Seismic Attributes Research Articles

Sensitivity analysis of seismic attributes parametrization to reduce misinterpretations: Applications to deepwater channel complexes

Geoscientists apply algorithms such as seismic attributes to better interpret depositional systems that enhance various aspects of the seismic data. However, they are limited by the original seismic amplitude or frequency content, data quality, and algorithm parameters considered.Additionally, our capacity to interpret depositional system architecture is limited by seismic resolution, which results in potential misinterpretations associated with the correct position of stratigraphic features. Resolution also sets a minimum thickness that can be observed in seismic data. This is particularly important as mapping reservoir architecture (geobody size, shape, and stacking patterns) in the subsurface is critical for exploring and producing hydrocarbons, CO2 storage, and geothermal resource development since it can define connectivity or compartmentalization of flow zones.To address these concerns, we investigated five synthetic seismic volumes from low to high-dominant frequencies of 15 Hz, 30 Hz, 60 Hz, 90 Hz, and 180 Hz based on an architectural model of an outcropping deepwater channelized slope system in the Magallanes Basin, Chile. We analyzed 1) how seismic bandwidth affects the resolution of stacked stratigraphic features (i.e., deepwater channel elements and Mass Transport Deposits (MTDs)) and their subsequent seismic interpretation, and 2) the effect of different seismic attributes commonly employed in channel interpretation on our data to understand the “mixing” or “vertical smearing” of stratigraphic features by comparing the seismic with the true geological model 3) we explored how the attributes’ parametrization affects the imaging of differently sized features by modifying the analysis window in each case from ± 2ms to ± 50 ms. Finally, 4) we evaluated the effect of different noise levels in the sensitivity analysis. Resultsshow that the “mixing” of events occurs mainly as a result of 1) the seismic bandwidth, 2) the algorithm used for each seismic attribute calculation, 3) the attribute vertical analysis window, and 4) the signal-to-noise ratio of the data. Broadband, higher frequency data, and small analysis windows provide clearer images of the stacked channels. In contrast, low-frequency data and larger analysis windows result in more mixing or “composite” appearances, affecting interpretations and net-to-gross estimates, especially in small-size stratigraphic features such as individual channel elements and Mass Transport Deposits (MTDs). Our observations warn of potential misinterpretations in applying default attributes to actual seismic data, especially in geometrical attributes and window-dependent ones. Recognizing these misinterpretations is paramount for reconstructing deepwater architecture (this study), sedimentary and structural studies for drilling locations, reserves estimation, and overall uncertainty assessment.

Subsurface fluid flow: A case study from the Indo-Gangetic peripheral foreland basin

Subsurface fluid flow involves migration of fluids from source to surface through a wide range of geologic structures, and thus plays a crucial role in unlocking potential hydrocarbon plays. Such studies have been mainly carried out in the marine environment by means of high-resolution seismic data in unravelling the fluid flow system and remains poorly documented for on-land data. This research attempts to explore the fluid flow activity in onshore petroliferous Indo-Gangetic peripheral foreland basin of India by using high-quality 3D seismic reflection data. Several seismic attributes have been utilised for efficiently describing the patterns of subsurface fluid flow structures. The attribute-responses of fluid flow are fused into a single attribute, called the Fluid Cube meta-attribute, by designing a workflow based on artificial neural network, which has enabled to elucidate the subsurface fluid migration routes. The subsurface fluid flow features are imaged as relatively vertical mounded structure with conical vent-like morphology. Internally, these features are associated with moderately distorted reflections. The reflections at the top are vertically stacked with medium to high amplitudes. The Fluid Cube meta-attribute demonstrates that subsurface fluid vertically migrates from the Neoproterozoic strata through minute fracture networks and weaker strata of the overlying Tertiary succession. The analysis of surficial geochemical anomalies corroborates quite reasonably with these observations. Thus, the Ganga peripheral foreland basin could be considered a promising area with potential leads that can be unlocked for hydrocarbon exploration. The analysis presented through this research could be efficiently carried out over other onshore basins worldwide.

