Seismic image denoising is essential to enhance the signal-to-noise ratio (SNR) of seismic images and facilitate seismic processing and geological structure interpretation. With the development of deep learning (DL), several DL-based models have been proposed for seismic image denoising. However, the commonly used supervised DL-based denoising models require noise-free data as training labels, yet noise-free data are often difficult to be obtained in field application scenarios. By considering the similarity of seismic images, we propose a similarity-informed self-learning (SISL) to address seismic image denoising in the absence of noise-free seismic images. To accurately preserve valid seismic signals when constructing training pairs, we develop a specialized workflow, termed the similar image sampler. In this way, we can fully use the self-similarity of noisy seismic images to build training pairs and then train a denoising model. Moreover, to effectively attenuate random noise, we propose a hybrid loss function with a regularization constraint to availably retain valid seismic events. After comparing with the traditional denoising methods and several state-of-the-art unsupervised DL models, the experimental results from synthetic and field data quantitatively and qualitatively demonstrate the effectiveness and the stability of the proposed SISL model for seismic image denoising.

Lithofacies identification and classification are critical for characterizing the hydrocarbon potential of unconventional resources. Although extensive applications of machine learning models in predicting lithofacies have been applied to conventional reservoir systems, the effectiveness of machine learning models in predicting clay-rich, lacustrine shale lithofacies has yet to be tackled. Here, we apply machine learning models to conventional well log data to automatically identify the shale lithofacies of Gulong Shale in the Songliao Basin. The shale lithofacies were classified into six types based on total organic carbon and mineral composition data from core analysis and geochemical logs. We compared the accuracy of Multilayer Perceptron (MLP), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), and Random Forest models. We mitigated the bias of imbalanced data by applying oversampling algorithms. Our results show that ensemble methods (XGBoost and Random Forest) have a better performance in shale lithofacies identification than the other models do, with accuracies of 0.868 and 0.884, respectively. The organic siliceous shale proposed to have the best hydrocarbon potential in Gulong Shale can be identified with F1 scores of 0.853 by XGBoost and 0.877 by Random Forest. Our study suggests that ensemble machine learning models can effectively identify the lithofacies of clay-rich shale from conventional well logs, providing insight into the sweet spot prediction of unconventional reservoirs. Further improvements in model performances can be achieved by adding domain knowledge and employing advanced well log data.

Submarine landslides have attracted widespread attention, with the continuous development of ocean engineering. Due to the recent developments of in-situ investigation and modelling techniques of submarine landslides, significant improvements were achieved in the evolution studies on submarine landslides. The general characteristics of typical submarine landslides in the world are analyzed. Based on this, three stages of submarine landslide disaster evolution are proposed, namely, the submarine slope instability evolution stage, the large deformation landslide movement stage, and the stage of submarine landslide deposition. Given these three stages, the evolution process of submarine landslide disaster is revealed from the perspectives of in-situ investigation techniques, physical simulation, and numerical simulation methods, respectively. For long-term investigation of submarine landslides, an in-situ monitoring system with long-term service and multi-parameter collaborative observation deserves to be developed. The mechanism of submarine landslide evolution and the early warning factors need to be further studied by physical modelling experiments. The whole process of the numerical simulation of submarine landslides, from seabed instability to large deformation sliding to the impact on marine structures, and economizing the computational costs of models by advanced techniques such as parallel processing and GPU-accelerators, are the key development directions in numerical simulation. The current research deficiencies and future development directions in the subject of submarine landslides are proposed to provide a useful reference for the prediction and early warning of submarine landslide disasters.

Attenuation of incoherent noise is an effective way to improve signal-to-noise ratio (SNR) of seismic data. Recently, supervised deep learning based methods have been widely utilized for seismic image denoising, which often need plenty of noise-free data as training labels. However, noise-free seismic data are often unavailable in field applications. We propose an unsupervised learning method (NS2NS) to train a denoising network by using single noisy seismic data. The proposed model is based on two basic truths of seismic data: (1) High self-similarity of seismic data; (2) Spatially independence of incoherent noise in seismic data. To implement the proposed method, we first build a sampling workflow to generate paired noisy images based on single noisy seismic image. Moreover, we create similar noisy images that are similar but different with the original noisy image by using the proposed self-similar sampler. The original noisy images and generated similar noisy images are then fused by using a suggested Bernoulli sampler to create new paired noisy images. These new paired noisy images are used as the input and target of the denoising model, respectively. Next, an end-to-end convolutional neural network (CNN) is built for seismic image denoising, which aims to learn features of valid signals and suppress unpredictable random noise. Finally, we apply the proposed NS2NS method to both synthetic and field data. The results show that our proposed method can effectively suppress incoherent noise while preserving valid signals.

Faults in the subsurface can be an avenue of, or a barrier to, hydrocarbon flow and pressure communication. Manual interpretation of discontinuities on 3D seismic amplitude volume is the most common way to define faults within a reservoir. Unfortunately, 3D seismic fault interpretation can be a time-consuming and tedious task. Seismic attributes such as coherence help define faults, but suffer from “staircase” artifacts and nonfault-related stratigraphic discontinuities. We assume that each sample of the seismic data is located at a potential fault plane. The hypothesized fault divides the seismic data centered at the analysis sample into two subwindows. We then compute the coherence for the two subwindows and full analysis window. We repeat the process by rotating the hypothesized fault plane along a set of user-defined discrete fault dip and azimuth. We obtain almost the same coherence values for the subwindows and the full window if the analysis point is not located at a fault plane. The “best” fault plane results in maximum coherence for the subwindows and minimum coherence for the full window if the analysis point is located at a fault plane. To improve the continuity of the fault attributes, we finally smooth the fault probability attribute along the estimated fault plane. We illustrate the effectiveness of our workflow by applying it to a synthetic and two real seismic data. The results indicate that our workflow successfully produces a continuous fault attribute without staircase artifacts and stratigraphic discontinuities.

Today's 3-D seismic surveys usually contain hundreds of inline and crossline vertical seismic slices. Seismic interpreters usually need to spend weeks or even months for manually picking horizons on vertical seismic slices. Researchers have developed algorithms to accelerate horizon picking and most algorithms employ the seismic reflector's dip as the input. However, the computed seismic reflector's dip is usually inaccurate near and across the discontinuities in the seismic images. Note that the time samples which belong to the same horizon should have approximate similar seismic instantaneous phase values. We propose to automatically track the seismic horizon simultaneously considering the seismic reflector's dip and instantaneous phase attributes. Our algorithm aims to achieve three objectives: 1) minimizing the difference between the dip computed using tracked horizon and seismic dip attribute, 2) minimizing the difference among instantaneous phase value of the time samples on the tracked horizon, and 3) the tracked horizon exactly passes through user-defined control points (seeds). A constrained conjugate least-square algorithm is employed to solve our optimization problem. The applications show that the tracked horizon which only uses dip attribute would ``jump'' from one seismic event to another seismic event near the unconformity zone. However, the horizon tracked using the proposed method strictly follows the same seismic event over the whole seismic survey.