Logistic nonlinear convergence rate for subtle difference identification in seismic data

ABSTRACTSeismic detection of faults, dykes, potholes and iron‐rich ultramafic pegmatitic bodies is of great importance to the platinum mining industry, as these structures affect safety and efficiency. The application of conventional seismic attributes (such as instantaneous amplitude, phase and frequency) in the hard‐rock environment is more challenging than in soft‐rock settings because the geology is often complex, reflections disrupted and the seismic energy strongly scattered. We have developed new seismic attributes that sharpen seismic reflections, enabling additional structural information to be extracted from hard‐rock seismic data. The symmetry attribute is based on the invariance of an object with respect to transformations such as rotation and reflection; it is independent of the trace reflection amplitude, and hence a better indicator of the lateral continuity of thin and weak reflections. The reflection‐continuity detector attribute is based on the Hilbert transform; it enhances the visibility of the peaks and troughs of the seismic traces, and hence the continuity of weak reflections. We demonstrate the effectiveness of these new seismic attributes by applying them to a legacy 3D seismic data set from the Bushveld Complex in South Africa. These seismic attributes show good detection of deep‐seated thin (∼1.5 m thick) platinum ore bodies and their associated complex geological structures (faults, dykes, potholes and iron‐rich ultramafic pegmatites). They provide a fast, cost‐effective and efficient interpretation tool that, when coupled with horizon‐based seismic attributes, can reveal structures not seen in conventional interpretations.

Sand-lithofacies distributions are a vital component of reservoir characterisation and the generation of successful reservoir models, particularly in thin reservoirs thickness. The integration between seismic attributes and conceptual geologic models is important to identify the sand distributions and reservoir geometries. The aim of this work is illustrating the uses of seismic data in revealing the structural and stratigraphic elements of two main reservoir units. It is important to identify and list the attributes, which can be involved in analysing the reservoirs and their degree of involvement in the characterisation of the hydrocarbon accumulation zones. The study was carried out for the reservoirs: Abu Roash C and Abu Roach E in Sitra oil and gas concession, which is located in the northern part of the Egyptian Western Desert. Four seismic attributes are used: 1- Coherence (Variance) 2- 3D Curvature attributes were used for resolving mainly the structural features and also the stratigraphic features 3- Instantaneous phase 4- Instantaneous frequency attributes were used mainly for resolving the stratigraphic features. Through the integration of selected seismic attributes with post-stack seismic data, reservoir distributions and architectures are accurately delineated for the Upper Cretaceous reservoir (Abu Roash ‘C’ and ‘E’). Five wells were used to match the results of the seismic attributes and establish the regional trends of reservoir parameters. The integration of wells data, conventional seismic interpretation and extracted the relevant seismic attributes help in reaching a robust geologic model for the reservoirs of interest. Stratigraphic and structural interpretation objectives based on extracting the seismic attributes for reservoir characterisation become more dominant, as they go deeper in seismic data, to derive more subsurface geologic data than traditional interpretation. 3D structure model of the concern area was established by interpreting the available seismic data as both reservoirs implicated by two fault trends: NW-SE and ENE-WSW. By the end a 3D geological model was defined based on the results of the conventional interpretation of the seismic and seismic attributes to reflect all structural and stratigraphic elements of the two reservoir units.

This work has the purpose to understand the relation that are between seismic attributes extracted from 2D high-resolution seismic data and the sediment surveyed. To accomplish this goal, we performed a research using 2D high-resolution seismic data acquired on the Continental Shelf of the Antarctic Peninsula, nearby the South Shetland Islands, and sediment samples acquired with gravity corer overlapping each seismic line surveyed. The geologic samples were used as control samples to make the binding between seismic traces - sediment cores and to fulfill the sediment classification software training outline. One computational script was written to extract the seismic attributes from the seismic data and in sequence the attributes were statistically analyzed. The attributes were used as variables during the statistical process and they were: Amplitude (AMP), Instantaneous Phase (PHI), Instantaneous Frequency (FREQ), Envelope (ENV), Time Derivative of the Envelope (DTENV), Second Derivative of the Envelope (DT2ENV) and Acceleration of Phase (ACPHI). The statistical analyses were: Principal Components (PCA), Dendrograms and K-Means Classification. Four kinds of seismic attributes groups were coherently formed on statistical analyses. In theory basis, the PHI shouldn't be grouped with any other attribute because of its physical behavior completely distinct from the others attributes here studied. The sets of attributes formed that presented ACPHI were absolutely coherent, in a way that all others attributes grouped within ACPHI present the same physical and spatial behavior such as DTENV, AMP, ENV and FREQ, this is due to the absorption of both the amplitude and frequency of the wave propagated through the medium. Therefore, it is believed that both more coherent sets of attributes are AMP and ENV identified by all the statistical analyses and DTENV and ACPHI formed by the PCA and Dendrograms analyses, and are both well supported by the other analyses that assured an intrinsic physical and spatial relation between the grouped attributes. The subsequent step on the Semi-Automatic Sediment Classification Process using 2D high-resolution seismic data were to prepare the set of attributes that grouped coherently, to identify the range of values of each attribute grouped that correspond to the location of each core sample, locate on the reflectors in the seismic section where those ranges of values from the set of attributes are found at the same place and finally suppose that on these regions identified by the script we have the same kind of sediment found on the geological samples.