Collaborative Approach Solves Upstream R&D Challenges

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 211115, “IBTIKAR Digital Laboratory: A Collaborative Approach Toward Research and Development Challenges in Oil and Gas Upstream,” by Rauf Iqbal, Pranav Kumar, and Muhammad Aamir, ADNOC, et al. The paper has not been peer reviewed. _ The complete paper describes creation of an approach involving a research and development (R&D) environment that supports the business community. The operator’s Ibtikar laboratory is the result of an initiative to develop and provide in-house infrastructure (hardware and software) support within a national oil company’s operations. This synopsis is mostly devoted to a description of R&D challenges faced by the developers and the steps taken to overcome them. Efficient Seismic Attribute Extraction for Azimuthal or Angle Stacks An asset team acquired a few high-density broadband single-source, single-sensor seismic surveys characterized by an enormous amount of data. From acquisition to processing, multiple subcubes (90 GB/cube) were produced for the purpose of capturing subtle geological features. On a conventional machine, loading of the seismic cubes took 40 hours and realizing them took more than 7 days—and this was for only one seismic attribute. To fully use the data in the static and dynamic models of the respective reservoirs, the asset team collaborated with the team behind the described “digital laboratory” to provide a solution best suited for such a large processing requirement. The latter team analyzed the existing system, estimated requirements, and provided a high-performance computing (HPC) system that could handle the massive workload. Table 1 of the complete paper shows the results of a comparison of the conventional workstation with the HPC system. Computationally Intensive Fault-Likelihood Attribute Generation and Fault-Extraction Process In complex geological settings, fault identification and extraction of fault planes has always been a challenging task using traditional seismic-interpretation techniques. Conventional discontinuity-based seismic attributes often face limitations in delineating complex fault patterns, especially in areas of strike/slip and compressional-tectonics regimes. To address these limitations, geoscientists often use sophisticated seismic attributes such as fault likelihood, which follows a sequential multiattribute work flow to generate a fault-imaging volume used later to extract fault planes directly. The key challenge associated with running a computationally intensive attribute with a traditional workstation is that the process is time-consuming and performance is highly compromised because of hardware limitations. The fault-likelihood process itself generates six different seismic volumes during the automatic computation process. The task becomes even more challenging when multiple scenarios are required with varying parameters to assess and finalize the outcomes of fault-imaging volumes and extraction of fault planes. The digital laboratory was used to run this computing-intensive fault-likelihood process on a high-end machine with 800 GB random access memory and 96 cores with improved efficiency. It took only 10 days to run this multiattribute work flow on a seismic volume after running multiple scenarios using different attribute parameters to reduce fault uncertainty and automatically extract fault planes.

Seismic attribute and well-log analysis for channel characterization in the upper San Andres and Grayburg formations of the Midland Basin, Texas

The Permian Basin is one of the most prolific, and currently one of the most active, oil and gas basins in the USA. The Lower Permian strata in the Permian Basin have produced more than 14 billion barrels of oil (BBO), making it the largest volume of hydrocarbon in the basin. Sedimentation in the Midland Basin during late Leonardian through early Guadalupian (ca. 272–269 Ma) resulted in progradation of shelf edge and ultimately closure of the basin by Middle Permian time. We analyzed a merged seismic survey covering parts of the Permian Basin (i.e., Central Basin Platform and Midland Basin) in Andrews, Ector, and Midland Counties, Texas. The seismic survey and well logs show the presence of gently dipping (ca. 1°) clinoforms in the Upper San Andres and Grayburg Formations on the eastern edge of the Central Basin Platform and western Midland Basin. The seismic attributes, curvature, and spectral decomposition identify low sinuosity slope channels oriented north-south, but such channels do not appear beyond the slope. The shelf edge shifts from north to south during deposition of the Upper San Andres and Grayburg Formations. We identify five basinward shifts noted by the migration of the shelf edge toward the basin center and the presence of channel features along the depositional slope. The petrophysical analysis indicates that channels cut into carbonate rocks and are filled by siliciclastic sediments; this interpretation is supported by the most negative curvature anomalies along the channel axes caused by the differential compaction between the carbonate and siliciclastic rocks. There are a few channels with a northwest-southeast strike, which matches the direction of the Concho Lineament observed by satellite data. Such observations are consistent with previous interpretations of the northern Midland Basin closure during Middle-Late Permian time.

