Seismic geomorphology-based facies classification

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 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.

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.

In a companion paper (Nivlet et al.,2005), we have described an application of seismic pre‐stack inversion to retrieve acoustic and shear impedance models in Girassol field. The next step, which will be discussed in this paper, is the lithoseismic interpretation of the inversion results. Traditional seismic facies analysis consists in extracting various amplitude maps of a reservoir level, and producing probabilistic or deterministic seismic facies maps. In this paper, we will present a workflow reaching beyond this traditional approach, by producing a 3D geological facies proportion model, which can then be used as a direct constraint to high resolution geological modeling techniques. The workflow consists of 5 main steps: • well log interpretation calibrated on core data; • upscaling of the facies database at the seismic scale; • choice of appropriate seismic attributes to discriminate betweenn the different facies, compensating for instance for the shale compaction with depth; • probabilistic seismic facies analysis from the chosen seismic attribute cubes supervised by upscaled well information; • post‐interpretation of seismic facies analysis results in terms of mean proportions of geological facies; • consistency check between the facies model and seismic impedances.

To solve the problem of low resolution of seismic data from granite weathering crust and difficulty in predicting effective reservoir distribution. Based on the research idea of combining geology and seismology, seismic facies characteristics and effective reservoir distribution of granite weathering crust are studied by means of frequency division reconstruction processing, seismic facies analysis and sensitive attributes interpretation. This method consists of the following four main processes: (1) Using improved Morlet wavelet function for frequency division and reconstruction processing of seismic data; (2) From the perspective of geological evolution, analyzing the geological origin of seismic facies; (3) Based on the results of well- seismic combination, clarifying the types of effective reservoirs and predicting their distribution; (4) Comprehensive research indicating favorable reservoirs distribution. The results indicate that (1) a considerably improved resolution of seismic data processed by frequency division reconstruction is obtained based on the improved Morlet wavelet function. Thus, granite reservoirs then have clearer internal reflection characteristics. (2) Affected by differences in weathering, the seismic facies in granite weathering crusts can be classified as follows: Strong-amplitude facies with amplitudes of 23,700 − 32,800, corresponding to eluvium, represents the late stage of weathering crust evolution with high-intensity weathering. Medium-amplitude facies with amplitudes of 5,250 − 24,000, corresponding to dissolution layer, represents a middle stage of weathering crust evolution with moderate weathering. And weak-amplitude-blank facies with amplitudes of 0–5,330, corresponding to disintegration layer, represents an early stage of weathering crust evolution with weak weathering. (3) Effective reservoirs can be classified according to well-seismic calibrations and combined with reservoir spaces, porosities and permeabilities, and the logging and oil-bearing characteristics of the reservoirs. Type I effective reservoir comprises medium-amplitude facies, with favorable reservoir conditions and high-yield oil-bearing properties; this type is distributed in the western slopes, with an elevation of 1,300-2,500 m. Type II effective reservoir comprises strong-amplitude facies, with poor oil-bearing properties; this type is mainly distributed in the southeast and north slopes of the study area, with a relatively limited range, and the elevation is mostly below 1,500 m. Type III effective reservoir comprises weak-amplitude-blank facies, with poor reservoir conditions and poor oil-bearing properties; this type is widely distributed at various elevations. (4) The areas where medium-amplitude facies have developed are the most favorable reservoir exploration zones for granite in Dongying Sag, Bohai Bay Basin.

