Seismic trace interpolation is an important technique because irregular or insufficient sampling data along the spatial direction may lead to inevitable errors in multiple suppression, imaging, and inversion. Many interpolation methods have been studied for irregularly sampled data. Inspired by the working idea of the autoencoder and convolutional neural network, we have performed seismic trace interpolation by using the convolutional autoencoder (CAE). The irregularly sampled data are taken as corrupted data. By using a training data set including pairs of the corrupted and complete data, CAE can automatically learn to extract features from the corrupted data and reconstruct the complete data from the extracted features. It can avoid some assumptions in the traditional trace interpolation method such as the linearity of events, low-rankness, or sparsity. In addition, once the CAE network training is completed, the corrupted seismic data can be interpolated immediately with very low computational cost. A CAE network composed of three convolutional layers and three deconvolutional layers is designed to explore the capabilities of CAE-based seismic trace interpolation for an irregularly sampled data set. To solve the problem of rare complete shot gathers in field data applications, the trained network on synthetic data is used as an initialization of the network training on field data, called the transfer learning strategy. Experiments on synthetic and field data sets indicate the validity and flexibility of the trained CAE. Compared with the curvelet-transform-based method, CAE can lead to comparable or better interpolation performances efficiently. The transfer learning strategy enhances the training efficiency on field data and improves the interpolation performance of CAE with limited training data.

Modeling the elastic properties of organic shale has been of long-standing interest for source rocks and unconventional reservoir characterization. Organic shales exhibit significant variabilities in rock texture and reservoir properties at different maturity stages, subsequently affecting their elastic responses. We have developed a new rock-physics modeling scheme honoring the maturity levels (immature, mature, and overmature), which are constrained by the evolution of the physical properties of organic shale upon kerogen maturation. In particular, at different maturity stages, the manners in which the compliant organic materials interact with the inorganic mineral matrix are characterized by different effective medium theories. On the basis of the developed rock-physics templates, organic shales have different elastic behaviors at different maturity stages. Ignoring the impact of kerogen maturation is insufficient to adequately characterize the elasticity of the whole organic shale system. Modeling results suggest that the elastic responses of organic shale are sensitive to two dominant factors — organic matter content and mineralogical composition. The elastic anisotropy characteristics are not only affected by the kerogen content and clay alignment but also depend on the morphology of kerogen distribution. Our results compare satisfactorily with data from ultrasonic velocity and log measurements, confirming validity and applicability of our modeling framework.

Inferring porosity of subsurface from seismic data is of profound significance to many fields of Earth science and engineering applications, including but not limited to: hydrocarbon reservoir characterization, underground water flow modeling, geological CO₂ storage, and geothermal energy exploitation. Traditional model‐driven approaches confront the problems of strong nonlinearity and geological heterogeneity, while machine learning is good at nonlinear mapping, providing higher efficiency and accuracy as well. We propose a Random Forest (RF) based method using multiple seismic attributes to predict the underground porosity distribution with uncertainty quantification. The standard deviation of base models' predictions is used to quantify the regression uncertainty of RF. The uncertainty can robustly indicate the prediction quality in numerous experiments, where low uncertainty corresponds to relatively precise prediction and high uncertainty gives a possibility of larger errors. Furthermore, we utilize the quantified uncertainty to improve the RF regression accuracy by correcting the originally predicted porosity according to the statistical relationship between the absolute error and the standard deviation. The application of the proposed method on seismic data shows its potential to characterize spatially varying reservoir parameters, and the quantified uncertainty profile offers insights into risk evaluation for hydrocarbon exploration and development.

Seismic prediction of fluid and lithofacies distribution is of great interest to reservoir characterization, geologic model building, and flow unit delineation. Inferring fluids and lithofacies from seismic data under the framework of machine learning is commonly subject to issues of limited features, imbalanced data sets, and spatial constraints. As a consequence, an extreme gradient boosting-based workflow, which takes feature engineering, data balancing, and spatial constraints into account, is proposed to predict the fluid and lithofacies distribution by integrating well-log and seismic data. The constructed feature set based on simple mathematical operations and domain knowledge outperforms the benchmark group consisting of conventional elastic attributes of P-impedance and VP/VS ratio. A radial basis function characterizing the weights of training samples according to the distances from the available wells to the target region is developed to impose spatial constraints on the model training process, significantly improving the prediction accuracy and reliability of gas sandstone. The strategy combining the synthetic minority oversampling technique and spatial constraints further increases the F1 score of gas sandstone and also benefits the overall prediction performance of all of the facies. The application of the combined strategy on prestack seismic inversion results generates a more geologically reasonable spatial distribution of fluids, thus verifying the robustness and effectiveness of our workflow.

We mapped probabilities of lithology and fluid based on seismic and well observations in a heterogeneous carbonate reservoir of the Sichuan Basin, southwest China, thus characterizing the reservoir complexity and identifying the potential sweet spot. Rock physical analysis showed that the presence of a vuggy-fracture porosity system combined with the fluids’ effect complicate the elastic responses of heterogeneous carbonates. This also gives physical insight into how the elastic properties of different lithofluid classes can be distinguished. With the application of Bayesian linearized amplitude variation with offset inversion, we found that the posterior distribution of P- and S-wave velocities were reliably extracted, whereas the inverted density tended to show high uncertainty due to a lack of wide angle data in the migrated seismic gathers. Finally, a Bayesian approach was implemented to propagate uncertainty from prestack seismic data to lithofluid classes in an integrated framework. The gas-carbonate predictions with high posterior probabilities were consistent with well observations at the well location and the present geological continuity. The probability distributions for the anhydrite, limestone, and dolostone were essential to understand the reservoir architecture and delineate the reservoir heterogeneities. We also ascertained that the uncertainty of lithofacies prediction is determined by the uncertainty of seismic inversion as well as the uncertainty of the link between lithofacies to their corresponding elastic responses.

Conventionally, multicomponent geophones used to record the elastic wavefields in the solid seabed are necessary for ocean bottom seismic (OBS) data elastic reverse time migration (RTM). Particle velocity components are usually injected directly as boundary conditions in the elastic-wave equation in the receiver-side wavefield extrapolation step, which causes artifacts in the resulting elastic images. We have deduced a first-order acoustic-elastic coupled equation (AECE) by substituting pressure fields into the elastic velocity-stress equation (EVSE). AECE has three advantages for OBS data over EVSE when performing elastic RTM. First, the new equation unifies wave propagation in acoustic and elastic media. Second, the new equation separates P-waves directly during wavefield propagation. Third, three approaches are identified when using the receiver-side multicomponent particle velocity records and pressure records in elastic RTM processing: (1) particle velocity components are set as boundary conditions in receiver-side vectorial extrapolation with the AECE, which is equal to the elastic RTM using the conventional EVSE; (2) the pressure component may also be used for receiver-side scalar extrapolation with the AECE, and with which we can accomplish PP and PS images using only the pressure records and suppress most of the artifacts in the PP image with vectorial extrapolation; and (3) ocean-bottom 4C data can be simultaneously used for elastic images with receiver-side tensorial extrapolation using the AECE. Thus, the AECE may be used for conventional elastic RTM, but it also offers the flexibility to obtain PP and PS images using only pressure records.