Visualizing subtle structural and stratigraphic features on 3D seismic-reflection data: A case study from offshore Libya

Seismic-reflection data contain residual noise after processing, which can cause geologic misinterpretation of noise-sensitive seismic attributes. For detailed subsurface imaging, seismic data conditioning can enhance the visualization of subsurface reflection features down to the limit of seismic resolution. This study on the Gabes-Tripoli Basin of western offshore Libya demonstrates the potential of postprocessing seismic data conditioning using a standard industry 3D-seismic data set. The seismic data exhibit distinct reflection discontinuities and configurations interpreted as folds, reverse faults, and crustal- and gravity-driven normal faults. Other reflection discontinuities are interpreted as imaging the external form and internal architecture of buried carbonate platforms. The seismic-reflection interpretation finds that seismic data conditioning by filtering strongly supports the interpretation of the 3D geometry, type, and trend of distinct subsurface reflection features, particularly if used as input for a structural-attribute generation. Postprocessing seismic conditioning initially improved the signal-to-noise ratio by structure-oriented filtering with edge preservation. Application of this filter configuration emphasized subtle geologic features supporting, e.g., the detection of faults close to the limit of the seismic resolution. At the same time, the filtering resulted in a higher lateral continuity of the individual seismic reflectors, supporting the autotracking of the horizons. Structural attributes generated from the conditioned data such as the variance and curvature imaged more subsurface reflection detail when compared with the structural attributes generated from nonconditioned data. The filter-based workflow proposed can be applied in most seismic interpretation software packages and is recommended to be used as a standard procedure preceding a structure-attribute calculation and structural interpretation of the seismic-reflection data of limited quality.

Machine learning elucidates the anatomy of buried carbonate reef from seismic reflection data

A carbonate build-up or reef is a thick carbonate deposit consisting of mainly skeletal remains of organisms that can be large enough to develop a favourable topography. Delineation of such geologic features provides important input in understanding the basin's evolution and petroleum prospects. Here, we introduce a new attribute called the Reef Cube (RC) meta-attribute that has been computed by fusing several other seismic attributes that are characteristics of the reef through a supervised machine-learning algorithm. The neural learning resulted in a minimum nRMS error of 0.28 and 0.30 and a misclassification percentage of 1.13% and 1.06% for the train and test data sets. The Reef Cube meta-attribute has efficiently captured the anatomy of carbonate reef buried at ∼450 m below the seafloor from high-resolution 3D seismic data in the NW shelf of Australia. The novel approach not only picks up the subsurface architecture of the carbonate reef accurately but also accelerates the process of interpretation with a much-reduced intervention of human analysts. This can be efficiently suited for delimiting any subsurface geologic feature from a large volume of surface seismic data.

An integrated study for seismic structural interpretation and reservoir estimation of Sawan gas field, Lower Indus Basin, Pakistan

The information about the subsurface structure, type of fluids present in the reservoir, and physical properties of the rocks is essential for identifying potential leads. The integrated approach of petrophysical analysis, seismic data interpretation, seismic attributes analysis, lithology, mineralogy identification, and Gassmann fluid substitution were used for this purpose. The structural interpretation with the help of seismic data indicated the extensional regime with horst and graben structures in the study area. The two negative flower structures are cutting the entire Cretaceous deposits. The depth contour map also indicate favorable structures for hydrocarbon accumulation. The four possible reservoir zones in the Sawan-01 well and two zones in the Judge-01 well at B sand and C sand levels are identified based on well data interpretation. The main lithology of the Lower Goru Formation is sandstone with thin beds of shale. The clay types confirm the marine depositional environment for Lower Goru Formation. The water substitution in the reservoir at B sand and C sand levels indicated increased P-wave velocity and density. The water substitution affected the shear wave velocity varies slightly due to density changes. The cross plots of P-impedance versus Vp/Vs ratio differentiate the sandstone with low P-impedance and low Vp/Vs ratio from shaly sandstone with high values in the reservoir area. The P-impedance and S-impedance cross plot indicate increasing gas saturation with a decrease in impedance values. The low values of Lambda-Rho and Mu-Rho indicated the gas sandstone in the cross plot.