As petroleum resources run out, subtle stratigraphic reservoirs have become an important source of future discoveries. Continental tertiary deposits of fluvio-deltaic origin are known to produce such complex reservoirs and pose a challenge for characterization. We propose a novel bio-integrated framework called multi-modal machine learning fusion to tackle this challenge. The approach bases reservoir characterization on biological interpretation of sensory data. Using seismic attributes as varied modal inputs and rock physics relationships as input data, we perform multi-modal machine learning fusion to predict the gamma ray spatial distribution. Before processing however, the seismic data is filtered with a dip steered filtering algorithm to remove random noise. The resulting volume is then used in computing the seismic attributes. Effective seismic attributes utilized were, instantaneous amplitude, instantaneous frequency, spectral decomposition, seismic inversion and similarity. Results from a case study in the Bohai Bay Basin in Eastern China demonstrate that multi-modal machine learning fusion can predict the spatial distribution of lithology using gamma ray logs. The conclusions from this study can be applied in analogous reservoirs in similar petroleum provinces in other regions of the world and can also be applied to other geophysical interpretational methods.

Ore-controlling structures of sandstone-hosted uranium deposit in the southwestern Ordos Basin: Revealed from seismic and gravity data

The Tuaheni Landslide Complex, located on the upper slope of the northern Hikurangi Margin in New Zealand, is a unique place to research on slow slip creep-like deformation and seabed failure, as well as their possible relationship with the presence of gas hydrates, cold seeps, and fluid migration. Based on the visual interpretation of seismic data, it is sometimes very difficult to identify various subsurface structures and tectonic features. We study certain seismic attributes, namely the reflection strength, instantaneous frequency, instantaneous phase, and the Hilbert transform, in the Tuaheni Landslide Complex and observe that these attributes play a very important role in identifying and interpreting various subsurface geological features and bed boundaries that are not clearly visible in the seismic sections. In general, these seismic attributes are studied to identify hydrocarbons such as oil and gas. However, in this present study these seismic attributes nicely illustrate the fluid migration pathways, the decollement of the sediment slide, the base of the debris flow, the base of the deformed sediment and gas migration, etc., along two perpendicular seismic profiles crossing the Site U1517 of IODP Expedition 372. The instantaneous phase and Hilbert transform attribute depict the bed boundaries and discontinuities, whereas the reflection strength and instantaneous frequency attributes characterize the various strata in terms of whether they are associated with fluid at their bases. The possible role of tectonic activity and seafloor slope failure due to gas hydrate dissociation and vice versa is clearly visible through fluid-filled weak zones in the seismic attribute volumes. Gas hydrates are dissociating and BSRs are abruptly pinching out towards the seafloor due to the movement of hot fluid and free gas, enhancing seafloor sliding and local tectonic activities together.

Recently, ideas of color blending have brought the enlightenment for subsurface geobody imaging in petroleum engineering. In this paper, we present this approach of CMY color blending and its application in subsurface geobody characterization by using seismic attributes data. The first step is to calculate three types of seismic attributes based on the Hilbert transform algorithm, including envelop, instantaneous phase, and instantaneous frequency. Then scale the three attributes and combine them together using CMY color model in three-dimensional environment, with each attribute corresponding to one primary color channel. Adjust the scale and offset for each color component and then mix them optimally to create one color-blended volume. The blended volume in CMY mode has plenty of geological information coming from the three input attributes, resulting in high resolution and accurate image for subsurface geobodies. Applications show good performances in buried channels, caves, and faults imaging. Based on the blended slice, the geological targets can be easily but accurately interpreted and depicted.

