Stratigraphic Texture-Enhanced Seismic Facies Analysis via Unsupervised Dictionary Learning

Summary Deep learning have led to a certain extent breakthrough in seismic facies analysis(SFA). The greatest challenge in the SFA domain is how to cope with large seismic datasets and the limited amount of annotated samples, especially when employing supervised machine learning algorithms that require labelled data and larger training sample. we develop a supervised seismic facies analysis via an improved Auxiliary Classifier Generative Adversarial Network(ACGAN).To solve the problem of imbalanced seismic facies class distribution, and limited training data with labels, we used an improved ACGAN to generate synthetic seismic data sets, and then train Convolutional Neural Networks(CNN) using theses generated synthetic data. Fault data and seismic facies data test demonstrated that generated seismic data via the improved ACGAN can be used for synthetic data augmentation as for increased the performance of CNN and avoided overfitting compared to original seismic data training set. Therefore, we believe that the proposed method can effectively alleviate the problems of insufficient and imbalanced seismic data in SFA, and also can generalize to other seismic data classification applications.

Seismic reflection pattern analysis or seismic facies analysis is crucial for subsurface reservoir prediction. Supervised pre-stack reflection pattern analysis utilizing well-logging data can fully utilize abundant reservoir information in pre-stack data, and provide clearer physical interpretations than unsupervised methods. However, pre-stack seismic data is high-dimensional, and the well-logging data labels are limited. Traditional convolutional neural network-based approaches face challenges in capturing long-range dependencies across different angle-gathers in pre-stack seismic data due to the limitations of their receptive fields. Additionally, existing data augmentation methods lack constraints and physical guidance. To tackle these problems, we introduce a supervised pre-stack seismic reflection pattern analysis method based on the ConvNext network and incorporating physics-guided and active learning for label augmentation. The ConvNext model incorporates the large kernel attention (LKA) mechanism, enhancing the model's sensitivity to stratigraphic and spatial features in pre-stack seismic data. To reduce model ambiguity, we develop a label augmentation algorithm that combines active learning and physical attributes. The experiments on synthetic data and real data demonstrate that our method has better performance than the traditional approaches in pre-stack seismic reflection analysis.

Machine learning-based automatic seismic facies analysis has increased significantly over the past few decades. The key is to select the most representative features (such as the commonly used poststack amplitude data or the induced seismic attributes) as the input of the different machine learning algorithms. As an advanced branch of machine learning, deep learning can be used to extract the discriminatively deep features from seismic data similar to those used in image classification. In this study, we havedeveloped an unsupervised seismic facies analysis method by using a recurrent autoencoder model. First, we have constructed and trained an autoencoder architecture combined with long short-term memory-based recurrent operation. Its main aim is to learn the deep discriminative features by taking the windowed poststack seismic data as the input time series data. This type of unsupervised learning takes advantage of the no labeling requirement of seismic data. In addition, the recurrent operation is beneficial in delineating the time-sequential characteristics of seismic data. Second, we have taken the learned features as the input of simple K-means clustering and analyzed the corresponding seismic facies. In other words, the clustering is executed in the learned feature space (learned feature-based clustering). Real data results have demonstrated that our method reveals more details than the original amplitude-based K-means clustering, depending on the cluster calibration. In particular, according to the known natural gamma-ray logs and lithological descriptions of five wells, different amounts of sandstone and mudstone deposit are more accurately discriminated, which is substantially informative in reservoir prediction and hydrocarbon exploration. Furthermore, the estimated average silhouette scores have quantitatively shown the effectiveness of our method.