Anelastic dispersion and attenuation of P- and SV-wave scattering by aligned nonisothermal inclusions of finite thickness

We have investigated the anelastic dispersion and attenuation of P- and SV-wave scattering by nonisothermal inclusions of finite thickness. The inclusions, which are aligned and sparsely embedded in an isotropic medium, induce an initial static stress field (acoustoelasticity) and a nonlinear dependence of the velocities on this stress. Moreover, we describe the anelastic properties as a function of frequency by incorporating the displacement discontinuities across the inclusion into the representation theorem by using the Foldy approximation. The response as a function of temperature is calculated for different incidence angles (anisotropy), and the results find that anelasticity increases with an increasing temperature difference between the inclusions and the background medium. The SV wave in the solid inclusions indicates a stronger sensitivity along the inclusion normal, and it is more affected than the P wave. The P- and SV-wave scattering by fluid-saturated inclusions behave in an opposite manner. This theory can be useful to evaluate the distribution of temperature from seismic attributes.

Seismic Tiles, a data format to facilitate analytics on seismic reflectors

Seismic Tiles is a new concept for representing seismic reflectors and attributes derived from seismic data, which makes it easy to automate seismic interpretation and analysis in general. Many subsurface problems can now be solved mostly automatically by computers, instead of mostly manually by humans, enabling the disruption of seismic interpretation at a scale similar to the disruption of road navigation facilitated by the closely analogous Google Map-style data structure. Some examples where this has been successfully implemented include screening for amplitude-variation-with-offset anomalies to identify prospects, automatic characterization of 4D effects according to rock physics modeling, and identifying faults and horizons verified by structural logic. Automation takes place not via machine learning but by direct and transparent application of well-known geophysical and geologic logic. This is achieved by automatically detecting all reflectors in the seismic data set and representing them by a set of tiles, which are small rectangular surfaces typically 200 m × 200 m in size. Tiles are calculated on a regular grid in lateral coordinates via a local 3D linear Radon transform. A set of tiles combine to describe a reflector surface, similar to how tiles in a bathroom combine to describe the walls and floor. Any attribute, from any seismic cube or other data sets, can be automatically extracted and assigned to the tiles and thus used in the logic. This logic can include the geometry of the tile itself and how it relates to tiles in the neighborhood, thus allowing easy implementation of any kind of interpretation logic that involves seismic reflectors. As the tile geometry and attributes are exposed to man and machine in a simple tabular data format, implementing logic using Seismic Tiles is easy, and much easier than applying logic directly on input in the form of raw seismic data.

Research on the characteristics of fluid escape near the seabed in Bohai Bay

AbstractSeabed fluid flow happens due to the complex interactionamong the accumulation, migration and escape of hydrocarbon and/or fluids. As shallow gas accumulations are anomalously developed in Bohai Bay, studying fluid flow features provides insights for understanding environment fluid in the Bohai Sea. Therefore, in our study, we aimed to make practical and effective use of single‐channel seismic reflection data to delineate and interpret subsurface fluid flow features by analysing the variance, sweetness, instantaneous frequency, and root mean square amplitude of seismic attributes. Four types of submarine fluid escape features, including pockmarks, mounds, fluid vents, dimmed or chaotic seafloor reflections, and two shallow fluid migration pathways (gas chimneys and faults) are distinguished on the single‐channel seismic reflection profiles. Gas chimneys are divided into strict columnar and irregular columnar chimneys, fluid migration along faults is critically dependent on the activity of faults. Enhanced reflections and bright spots may be good indicators of lateral fluid migration and accumulation of fluids in porous formations. We propose a shallow fluid flow model in the study area that illustrates the dynamic fluid migration process in terms of possible fluid sources, shallow migration pathways, and submarine fluid escape features.

