Quantifying the sensitivity of seismic facies classification to seismic attribute selection: An explainable machine learning study

During the past two decades, geoscientists have used machine learning to produce a more quantitative reservoir characterization and to discover hidden patterns in their data. However, as the complexity of these models increase, the sensitivity of their results to the choice of the input data becomes more challenging. Measuring how the model uses the input data to perform either a classification or regression task provides an understanding of the data-to-geology relationships which indicates how confident we are in the prediction. To provide such insight, the ML community has developed Local Interpretable Model-agnostic Explanations (LIME), and SHapley Additive exPlanations (SHAP) tools. In this study, we train a random forest architecture using a suite of seismic attributes as input to differentiate between mass transport deposits (MTDs), salt, and conformal siliciclastic sediments in a Gulf of Mexico dataset. We apply SHAP to understand how the model uses the input seismic attributes to identify target seismic facies and examine in what manner variations in the input such as adding band-limited random noise or applying a Kuwahara filter impact the models’ predictions. During our global analysis, we find that the attribute importance is dynamic, and changes based on the quality of the seismic attributes and the seismic facies analyzed. For our data volume and target facies, attributes measuring changes in dip and energy show the largest importance for all cases in our sensitivity analysis. We note that to discriminate between the seismic facies, the ML architecture learns a “set of rules” in multi-attribute space and that overlap between MTDs, salt, and conformal sediments might exist based on the seismic attribute analyzed. Finally, using SHAP at a voxel-scale, we understand why certain areas of interest were misclassified by the algorithm and perform an in-context interpretation to analyze how changes in the geology impact the model’s predictions.

Seismic attribute analysis of a fault zone in the Thebe field, NW Shelf, Australia

Seismic data is one of the main ways to characterize faults in the subsurface. Faults are 3D entities and their internal structure play a key role in controlling fluid flow in the subsurface. We aim to characterize a geologically sound fault volume that could be used for subsurface model conditioning. We present an attribute analysis of a normal fault from a high resolution seismic dataset of the Thebe Field, offshore NW Australia. We merge together a series of common attributes for fault characterization: dip, semblance and tensor (DST), and we also introduce a new Total Horizontal Derivative (THD) attribute to define the edges of the fault zone. We apply a robust statistical analysis of the attributes and fault damage definition through the analysis of 2D profiles along interpreted horizons. Using the THD attribute, we interpret a smaller width of the fault zone and a more straightforward definition of the boundaries than from the DST cube. Following the extraction of this fault volume, we define two seismic facies that are correlated to lithologies extracted from our conceptual model. We observe a wider fault zone at larger throws, which corresponds also to syn-rift sequence, hence more complex internal fault damage. Our method provides volumes at adequate scale for reservoir modeling and could therefore be used as a proxy for property conditioning.

Arctic Shelf Oil and Gas Prospects from Lower-Middle Paleozoic Sediments of the Timan–Pechora Oil and Gas Province Based on the Results of a Regional Study

The Timan–Pechora oil and gas province (TPP), despite the good geological and geophysical knowledge of its central and southern regions, remains poorly studied in the extreme northwestern part within the north of the Izhma–Pechora depression and the Malozemelsk–Kolguev monocline, and in the extreme northeast within the Predpaikhoisky depression. Assessing the oil and gas potential of the Lower Paleozoic part of the section is urgently required in the northwestern part of the TPP, the productivity of which has been proven at the border and in the more eastern regions of the province (Pechora–Kolva, Khoreyverskaya, Varandei–Adzva regions), that have been evaluated ambiguously. A comprehensive interpretation of the seismic exploration of regional works was carried out, with the wells significantly clarifying the structural basis and the boundaries of the distribution of the main seismic facies’ complexes. The capabilities of potentially oil- and gas-producing strata in the Silurian–Lower Devonian were studied. An analysis of migration routes in transit strata used for basin modeling in order to reconstruct the conditions of oil and gas formation that are common in the land and water areas of the Arctic zone of the TPP was carried out. Modeling allowed us to reach an understanding of the formation of large zones with possible accumulations of hydrocarbons, including the time at which the formation occurred and under what conditions, to establish space–time links with possible centers of generation to identify migration directions and, based on a comparison with periods of intensive generation of hydrocarbons both directly located within the work area and beyond (noting the possible migration), to identify zones of the paleoaccumulation of hydrocarbons. The body of existing literature on the subject made it possible to outline promising oil and gas accumulation zones, with the allocation of target objects for further exploration in the Lower Paleozoic part of the section.