Seismic facies analysis makes use of different seismic parameters in order to get other than structural information. A review is given of possibilities and usefulness of seismic facies analysis in oil exploration.A seismic facies unit can be defined as a sedimentary unit which is different from adjacent units in its seismic characteristics. Parameters that should be taken into consideration in the seismic facies analysis are as follows: reflection amplitude, dominant reflection frequency, reflection polarity, interval velocity, reflection continuity, reflection configuration, abundance of reflections, geometry of seismic facies unit, and relationship with other units.Interpretation of seismic facies data may be either direct or indirect. The purpose of the direct interpretation is to find out geological causes responsible for the seismic signature of a seismic facies unit. So, the direct interpretation may be aimed at predicting lithology, fluid content, porosity, relative age, overpressured shales, type of stratification, geometry of the geological body corresponding to the seismic facies unit and its geological setting. The indirect interpretation is intended to reach some conclusions on depositional processes and environments, sediment transport direction, and some aspects of geological evolution (transgression, regression, subsidence, uplift, erosion).The results of the seismic facies analysis may be shown on seismic facies cross‐sections and seismic facies maps. Depending on the available seismic data and geological conditions in the area under consideration, the seismic facies maps may be of different types such as general seismic facies maps showing distribution of different seismic facies units, sand‐shale ratio maps, direction of cross‐bedding and paleo‐transport maps etc.Several kinds of seismic facies units and their geological interpretation are discussed as examples of seismic facies analysis.

Seismic facies analysis plays a key role in stratigraphy and has developed rapidly in the context of deep learning methods. However, seismic facies analysis cannot only be regarded as an image segmentation or classification task because this would exclude prior knowledge of geophysicists on seismic interpretation. Seismic parameters summarized by geophysicists were included in the seismic facies analysis model. Seismic parameters were used to calibrate the different seismic facies units, and well logging data and sequence stratigraphy were integrated to perform geological interpretation and to establish a training dataset. A multi-head seismic parameter classification (MSPC) model was then established to divide the seismic facies units. A seismic facies geological interpretation (SFGI) model was constructed for the geological interpretation of seismic facies units, and an end-to-end seismic facies analysis model was designed based on the MSPC model and the SFGI model. The proposed model effectively improves the accuracy of the seismic interpretation, solves the intra-facies noise problem, and is easy to train and use.

Seismic facies analysis is a powerful, yet largely subjective, methodology for hydrocarbon exploration and development. The term “seismic facies” means different things to different people. For this study, however, seismic facies is defined as a seismic description of depositional facies confined in a seismically thin bed at a normal seismic sampling rate (e.g., 4 ms), making this study's seismic facies analysis different from classical seismic facies analysis of much thicker depositional sequences. As a subjective interpretation of seismic attributes, seismic facies analysis is often laborious and inconsistent, especially if multiple mapping units are involved. How to automate the process via computer pattern recognition so that consistency could be improved and the 3D cycle time reduced has been a hot topic in recent years.

Seismic facies analysis is increasingly being used in the interpretation of high-resolution seismic reflection data. Existing depositional systems models, based primarily on seismic data, have been constructed on the basis that acoustic character can be directly correlated with relatively consistent lithologies and geotechnical properties. However, borehole and shallow core data from the northern UK continental margin indicate significant inconsistencies in the use of this technique as a predictive tool. Changes in acoustic texture can occur laterally and vertically, in both lithologically homogeneous and heterogeneous sequences. The successful application of this technique can only be achieved by the integration of seismic data with other subsurface information, with interpretation based on sound geological concepts and models.

Seismic deconvolution is a typical ill-posed inverse problem. The regularization technique in terms of different prior information is used for a unique and stable solution. Due to the difference between prior information and the actual subsurface situation, it is hard to obtain a solution with satisfactory accuracy and resolution. This letter presents the dictionary learning as an efficient adaptive deconvolution method for the reflectivity reconstruction problem. Considering the curse of dimensionality of conventional dictionary learning and the suboptimal solution of the patch-based dictionary learning, we take the convolutional sparse coding (CSC) model as the dictionary learning method. In this method, the prior information can be obtained from the well-log data in the form of sparse CSC dictionary of reflectivity. On the assumption that the deposition of the subsurface layers is stable, the CSC dictionary extracted from the well-log data can also be applied in the whole work area. The CSC-based deconvolution can be seen as the adaptive deconvolution due to the independence of the assumption made about the reflectivity and seismic data. The process of the adaptive CSC-based deconvolution is divided into three parts. First, the CSC dictionary is learned from the well-log data. Then, the objective function is formulated by combining the CSC dictionary and the single-channel seismic record misfit term for the reconstruction of reflectivity. Finally, the objective function is efficiently solved with the coordinate descent approach. We illustrate the performance of our adaptive deconvolution with synthetic and field seismic data.