Seismic stratigraphy and attenuation of gas-hydrate zones within Hikurangi and Gondwana margins, eastern New Zealand

SUMMARY Gas hydrates that occur on many continental margins have received global attention. In reflection seismic imaging, the bottom-simulating reflector (BSR) is a common indicator of gas hydrates. However, it is difficult to identify gas hydrates and quantify their amounts through the BSR alone. For gas-hydrate characterization, it is therefore useful to measure seismic stratigraphic and attenuation attributes. Short-scale patterns of layering that contain information about the amount and mechanism of gas hydrates can be identified through stratigraphic and attenuation attributes. We measure the complete time-variant spectra by using sparse strongest peaks, and the spectral differences at different times through attenuation parameters Q–1 and γ. The traditional Q–1 is associated with the attenuation of the frequency-dependent part of wavefield, and the γ characterizes the frequency-independent attenuation. The measurement approach is straightforward and requires no sophisticated inverse algorithm and is applied to surface seismic data acquired over the Hikurangi and Gondwana margins, eastern New Zealand. High-quality spectral and attenuation images are obtained. Spectral attributes correlate with BSRs and large positive Q–1 and negative γ-values are below and above the BSRs, which are interpreted as being related to free-gas and gas-hydrate accumulations. These results will aid the quantification of gas hydrates and the assessment of their roles as an energy resource, as a potential geological hazard, and in climate change and ocean warming.

Evidencing subtle faults in deep seismic reflection profiles: Data pre-conditioning and seismic attribute analysis of the legacy CROP-04 profile

Legacy seismic reflection data constitute infrastructure of tremendous value for basic research. This is especially relevant in seismically hazardous areas, as such datasets can significantly contribute to the seismotectonic characterization of the region. The quality of the data and the resulting image can be effectively improved by using modern tools, such as pre-conditioning techniques and seismic attributes. The latter are extensively used by the hydrocarbon exploration industry, but are still only poorly applied to the study of active faults. Pre-conditioning filters are effective in removing random noise, which hampers the detection of subtle geologic structures (i.e., normal faults). In this study, a workflow including pre-conditioning and extraction of seismic attributes is used to improve the quality of the CROP-04 deep seismic reflection profile. CROP-04 was acquired in the 1980s across the Southern Apennines mountain range, one of the most hazardous seismically active regions in Italy. The results show the capacity of this method to extract, from low-resolution legacy data, subtle seismic fabrics that correspond to a dense network of fault sets. These seismic signatures and the enhanced discontinuities disrupting the reflections, which were invisible in the original data, correlate well with the main regional normal faults outcropping at the surface. Moreover, the data reveal higher structural complexity, due to many secondary synthetic and antithetic structures, knowledge of which is useful in modeling of the local and regional distribution of the deformation and potentially in guiding future field mapping of active faults. This proposed approach and workflow can be extended to seismotectonic studies of other high-hazard regions worldwide, where seismic reflection data are available.

Modelling of geomorphological features of fluvial systems in Eromanga Basin and North Sea using 3D seismic attributes

Modelling of geomorphological features of fluvial systems in Eromanga Basin and North Sea using 3D seismic attributes

Integrating seismic attributes and rock physics for delineating Pliocene reservoir in Disouq field, Nile Delta, Egypt

The Nile Delta is known as the most prolific gas province in North Africa. Gas exploration and production were the main targets over decades in the Nile Delta. The integration of petrophysical analysis, seismic interpretation, and seismic attributes could indicate the presence of gas in Disouq field, which is located in Nile Delta, Egypt. Well log data showed that the reservoir is classified into three zones (A, B, and C). These zones are characterized by low gamma ray (around 35%), with relatively high porosity (around 29%) and low water saturation (36–38%). Structural and stratigraphic interpretation was conducted in order to investigate the extension and geometry of the reservoir, in addition to the faults affecting the study area. The reservoir is represented by a channel of Pliocene age, characterized by the low impedance sand saturated by gas. Near and far offset data show the gas effect. There is an increase of the negative amplitude from near to far offset. The amplitude map and structural contours match well and show good conformance. Seismic attributes including surface attributes and “Red Green Blue/colour blending” help in delineating the extent and geometry of the reservoir, in addition to the possible prospect. Finally, the volume of original gas in place was estimated what showed that the study area is of great economic interest.