In order to accurately identify the distribution of the karst water conduction channel after water inrush, the ground penetrating radar (GPR) is used for fine detection of development of karst water conduction in front of the heading surface after water inrush, and the instantaneous amplitude, instantaneous phase and instantaneous frequency are extracted from the radar signal by Hilbert transform. Then, the three instantaneous properties are combined for multi-parameter comprehensive analysis of the karst water conduction channel. As shown in the study results, the instantaneous amplitude can better show the boundary characteristics and morphological size of the karst water conduction channel. The instantaneous phase can clearly show the heterogeneous characteristics inside the karst water conduction channel. The instantaneous frequency can reflect the water richness of the karst water conduction channel from the angle of energy decay. The combination of the three instantaneous attributes can avoid the explanatory bias due to the analysis of single waveform characteristics, and effectively improve the resolution and accuracy of the detection of the karst water conduction channel.

A robust algorithm for the simultaneous estimation of dip, azimuth and curvature in the depth migrated domain is developed. The algorithm makes use of instantaneous‐attribute beamforming in small windows aligned with locally coherent migrated events. This is the following two‐step procedure: inline and cross‐line linear beam stacks to compute the two local dips at image points; 360‐degree azimuthal parabolic beamforming to estimate azimuthally dependent curvatures with respect to the plane tangent to the instantaneous‐phase surface at the image point. Amplitude and phase content of migrated traces are processed independently using complex trace analysis in order to facilitate this procedure. The instantaneous phase is of central importance since it describes the lateral continuity of events and defines instantaneous dips corresponding to instantaneous apparent phase velocity and group velocity. This attribute is used to compute the instantaneous frequency that defines the beam wavelet at the peak of waveform envelope. The instantaneous amplitude or envelope represents an input to model‐based wavefield (Born‐Kirchhoff) de‐remigration of picked events, useful for dip QC. The QC analysis involves thresholding of modeled migrated amplitudes predicted by Kirchhoff remigration of the Born demigrated wavefield. This also includes simple thresholding of the instantaneous frequency. The analysis is shown to be capable of eliminating undesired events. The algorithm was tested using a 3‐D marine streamer dataset with high noise. The success with field migrated data demonstrates that the algorithm is particularly useful for slope‐limited tomographic velocity updates and structural interpretation when the signal‐to‐noise ratio (SNR) is low.

The paper is devoted to predicting petrophysical parameters of productive series of Kurovdagh field by use of attribute analysis of seismic data to define the direction of reconnaissance works in this field. The paper considers geographical position of the study area, its cover by geological and geophysical studies and underlines, in particular, the importance of 3D seismic data acquisition for more detailed study of Kurovdagh structure. Lithological and stratigraphic characteristics of the section is also given in the paper with detailed description of Productive Series deposits. In addition, tectonics of the area is considered in more detail. It is noted that the area is attributed to the Low Kur basin — the compound of the large tectonic unit. The tectonic zone of the south-eastern Shirvan is embracing four anticlinal zones: Pirsaat-Khamamdagh; Kharami—Mishovdag—Kalmas—Khydirly—Aghaevir-Byandovan; Kursangya; Padar—Kurovdagh—Karabaghly—Babazanan—Duzdagh-Neftchala. The latter anticlinal zone is characterized by a significant length. In the north-western part between the folds of Padar and Karabaghly the brachyanticline of Kurovdagh is located. In the north it has the border with M. Kharami uplift, in the north-east with Mishovdagh fold, it borders with Kursyanga anticline in the south-east and in the south-west with wide Salyan trough. In the Near-Kur depression the existence of two tectonic stripes has been established. One of them embraces the south-eastern Shirvan, the other covers the eastern Mughan and the western portion of Salyan steppe. The detailed description of fold setting is given on the basis of 3D seismic survey data. It has been indicated that the results of 3D data interpretation made it possible to study in more detail and make changes in the scheme of faults location accepted earlier. The other problem considered in the paper is the oil-and-gas presence in Kurovdagh field, which is related to the Absheron stage of Pleistocene, Akchagyl stage and Productive series (horizons of PS01—PS20) of Neogene, with lithology represented by sandy-clay rocks with various degrees of calcareousness. The structure of each of indicated horizons is rather complicated and variable in lateral. The most complicated of them is the Middle Absheron sub-stage, with identified 11 oil-bearing layers. Study results are given in the end of the paper. For prediction purposes within the study area we have prepared normalized curves of relative parameter of SP-ASP, gamma-log — dGR and resistivity by use of well logging data. The analysis of dependence of seismic attributes on petrophysical parameters within target interval, identified the low information bearing ability of SP method and gamma-log across the study area and established a good correlation between resistivity curve and instantaneous amplitudes, frequencies and dip angles. The clay cubes have been designed. To outline productive layers, we have applied multidimensional filters with cut offs for reservoir and as a result we have acquired a cube for supposed distribution of productive layers. It is emphasized that the conducted studies led to the conclusion that due to the complexity and interference nature of the observed wave pattern in some parts of the Kurovdagh structure, it was not possible to reliably convert the attributes of the seismic wave field into petrophysical parameters.