Summary Seismic data interpolation is an essential prerequisite for multiple removal, imaging and seismic attributes analysis. We proposed a specific deep convolutional autoencoder (DCAE) for seismic data interpolation in an unsurpervised approach. DCAE is trained on undersampled seismic data with randomly 50% missing traces and only the reconstruction error on the uncurropted data is considered without generating labels on missing traces. Synthetic data and field data testing shows that the structure of DCEC is capble to capture of the low-level statics prior information from undersampled seismic data, and the tranied model has a good interpolation performance on regular and irrgular sampled seismic data.

Automatic prostate segmentation from MR images is an important task in various clinical applications such as prostate cancer staging and MR-guided radiotherapy planning. However, the large appearance and shape variations of the prostate in MR images make the segmentation problem difficult to solve. Traditional Active Shape/Appearance Model (ASM/AAM) has limited accuracy on this problem, since its basic assumption, i.e., both shape and appearance of the targeted organ follow Gaussian distributions, is invalid in prostate MR images. To this end, the authors propose a sparse dictionary learning method to model the image appearance in a nonparametric fashion and further integrate the appearance model into a deformable segmentation framework for prostate MR segmentation. To drive the deformable model for prostate segmentation, the authors propose nonparametric appearance and shape models. The nonparametric appearance model is based on a novel dictionary learning method, namely distributed discriminative dictionary (DDD) learning, which is able to capture fine distinctions in image appearance. To increase the differential power of traditional dictionary-based classification methods, the authors' DDD learning approach takes three strategies. First, two dictionaries for prostate and nonprostate tissues are built, respectively, using the discriminative features obtained from minimum redundancy maximum relevance feature selection. Second, linear discriminant analysis is employed as a linear classifier to boost the optimal separation between prostate and nonprostate tissues, based on the representation residuals from sparse representation. Third, to enhance the robustness of the authors' classification method, multiple local dictionaries are learned for local regions along the prostate boundary (each with small appearance variations), instead of learning one global classifier for the entire prostate. These discriminative dictionaries are located on different patches of the prostate surface and trained to adaptively capture the appearance in different prostate zones, thus achieving better local tissue differentiation. For each local region, multiple classifiers are trained based on the randomly selected samples and finally assembled by a specific fusion method. In addition to this nonparametric appearance model, a prostate shape model is learned from the shape statistics using a novel approach, sparse shape composition, which can model nonGaussian distributions of shape variation and regularize the 3D mesh deformation by constraining it within the observed shape subspace. The proposed method has been evaluated on two datasets consisting of T2-weighted MR prostate images. For the first (internal) dataset, the classification effectiveness of the authors' improved dictionary learning has been validated by comparing it with three other variants of traditional dictionary learning methods. The experimental results show that the authors' method yields a Dice Ratio of 89.1% compared to the manual segmentation, which is more accurate than the three state-of-the-art MR prostate segmentation methods under comparison. For the second dataset, the MICCAI 2012 challenge dataset, the authors' proposed method yields a Dice Ratio of 87.4%, which also achieves better segmentation accuracy than other methods under comparison. A new magnetic resonance image prostate segmentation method is proposed based on the combination of deformable model and dictionary learning methods, which achieves more accurate segmentation performance on prostate T2 MR images.