Automatic gas chimney detection from 3D seismic reflection data using a single amplitude attribute

Automatic gas chimney classification methods use selected seismic attributes as training inputs for machine-learning algorithms. Contrary to the common belief that a single attribute is insufficient, the scope of this paper is to introduce an algorithm for gas chimney detection and extraction based on the amplitude attribute. The method is applied to a 3D seismic cube and involves the use of a feature extraction process (maximum overlap discrete wavelet transform - MODWT), before applying a standard machine-learning algorithm (bidirectional long short-term memory - BiLSTM). The MODWT is used to extract four unique features of each trace, which are then used with the original trace for training. The idea is to introduce a new feature-extraction step that enhances the accuracy of a single attribute. When tested on the F3 Block seismic cube, the integration of this process with the BiLSTM deep-learning algorithm achieved a balanced accuracy of 93% in gas chimneys identification. Results calibration by an experienced seismic interpreter also show a good match between machine-learning algorithm and manual interpretation. Therefore, our results successfully demonstrate that gas chimneys can be automatically detected and extracted from 3D seismic data using a single attribute. The workflow presented here is applicable worldwide for studies of different economic and environmental purposes (e.g., natural resources of petroleum exploration and gas hydrate, subsurface storage characterization and geohazards identification).

Multi-scale geophysical characterization of microporous carbonate reservoirs utilizing machine learning techniques: An analog case study from an upper Jubaila formation outcrop, Saudi Arabia

Microporosity hosts a significant portion of the total hydrocarbon volume in the Middle Eastern carbonate reservoirs. An improved understanding of microcrystals' morphology that hosts micropores, their impact on reservoir properties, and their spatial distributions will contribute to a more accurate reservoir quality prediction. This study proposes a methodology that utilizes machine learning approaches for microporosity characterization and integrates a multi-scale dataset ranging from micrometer-scale SEM images to meter-scale seismic attributes to build reservoir porosity models. We tested the methodology on an outcrop in Riyadh, Saudi Arabia, which exposes the upper part of the Jubaila Formation equivalent to the lower part of the Arab-D reservoir. We acquired a 35 m-long core and a 600 m-long 2D seismic line behind the outcrop. Laboratory-scale petrophysical measurements, including porosity, permeability, acoustic velocity, bulk and grain density, and x-ray diffraction, were performed over 106 horizontal core plugs drilled from the core. We quantitatively characterized the morphology and microtextures of micrite crystals using SEM images by performing Random Forest classifications trained on SEM image features. We performed unsupervised classification using Self-Organizing Map (SOM) to all lab-measured properties for data clustering. We investigated potential correlations between SOM clusters with micrite morphology, which resulted in a predictable relationship following the granularity and sphericity of microcrystals with porosity, permeability, and acoustic velocity. It was possible to represent multiple lithofacies with a single log-linear porosity-permeability relationship and a single value of equivalent differential effective medium (DEM) aspect ratio in velocity-porosity space. The inter-relationship between micrite morphology and microporosity in the well was then propagated to reservoir-grid scale using inverse differential effective medium (DEM) of acoustic impedance from inverted seismic data. The methodology developed in this study thus provides a practical way to integrate key sub-grid scale micro-and macro-heterogeneities into reservoir scale property models.

Quantitative Prediction of Braided Sandbodies Based on Probability Fusion and Multi-Point Geostatistics