One of the key approaches to look for a stratigraphic prospect in the carbonate reservoir is to map out areas that have changes in the rock facies laterally. A workflow of the 2009–2010 exploration study focuses on four key components: Regional new play concepts, Well log electrofacies, Seismic facies classification, and Prospect evaluation. This helps to predict stratigraphic traps and lithofacies changes within the study area. The electrofacies analysis calibrated to a reprocessed high resolution 3D seismic has been carried out in conjunction with Natih and Shuaiba prospecting concept and analogue fields compilation. The aim of this integration of well log electrofacies and Seismic facies is to generate a 3D lithofacies model in the known areas where many wells had been drilled, and then use 3D seismic facies classification to predict lithofacies in the areas where no wells drilled. Moreover, this integration will provide an understanding of lateral and vertical facies changes of reservoirs and seals within target intervals, which are critical elements for defining stratigraphic traps. The electrofacies analysis was carried out by integrating core data and well log to build a rock type model. This correlation allows rockfacies to be classified in the cored wells and predict those facies in the uncored wells. The results of this electrofacies analysis have demonstrated reservoir property predictions in terms of porosity and permeability distribution associated with each rock facies for all wells. On the other hand, the seismic facies classification was analyzed using Neural Network approach based on multiple calibration of seismic attributes, such as High resolution 3D seismic (AMP), Far offset angle stack (FAR), and Acoustic impedance (AI). The first pass of seismic facies classification was performed without well data input, so called "unsupervised model". The model will determine a possibility of seismic facies from all input attributes, and generate multiple 3D facies output for both Natih and Shuaiba Formations. The most important part of the workflow is the merging between well log electrofacies and the 3D seismic facies model. The electrofacies derived from well log provide a detailed specific rock types at well locations, and the seismic facies would fill the gap in the areas without well control. Initially, the seismic facies were calibrated to the electrofacies rock types at each well location, and the prediction was extended laterally throuh out the area following the defined structural framework. The final product of this workflow is a "Most Likely" 3D lithofacies within the target intervals. A 3D probability volume was also generated to quantify the level of uncertainty, especially in the areas lacking of well control. The resulting 3D lithofacies model has reasonably demonstrated the ability to predict rock facies distribution within the Natih and Shuaiba Formations. The output model has provided a key element to identify prospective areas for stratigraphic traps within the study area. The 3D lithofacies model based on this integrated workflow was also useful for the prospect evaluation and the ranking process of the Natih truncation and Shuaiba clinoform prospects identified in the exploration study. The results have confirmed the presence of the Natih-C1 reservoir and also the predicted rock facies distribution in the upper part of Shuaiba Clinoform called "shoal" anticipated to be good reservoir quality in each prospect. In addition, the results of this work have been utilized for development activities, such as the updated 2011 3D static and dynamic reservoir models, which have demonstrated an improvement in terms of production history matching, reserves determination, and production forecast.

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 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.

An integrated sedimentological workflow was applied to the methane-hydrate exploration project in the eastern Nankai Trough area, central Japan, as a useful tool to investigate the distributions and actual volume of methane hydrates. Since previous research revealed that most of the methane hydrates in the eastern Nankai Trough area occur in matrix pores of submarine-fan turbidite sandstones, the facies distribution of turbidite sandstones may be one of the important keys to evaluate the occurrence of methane hydrates. Accordingly, this study conducted a) core facies analysis, b) log facies analysis mainly using FMI, c) depositional sequence division, d) seismic facies analysis on the 2D/3D seismic sections, and e) seismic geomorphological analysis on the 3D seismic survey data to map submarine-fan turbidite facies distributions for 17 depositional sequence horizons in the targeted Pleistocene interval. The obtained facies maps reveal that submarine-fan depositional styles changed throughout Pleistocene from a braided channel type, through small radial fan, trough-fill fan, and muddy sheet fan types, to a channel-levee system type. As the next step, the facies maps of each depositional sequence were overlaid with bottom simulating reflector (BSR) distributions as a proxy of methane hydrates. The overlaid maps indicate that the BSRs occur on feeder channels, distributary channels, and proximal lobes of submarine fans, suggesting that methane hydrates selectively occur in coarser grained portions of a submarine fan. The facies maps were also used for calculation of average net-to-gross ratio of the methane hydrate occurrence zones, as the maps provide information on the spatial distributions of seismic facies class, which has individual value of net-to-gross ratio. Finally, this study analyzed reservoir-scale geologic bodies, such as depositional lobes and braided channels of a submarine fan, using 3D seismic survey data, and constructed a detailed geologic model for simulations of methane-hydrate generation and production.