Deconvolution is an essential step in seismic data processing. Sparse-spike deconvolution often is used to enhance the resolution of the seismic image by adding a model-driven regularization term. However, this method does not consider the features of the data, nor does it exactly describe the relationship between seismic data and the desired attribute (such as seismic reflectivity). We have developed a data and model dual-driven seismic deconvolution method based on error-constrained joint sparse representation (SR) using borehole measurement and surface seismic data. The combined features of the borehole reflectivity and the surface seismic data can be obtained through joint dictionary learning. With the help of the joint dictionary, the relationship between seismic waveforms and reflectivity is captured by the sparse coefficients. We construct the regularization term of deconvolution by alternately decomposing the error of the synthesized data via sparse reconstruction and the observed seismic data. Unlike model-driven methods, the constraint term of the new method can be established by the error-constrained SR. Based on this SR, the initial model of reflectivity is obtained to realize the sparse deconvolution of seismic data under the constraint of borehole data features. In general, this method is a data and model dual-driven deconvolution. Synthetic and field data tests demonstrate that this method can effectively improve the resolution and accuracy of deconvolution.

Slip/antiblocking agents

Image reconstruction from undersampled k-space data accelerates magnetic resonance imaging (MRI) by exploiting image sparseness in certain transform domains. Employing image patch representation over a learned dictionary has the advantage of being adaptive to local image structures and thus can better sparsify images than using fixed transforms (e.g. wavelets and total variations). Dictionary learning methods have recently been introduced to MRI reconstruction, and these methods demonstrate significantly reduced reconstruction errors compared to sparse MRI reconstruction using fixed transforms. However, the synthesis sparse coding problem in dictionary learning is NP-hard and computationally expensive. In this paper, we present a novel sparsity-promoting orthogonal dictionary updating method for efficient image reconstruction from highly undersampled MRI data. The orthogonality imposed on the learned dictionary enables the minimization problem in the reconstruction to be solved by an efficient optimization algorithm which alternately updates representation coefficients, orthogonal dictionary, and missing k-space data. Moreover, both sparsity level and sparse representation contribution using updated dictionaries gradually increase during iterations to recover more details, assuming the progressively improved quality of the dictionary. Simulation and real data experimental results both demonstrate that the proposed method is approximately 10 to 100 times faster than the K-SVD-based dictionary learning MRI method and simultaneously improves reconstruction accuracy.

Seismic facies analysis aims to identify clusters (groups) of similar seismic trace shapes, where each cluster can be considered to represent variability in lithology, rock properties, and/or fluid content of the strata being imaged. Unfortunately, it is not always clear whether the seismic data has a natural clustering structure. Cluster analysis consists of a family of approaches that have significant potential for classifying seismic trace shapes into meaningful clusters. The clustering can be performed using a supervised process (assigning a pattern to a predefined cluster) or an unsupervised process (partitioning a collection of patterns into groups without predefined clusters). We evaluate and compare different unsupervised clustering algorithms (e.g., partition, hierarchical, probabilistic, and soft competitive models) for pattern recognition based entirely on the characteristics of the seismic response. From validation results on simple data sets, we demonstrate that a self-organizing maps algorithm implemented in a visual data-mining approach outperforms all other clustering algorithms for interpreting the cluster structure. We apply this approach to 2D seismic models generated using a discrete, known number of different stratigraphic geometries. The visual strategy recovers the correct number of end-member seismic facies in the model tests, showing that it is suitable for pattern recognition in highly correlated and continuous seismic data.

Sparse representation based classification has led to interesting image recognition results, while the dictionary used for sparse coding plays a key role in it. This paper presents a novel dictionary learning (DL) method to improve the pattern classification performance. Based on the Fisher discrimination criterion, a structured dictionary, whose dictionary atoms have correspondence to the class labels, is learned so that the reconstruction error after sparse coding can be used for pattern classification. Meanwhile, the Fisher discrimination criterion is imposed on the coding coefficients so that they have small within-class scatter but big between-class scatter. A new classification scheme associated with the proposed Fisher discrimination DL (FDDL) method is then presented by using both the discriminative information in the reconstruction error and sparse coding coefficients. The proposed FDDL is extensively evaluated on benchmark image databases in comparison with existing sparse representation and DL based classification methods.