Stratigraphic Trap Potential in the Lower Cretaceous Ratawi Interval, Partitioned Zone PZ, Saudi Arabia and Kuwait

The Lower Cretaceous Ratawi Oolite Formation is among the most prolific reservoirs in the PZ, having produced a significant amount of oil since the 1950's. The Ratawi is interpreted as a low angle carbonate ramp, with high-energy grainstone facies developing on structural highs. Production is focused on these structural highs, with very few well penetrations off structure. Recent work has identified potential Ratawi stratigraphic traps in prograding clinoforms along the flanks of the North Fuwaris structural high. Core data from Ratawi wells illustrate the interplay of depositional environment and diagenesis on reservoir quality. Gross depositional environment (GDE) maps created from the integration of seismic facies and core observations indicate the stratigraphic trap lies in the ramp slope. Reservoir quality variability of the ramp slope across the PZ is explained by the diagenetic history of the Ratawi. Early equant calcite cement develops from substantial meteoric runoff and lowers porosity, while later dissolution enhances reservoir quality. The area of interest is isolated from potential meteoric inputs; we do not expect equant calcite cement or the associated reduction in reservoir quality. Seismic interpretation was performed on recently acquired PZ 3D data to map the Ratawi section. Clinoforms (inclined geometry) were mapped along the western flank of the North Fuwaris high. These facies appear to have developed as a result of progradation to the NW and are indicative of good reservoir development. Leads were generated using the depth structure and GDE maps, supported by amplitude extraction and seismic inversion volumes. Amplitudes extracted from the clinoform shows that the strongest anomaly is along the structurally highest part of the horizon and the anomaly weakens downdip. High amplitudes could be a proxy for reservoir (porosity), and sharp turn-off in amplitude might indicate that lateral and updip facies changes to non-reservoir which is needed for an effective seal. Recent seismic inversion performed on the Ratawi interval shows a good match between the Acoustic Impedance (AI) from logs and the computed AI from the seismic. The Ratawi Oolite appears as a low impedance interval between overlying Ratawi Limestone and underlying Makhul. Porosity estimated from AI volumes appear to support possible Ratawi reservoir development along the flanks of North Fuwaris and Wafra highs.

Sedimentology and depositional environments of Murihiku Supergroup sediments exposed in the Southland Syncline, New Zealand: Implications for reservoir potential in the Great South Basin

<p>A considerable amount is known about the biostratigraphy and organic geochemistry of the Murihiku Supergroup sediments exposed in coastal outcrops of the Southland Syncline, New Zealand. Much less work has been undertaken on the sedimentology of these strata, or understanding their depositional environments and depositional trends through time. What these implications are for reservoir prospectivity in the adjacent Great South Basin, has also had little study focused on it. This thesis addresses these issues by undertaking outcrop-based sedimentological and facies interpretations of these rocks, thin-section based petrographic composition and provenance analysis, augmented by X-Ray Diffraction (XRD) and Scanning Electron Microscopy (SEM), as well as porosity and permeability measurements from outcrop core plugs. Petroleum industry seismic data has additionally enabled seismic facies mapping of Murihiku rocks in the offshore Great South Basin. Outcrop observations point to a progressive change in depositional setting, from shelf / upper slope settings during the Late Triassic, to base of slope turbidite deposition in the Early Jurassic. This transgression is followed by regression into fluvial settings in the youngest outcropping Murihiku rocks in the study of Middle Jurassic age. Petrographically the sandstones are feldspathic and lithic arenites and feldspathic and lithic wackes. Provenance suggests derivation from an evolving, intermediate arc that was becoming more siliceous through Late Triassic and Middle Jurassic time. Diagenesis is characterised by early calcite and chlorite precipitation which have almost completely destroyed any primary porosity. Any secondary micro porosity has subsequently been infilled through dissolution of framework grains and zeolitization. SEM and core plug porosity and permeability measurements corroborate the diagenetic changes observed petrographically, with only fluvial facies of Middle Jurassic (Upper Temaikan) age showing any measureable porosity or permeability. As a result, reservoir potential for the Late Triassic to Middle Jurassic, Murihiku Supergroup rocks analysed in this study is low. Younger Murihiku sandstones which are postulated to occur offshore in the Great South Basin are likely to be less influenced by burial diagenesis. As shown from North Island occurrences, these younger successions hold some potential.The reservoir potential for these youngest portions of the Murihiku succession therefore remains positive, both in the Great South Basin, as well as other frontier areas of Zealandia, and continue to provide an exploration target for the petroleum industry.</p>