Predicting the spatial distribution of braided fluvial facies reservoirs is of paramount significance for oil and gas exploration and development. Given that seismic materials enjoy an advantage in dense spatial sampling, many methods have been proposed to predict the reservoir distribution based on different seismic attributes. Nevertheless, different seismic attributes have different sensitivities to the reservoirs, and informational redundancy between them makes it difficult to combine them effectively. Regarding reservoir modeling, multi-point geostatistics represents the distribution characteristics of the braided fluvial facies reservoirs effectively. Despite this, it is very difficult to build high-quality training images. Hence, this paper proposes a three-step method of predicting braided fluvial facies reservoirs based on probability fusion and multi-point geostatistics. Firstly, similar statistical data of modern sedimentation and field paleo-outcrops were processed under the guidance of the sedimentation pattern to construct reservoir training images suitable for the target stratum in the research area. Secondly, each linear combination of selected seismic attributes was demarcated to calculate the principal component value and work out the elementary conditional probability. Lastly, the PR probability integration approach was employed to combine all conditional probabilities and calculate the joint probability. Then the joint probability was combined with training images to build a reservoir distribution model through multi-point geostatistics. We illustrated the detailed workflow of our new method by applying it to a braided fluvial reservoir modeling case in the Bohai Bay Basin, East China. The new method reduced the error of prediction results by 32% and 46% respectively, and the error of water content by 36.5% and 60.3%. This method is a potentially effective technique to predict and characterize the reservoir spatial distribution and modeling in other oil fields with the same geological background.

Combination of seismic attributes using clustering and neural networks to identify environments with sandstone-type uranium mineralization

Combination of seismic attributes using clustering and neural networks to identify environments with sandstone-type uranium mineralization

Porosity and water saturation predicting beyond boreholes from electromagnetic sounding and core sample data: Soultz-sous-Forêts (France) case study

Porosity predicting beyond boreholes is carried out by means of neural network taught by correspondence of the mercury porosity values determined from core samples and electrical resistivity estimated by magnetotelluric sounding in its vicinity. Testing of the artificial neuronet porosity prediction beyond boreholes indicated that the total relative error of the forecasts is about 7% that is slightly less than the prediction errors (∼10%) obtained in laboratory using seismic attributes. Comparative analysis of the prediction accuracy in boreholes with porosity values determined by direct and indirect ways showed that building of the porosity model of the section using simultaneously porosity data determined by direct and indirect techniques is inexpedient. At the same time, accuracy of electromagnetic porosity predicting using neural networks calibrated with data measured by the mercury porosimetry turns to be about three times higher than the accuracy of porosity estimation by indirect methods (for example, neutron gamma logging). A new approach to building of a water saturation section is proposed. It is based on a comparison of total open porosity determined using porosity values estimated by mercury porosimetry in a reference well and fluid volume fraction determined from electromagnetic sounding data. The latter is estimated taking into account the dependence of the brine electrical resistivity on the temperature determined using electromagnetic geothermometry. The errors of predicting water saturation beyond boreholes are around 20–22%, which are comparable with its estimation from seismic data in laboratory conditions.

Seismic identification of carbonate reservoir sweet spots using unsupervised machine learning: A case study from Brazil deep water Aptian pre-salt data

The pre-salt high-yield carbonate reservoirs of Brazil are responsible for about 75% of national oil production. Still, seventeen years since its discovery, detailed seismic characterization of these reservoirs remains challenging, with producing and nonproducing lithologies often exhibiting the same amplitude response. Those reservoirs are complex, composed of high productivity lacustrine carbonates (stromatolites and grainstones), mixed with non-producing mudstones, clays, and igneous intrusions. To address this heterogeneity, we evaluate unsupervised machine learning clustering techniques including Self Organizing Maps (SOM), Generative Topographic Maps (GTM), and K-means and apply them to a suite of instantaneous, geometric, interval velocity, and AVO seismic attributes. We use both static and dynamic reservoir data from 13 wells at the Mero field that penetrated lacustrine carbonates from Barra Velha formation to validate the accuracy of our unsupervised learning models. These wells are excellent blind tests as they were not used for any algorithm training. In this pioneer work, the first done using machine learning in the Mero field, we find that SOM, GTM, and K-means models provide clusters that can be linked to sweet spots associated with fracturing, karstification, and hydrothermal alteration. Additionally, clusters related to sill intrusions, platform facies and distal fine grained non-reservoirs were identified. The results can be extrapolated to the adjacent Central Libra Area (still in exploration phase), where we have identified the same clusters of the sweet spots reservoirs in Mero Field and, thus, they can be considered as future targets. For accuracy assessment, the same attribute combination SOM clustering was performed in another Petrobras’ pre-salt confidential dataset and the results were satisfactory. So, for exploration projects optimization purposes, the new proposed workflows can also be applied to other areas with geological similarity to the Mero Field.