This paper proposes a method for quantitative integration of seismic (elastic) anisotropy attributes with reservoir performance data as an aid in characterization of systems of natural fractures in hydrocarbon reservoirs. This method is demonstrated through application to history matching of reservoir performance using synthetic test cases. Discrete Feature Network (DFN) modeling (Dershowitz et al.1) is a powerful tool for developing field-wide stochastic realizations of fracture networks in petroleum reservoirs. Such models are typically well conditioned in the vicinity of the wellbore through incorporation of core data, borehole imagery, and pressure transient data. Model uncertainty generally increases with distance from the borehole. Threedimensional seismic data provides uncalibrated information throughout the inter-well space. Some elementary seismic attributes such as horizon curvature and impedance anomalies have been used to guide estimates of fracture trend and intensity (P32) (Dershowitz and Herda2) in DFN modeling through geostatistical calibration with borehole and other data. However, these attributes often provide only weak statistical correlation with fracture system characteristics. The presence of a system of natural fractures in a reservoir induces elastic anisotropy that can be observed in seismic data. Elastic attributes such as azimuthally dependent normal moveout velocity (ANMO), reflection amplitude versus azimuth (AVAZ), and shear wave bi-refringence can be inverted from 3D seismic data. Anisotropic elastic theory provides physical relationships among these attributes and fracture system properties such as trend and intensity. Effective elastic media models allow forward modeling of elastic properties for fractured media. A technique has been developed in which both reservoir performance data and seismic anisotropic attributes are used in an objective function for gradient-based optimization of selected fracture system parameters. The proposed integration method involves parallel workflows for effective elastic and effective permeability media modeling from an initial DFN estimate of the fracture system. The objective function is minimized through systematic updates of selected fracture population parameters. Synthetic data cases show that 3D seismic attributes contribute significantly to reduction of ambiguity in estimates of fracture system characteristics in the inter-well rock mass. The method will benefit enhanced oil recovery (EOR) program planning and management, optimization of horizontal well trajectory and completion design, and borehole stability studies.

Earthquake is one of the main factors causing structural disasters in current buildings. Under earthquake action, adjacent building structures are generally in different vibration stages, and collisions may occur in the structures, which may cause serious damage to the structure. In order to prevent certain earthquakes from damaging the designed buildings, this article mainly introduced the design and analysis of a numerical algorithm for seismic nonlinear structural dynamic response based on the HVSR algorithm. This article evenly divided the acceleration response time series into 15 time periods and then selected the position corresponding to the peak point of instantaneous amplitude within each period as the selected data point position. The same seismic load can be applied at the bottom of the established nonlinear model to extract the dynamic response data of the top layer of the structure, and then, the instantaneous amplitude and corresponding instantaneous phases and frequencies of the main components of the structural dynamic response can be extracted through the time-varying filter and Hilbert transform based on the discrete analytic mode decomposition. Under the influence of these four ground motions, the collision force within the range of 0-50 kN accounted for over 87% of the total number of collisions. In the comparison results of collision response to peak displacement, the four ground motions all led to structural collision, and the collision inhibited the positive peak displacement response of node 1512. Compared with noncollision, the peak displacement was reduced by 27.273, 33.675, 27.727, and 37.248%, respectively. The peak displacement of 1512 nodes was suppressed and reduced by 18.856%. The results indicate that the HVSR algorithm can obtain the instantaneous characteristic parameters of nonlinear structural dynamic response and achieve model correction.