Sedimentology and depositional environments of Murihiku Supergroup sediments exposed in the Southland Syncline, New Zealand: Implications for reservoir potential in the Great South Basin

<p>A considerable amount is known about the biostratigraphy and organic geochemistry of the Murihiku Supergroup sediments exposed in coastal outcrops of the Southland Syncline, New Zealand. Much less work has been undertaken on the sedimentology of these strata, or understanding their depositional environments and depositional trends through time. What these implications are for reservoir prospectivity in the adjacent Great South Basin, has also had little study focused on it. This thesis addresses these issues by undertaking outcrop-based sedimentological and facies interpretations of these rocks, thin-section based petrographic composition and provenance analysis, augmented by X-Ray Diffraction (XRD) and Scanning Electron Microscopy (SEM), as well as porosity and permeability measurements from outcrop core plugs. Petroleum industry seismic data has additionally enabled seismic facies mapping of Murihiku rocks in the offshore Great South Basin. Outcrop observations point to a progressive change in depositional setting, from shelf / upper slope settings during the Late Triassic, to base of slope turbidite deposition in the Early Jurassic. This transgression is followed by regression into fluvial settings in the youngest outcropping Murihiku rocks in the study of Middle Jurassic age. Petrographically the sandstones are feldspathic and lithic arenites and feldspathic and lithic wackes. Provenance suggests derivation from an evolving, intermediate arc that was becoming more siliceous through Late Triassic and Middle Jurassic time. Diagenesis is characterised by early calcite and chlorite precipitation which have almost completely destroyed any primary porosity. Any secondary micro porosity has subsequently been infilled through dissolution of framework grains and zeolitization. SEM and core plug porosity and permeability measurements corroborate the diagenetic changes observed petrographically, with only fluvial facies of Middle Jurassic (Upper Temaikan) age showing any measureable porosity or permeability. As a result, reservoir potential for the Late Triassic to Middle Jurassic, Murihiku Supergroup rocks analysed in this study is low. Younger Murihiku sandstones which are postulated to occur offshore in the Great South Basin are likely to be less influenced by burial diagenesis. As shown from North Island occurrences, these younger successions hold some potential.The reservoir potential for these youngest portions of the Murihiku succession therefore remains positive, both in the Great South Basin, as well as other frontier areas of Zealandia, and continue to provide an exploration target for the petroleum industry.</p>

Adaptive Gaussian Mixture Model and Convolution Autoencoder Clustering for Unsupervised Seismic Waveform Analysis

Seismic facies analysis can effectively estimate reservoir properties and seismic waveform clustering is a useful tool for facies analysis. We developed a deep learning-based clustering approach called the modified deep convolutional embedded clustering with adaptive Gaussian mixture model (AGMM-MDCEC) for seismic waveform clustering. Trainable feature extraction and clustering layers in AGMM-MDCEC are implemented using neural networks. The two independent processes of feature extraction and clustering are fused, such that extracted features are modified simultaneously with the results of clustering. A convolutional autoencoder is used in the algorithm for extracting features from seismic data and reduce data redundancy. At the same time, weights of clustering network are fined-tuned through iteration to obtain state-of-the-art clustering results. We apply our new classification algorithm to a data volume acquired in western China to map architectural elements of a complex fluvial depositional system. Our proposed method obtains superior results over those provided by traditional K-means, Gaussian mixture model, and some machine learning methods, and improves the mapping of the extent of the distributary system.

Palaeogeography of a Mid Miocene Turbidite Complex, Moki Formation, Taranaki Basin, New Zealand.