Facies models of two coalfields in Canada have been developed through seismic facies analysis. Fluvial-deltaic Upper Cretaceous coals near Camrose, Alberta, are characterized by hummocky, subparallel, high-amplitude reflections that are laterally continuous over distances of several kilometers. Deposition in an extensive delta-plain swamp is inferred. Coals of Pennsylvanian age in the Pictou coalfield, Nova Scotia, occur in two distinct seismic facies. Parallel to subparallel, relatively continuous reflections are interpreted to define a basinal facies comprising thick coals and lacustrine sediments. Basinal margin facies contain a greater abundance of variable clastic sediments and are recognized by discontinuous, subparallel to chaotic reflections. Further improvements n stratigraphic analysis will require synthetic seismograms to analyze individual reflections in terms of the thickness and separation of seams, and to enable interseam multiples to be recognized.

Seismic facies analysis was performed on approximately 1540 km of 2D high-resolution multi-channel seismic data from the German North Sea. Six seismic facies were defined in the study area. Facies 1 covers the Quaternary sand depositional system. Facies 2, 3, and 4 represent three types of tunnel-valley infill. These valleys were formed during different glacial periods in the study area. All of them are characterized by erosional incision into underlying deposits. Facies 5 probably represents the youngest post-glacial sediments after the retreat of last glacier. Facies 6 is a typical seismic reflector for peat.

Seismic stratigraphic interpretation of megasequence and sequence framework in the Upper Cambrian through Devonian formations in the central part of the Darling Basin, western New South Wales, Australia

Seismic facies analysis is not a deterministic and simple task. Usually, facies analysis is performed through the following steps [1]: 1. Geological oriented spatial and temporal segmentation of seismic traces (input space); 2. Seismic attributes selection (variable space); 3. Choosing the optimal number of classes (facies) and algorithm iterations; 4. Training and classification of the selected attributes using some statistical or neural networks methods (pattern space); 5. Building and interpreting facies map. Normally, the geological oriented spatial and temporal segmentation should be carefully done, because any horizon interpretation error could lead to wrong or very noisy facies results. The attributes selection is another complex task, because they should be physically consistent and statistically independent. It’s common to use the whole seismic trace amplitudes around the region of interest [2]. Nowadays, Self Organizing Maps (SOM), or Kohonen Maps [3], has become one of the most popular tools to build seismic facies maps [4]. But, it’s still empirical how to choose the number of classes and the best seismic attributes to discriminate geological features from seismic data. This project presents a new alternative to extract seismic pattern attributes and a new methodology to build seismic facies maps. We propose using Wavelet Transform to identify singularities in each geological oriented segment of the temporal seismic trace and then using SOM in a two-level approach [5]. To illustrate this technique it was applied to real data from a deep-water field in the Campos Basin, Brazil.

Seismic stratigraphy has developed almost as three separate disciplines. The first, pioneered by Vail and others (1977), is seismic sequence analysis. Sequence analysis is the study of recognizing and correlating regional stratal surfaces to define genetically related rock units that represent discrete chronostratigraphic intervals (Van Wagoner et al., 1988). Seismic data are used extensively in sequence stratigraphy, as primary seismic reflections are generated by time-correlative bedding surfaces, and surfaces that separate sets of contemporaneous depositional systems (systems tracts) (Van Wagoner et al., 1988).

Classification of seismic facies is done by clustering seismic data samples based on their attributes. Year after year, 3D datasets used by exploration geophysics increase in size, complexity, and number of attributes, requiring a continuous rise in the classification performance. In this work, we explore the use of Graphics Processing Units (GPUs) to perform the classification of seismic surveys using the well-established Machine Learning (ML) method k-means. We show that the high-performance distributed implementation of the k-means algorithm available at the RAPIDS library can be used to classify facies in large seismic datasets much faster than a classical parallel CPU implementation (up to 258-fold faster in NVIDIA V100 GPUs), especially for large seismic blocks. We tested the algorithm with different real seismic volumes, including Netherlands, Parihaka, and Kahu (from 12GB to 66GB).