recent deep learning-based denoising models have shown strong performance, these models are mostly trained on simulated or single real-world scenario data, overlooking the significant differences in noise characteristics from different geographical regions. Intuitively, models trained in one environment often struggle to perform well in new settings due to differences in geological conditions, equipment, and external interference, leading to reduced denoising performance. To this end, we propose the Dictionary-driven Prior Regularization Test-time Adaptation (DP-TTA). Our key insight is that TEM signals possess intrinsic physical characteristics, such as exponential decay and smoothness, which remain consistent across different regions regardless of external conditions. These intrinsic characteristics serve as ideal prior knowledge for guiding the TTA strategy, which helps the pre-trained model dynamically adjust parameters by utilizing self-supervised losses, improving denoising performance in new scenarios. To implement this, we customized a network, named DTEMDNet. Specifically, we first use dictionary learning to encode these intrinsic characteristics as a dictionary-driven prior, which is integrated into the model during training. At the testing stage, this prior guides the model to adapt dynamically to new environments by minimizing self-supervised losses derived from the dictionary-driven consistency and the signal one-order variation. Extensive experimental results demonstrate that the proposed method achieves much better performance than existing TEM denoising methods and TTA methods.