<p>The Moki Formation, Taranaki Basin, New Zealand, is a Mid Miocene (Late Altonian to Early Lillburnian) sand-rich turbidite complex bounded above and below by the massive bathyal mudstone of the Manganui Formation. The Moki Formation is a proven hydrocarbon reservoir with its stacked, thick, tabular sandstone packages totalling more than 300 m in places. Previous regional studies of the formation have been based primarily on well data and resulted in varying palaeogeographic interpretations. This study, restricted to the southern offshore region of the basin, better constrains the spatial and temporal development of the Moki Formation by combining well data with seismic interpretation to identify key stratal geometries within the sediment package. Nearly 30,000 km of 2D seismic reflection profiles and two 3D surveys, along with data from 18 wells and three cores were reviewed and key sections analysed in detail. Seismic facies have been identified which provide significant insights into the structure, distribution and progressive development of the Moki Formation. These include: a clearly defined eastern limit of the fan complex, thinning and fining of the distal turbidite complex onto the basin floor in the north and west, evidence of fan lobe switching, spectacular meandering channel systems incised into the formation at seismic scales, and the coeval palaeoshelf-slope break in the south east of the basin. In addition, a Latest Lillburnian / Waiauan turbidite complex has been mapped with large feeder, fan and bypassing channels traced. This study presents an improved palaeogeographic interpretation of the Moki Formation and the younger, Latest Lillburnian / Waiauan-aged, turbidite complex. This interpretation shows that during the Late Altonian, sandstone deposition was localised to small fan bodies in the vicinity of Maui-4 to Moki-1 wells. A bathymetric deepening during the Clifdenian is identified, which appears to have occurred concurrently as the establishment of the Moki Formation fan system, centred around the southern and central wells. With continued sediment supply to the basin floor, the fan system prograded markedly northward and spilled onto the Western Stable Platform during the early Lillburnian. Sand influx to the bathyal basin floor abruptly ceased and large volumes of mud were deposited. By the Waiauan stage, sands were again deposited at bathyal depths on fan bodies and carried to greater depths through a complex bypassing channel system.</p>

Palaeogeography of a Mid Miocene Turbidite Complex, Moki Formation, Taranaki Basin, New Zealand.

<p>The Moki Formation, Taranaki Basin, New Zealand, is a Mid Miocene (Late Altonian to Early Lillburnian) sand-rich turbidite complex bounded above and below by the massive bathyal mudstone of the Manganui Formation. The Moki Formation is a proven hydrocarbon reservoir with its stacked, thick, tabular sandstone packages totalling more than 300 m in places. Previous regional studies of the formation have been based primarily on well data and resulted in varying palaeogeographic interpretations. This study, restricted to the southern offshore region of the basin, better constrains the spatial and temporal development of the Moki Formation by combining well data with seismic interpretation to identify key stratal geometries within the sediment package. Nearly 30,000 km of 2D seismic reflection profiles and two 3D surveys, along with data from 18 wells and three cores were reviewed and key sections analysed in detail. Seismic facies have been identified which provide significant insights into the structure, distribution and progressive development of the Moki Formation. These include: a clearly defined eastern limit of the fan complex, thinning and fining of the distal turbidite complex onto the basin floor in the north and west, evidence of fan lobe switching, spectacular meandering channel systems incised into the formation at seismic scales, and the coeval palaeoshelf-slope break in the south east of the basin. In addition, a Latest Lillburnian / Waiauan turbidite complex has been mapped with large feeder, fan and bypassing channels traced. This study presents an improved palaeogeographic interpretation of the Moki Formation and the younger, Latest Lillburnian / Waiauan-aged, turbidite complex. This interpretation shows that during the Late Altonian, sandstone deposition was localised to small fan bodies in the vicinity of Maui-4 to Moki-1 wells. A bathymetric deepening during the Clifdenian is identified, which appears to have occurred concurrently as the establishment of the Moki Formation fan system, centred around the southern and central wells. With continued sediment supply to the basin floor, the fan system prograded markedly northward and spilled onto the Western Stable Platform during the early Lillburnian. Sand influx to the bathyal basin floor abruptly ceased and large volumes of mud were deposited. By the Waiauan stage, sands were again deposited at bathyal depths on fan bodies and carried to greater depths through a complex bypassing channel system.</p>