Simulating electromagnetic (EM) fields can obtain the EM responses of geoelectric models at different times and spaces, which helps to explain the dynamic process of EM wave propagation underground. EM forward modeling is regarded as the engine of inversion. Traditional numerical methods have certain limitations in simulating the EM responses from large-scale geoelectric models. In recent years, the emerging physics-informed neural networks (PINNs) have given new solutions for geophysical EM field simulations. This paper conducts a preliminary exploration using PINN to simulate geophysical frequency domain EM fields. The proposed PINN performs self-supervised training under physical constraints without any data. Once the training is completed, the responses of EM fields at any position in the geoelectric model can be inferred instantly. Compared with the finite-difference solution, the proposed PINN performs the task of geophysical frequency domain EM field simulations well. The proposed PINN is applicable for simulating the EM response of any one-dimensional geoelectric model under any polarization mode at any frequency and any spatial position. This work provides a new scenario for the application of artificial intelligence in geophysical EM exploration.

Natural gas hydrate is one of the most important clean energies and part of carbon cycle, due to the least carbon content. Natural gas hydrates depend on high pressure and low temperatures, located under seabed or permafrost. Small changes in temperature and pressure may lead gas hydrates to separate into water and gas, commonly as methane. As a powerful greenhouse gas, methane is much stronger than carbon dioxide. Therefore, it is necessary to detect the gas hydrates stable zone (GHSZ) before the methane gas escapes from GHSZ. Marine controlled source electromagnetic method (CSEM) is a useful tool to detect gas hydrate in offshore. The results from 3D CSEM method are a resistivity cube to describe the distribution of gas hydrates. In order to study the detectability of CSEM method, we simulate the sensitivity and resolution of marine CSEM synthetic data. By using the sensitivity and resolution, a simple statement may be quickly judged on the existence and occurrence range of the natural gas hydrate. In this paper, we compare the resolution of marine CSEM method with various transverse resistance. This information may help researchers find out whether the GHSZ exists or not.

Acoustic wave features, including velocity dispersion and attenuation, induced by fluid flow in porous media have attracted significant attention in reservoir exploration. To enhance the quantitative understanding of these features, various wave propagation mechanisms have been developed. It has been discovered that wave dispersion and attenuation are associated with multiple reservoir parameters, each with different sensitivity. It is difficult to distinguish the effects of individual physical parameters on acoustic features by traditional wave equations. Considering the ability of deep neural networks (DNNs) in establishing the relationships between two datasets, a fully connected DNN has been employed as a surrogate rock physics model, and the Shapley additive explanations model (SHAP) based on this DNN has been introduced to evaluate the contributions of different parameters. In this study, the classic White model is used to generate datasets for training the DNN. Datasets include seven parameters (bulk modulus, shear modulus, and density of the solid matrix, frequency, porosity, fluid saturation, and permeability), along with velocity dispersion and attenuation. By embedding SHAP into the trained DNN, the presented ShaRock algorithm allows for a clear quantification of the contributions of various reservoir parameters to acoustic features. Furthermore, we analyze the underlying interactions between two parameters by using their combined quantified contributions to the features. The application of this proposed algorithm, which is based on wave propagation mechanisms, demonstrates its potential in providing valuable insights for parameter inversions in hydrocarbon exploration.

Characterizing seismic wave propagation in a fluid-saturated porous media well enhances the precision of interpreting seismic data, bringing benefits to understanding reservoir properties better. Some important indicators, including wave dispersion and attenuation, along with the wavefield, are widely used for interpreting the reservoir, and they can be obtained from a rock physics model. In existing models, some of them are limited in scope due to their complexity, for example, numerical solutions are difficult or costly. In view of this, this study proposes an approach of establishing equivalent dynamic equations of existing models. First, the framework of the equivalent model is derived based on Biot's theory, while the elastic coefficients are set as unknown factors. The next step is to use deep neural networks (DNNs) to predict these coefficients, and surrogate models of unknowns are established after training DNNs. The training data is naturally generated from the original model. The simplicity of the equation forms, compared to the original complex model and some other equivalents such as the viscoelastic model, enables the framework to perform wavefield simulation easier. Numerical examples show that the established equivalent model can not only predict similar dispersion and attenuation, but also obtain wavefields with small differences. This also indicates that it may be sufficient to establish an equivalent model only according to dispersion and attenuation, and the cost of generating such data is very small compared to simulating the wavefield. Therefore, the proposed approach is expected to effectively improve the computational difficulty of some existing models.