The same geophysical and geologic data have been interpreted for both unconventional shale-gas and conventional sandstone reservoirs simultaneously by framing and using a customized workflow. Customization was needed because of property differences between the two types of resources. Unconventional shale reservoirs differ significantly from conventional reservoirs in depositional environment, diagenesis, physical properties, and production techniques. Magnitudes and patterns of reservoir properties (e.g., velocity, resistivity, radioactivity, porosity, permeability, etc.) of shales differ significantly from those of sandstone/carbonate reservoirs. In a conventional play, we generally search for intervals with relatively low gamma, high resistivity, and low neutron porosity, whereas in shale gas we search for intervals with very high gamma, relatively higher resistivity than lean shales, low velocity, and very high neutron porosity. Production from shales is largely due to horizontal drilling, hydrofracturing, and stimulations. The shale-gas reservoir properties are governed mainly by organic richness and its thermal maturation. Organic material has significant impact on the elastic/petrophysical properties, which are manifested in log and seismic responses. Thus, identification and mapping of shale-gas reservoirs is based mainly on mapping of total organic carbon- (TOC) rich zones. The customized workflow was developed for identification of shale-dominant intervals and organic-rich-zones within them by integrating log and seismic data. Shales generally show low-amplitude signals, and internal reflection configurations are not easily identified. Seismic data conditioning and seismic attributes, e.g., phase and perigram, were applied for internal geometry mapping. Once shale-gas markers were identified, mapping of shale boundary, depth, thickness, and areal extent was done similarly to conventional plays. Sweet spots were identified by generating and analyzing log-property volumes like impedance, sonic, resistivity, and neutron porosity. Natural fractures, which are helpful in designing well trajectory, were mapped through geometric seismic attributes like curvature and dip azimuth. This study has enhanced geologic understanding and, hence, mappability of shale-gas plays along with conventional plays. The customized workflow is applicable in all stages of shale-gas-play exploration and development.

Accurate three-dimensional acoustic impedance modeling in offshore clastic reservoirs remains a significant challenge due to sparse well control and the highly nonlinear relationship between seismic attributes and subsurface elastic properties. This study introduces an integrated, physics-guided machine learning (ML) workflow that combines rock-physics-driven pseudo-well generation with neural networks to directly map seismic attributes to acoustic impedance under data-limited conditions. A soft-sand rock physics workflow was applied, in which grain moduli were determined using the Voigt-Reuss-Hill average. The dry rock frame was modeled at critical porosity by Hertz-Mindlin contact theory and then interpolated toward zero porosity using the Modified Hashin-Shtrikman lower bound. Gassmann fluid substitution was subsequently performed. Using this approach, 45 pseudo-wells were generated and conditioned through lithofacies classification and spatial statistics, mitigating the risk of overfitting associated with the three available real wells. Six seismic attributes-envelope, RMS amplitude, instantaneous phase, instantaneous frequency, quadrature trace, and sweetness-were selected as predictors. Two neural architectures, a multi-layer feedforward network (MLFN) and a radial basis function network (RBFN), were trained and benchmarked using a leave-one-well-out cross-validation scheme. The MLFN achieved higher predictive accuracy (CC = 0.87, NRMSE = 0.493) compared to the RBFN (CC = 0.79, NRMSE = 0.613), which may reflect its greater capacity to model broader hierarchical relationships between seismic attributes and acoustic impedance. The resulting impedance volume delineates laterally coherent high-impedance sandstone units and low-impedance porous intervals consistent with geological interpretation. These results suggest that integrating physics-guided pseudo-well augmentation with feed-forward neural networks offers a practical and computationally efficient approach for acoustic impedance inversion in data-limited offshore settings. Future work may explore validation across diverse geological settings to assess the robustness and transferability of the proposed methodology. This study provides a basis for hybrid and uncertainty-aware inversion frameworks that may help address complexities in heterogeneous reservoir systems, highlighting the importance of reproducible and widely applicable data-driven seismic inversion methods under sparse well control.

Seismic facies classification plays a critical role in characterizing & delineating the various features present in the reservoirs. It aims at determining the number of facies & their description for the available seismic data. During the past few decades, seismic attributes have been widely used for the task of seismic facies identification. It helps geologists to determine different lithological and stratigraphical changes in the reservoir. With the increase in the seismic data volume & attributes, it becomes difficult for the interpreters to examine each seismic line. One of the solutions given to this problem was to use some computer-assisted methods such as k-means, self-organizing map, generative topographic map and artificial neural network for analyzing the seismic data. Even though these computer-assisted methods performed well but due to the size of the 3-D seismic data the overall classification process becomes very protracted. In this paper, we introduce a parallel framework for unsupervised classification of the seismic facies. The method begins by calculating four different seismic attributes. Spark & Tensorflow based implementation of unsupervised facies classification algorithms are then used to identify the seismic facies based on the 4-D input attributes data. Further, the comparison of results (in terms of execution time & error) of Spark & Tensorflow based algorithms with already existing approach show that the proposed approach provides results much faster than previously existing MPI based approach.