Search

Quantitative prediction of reservoir properties (e.g., gas saturation, porosity, and shale content) of tight reservoirs is of great significance for resource evaluation and well placements. However, the complex pore structures, poor pore connectivity, and uneven fluid distribution of tight sandstone reservoirs make the correlation between reservoir parameters and elastic properties more complicated and thus pose a major challenge in seismic reservoir characterization. We have developed a partially connected double porosity model to calculate elastic properties by considering the pore structure and connectivity, and to analyze these factors' influences on the elastic behaviors of tight sandstone reservoirs. The modeling results suggest that the bulk modulus is likely to be affected by the pore connectivity coefficient, while the shear modulus is sensitive to the volumetric fraction of stiff pores. By comparing the model predictions with the acoustic measurements of the dry and saturated quartz sandstone samples, the volumetric fraction of stiff pores and the pore connectivity coefficient can be determined. Based on the calibrated model, we have constructed a 3D rock physics template that accounts for the reservoir properties' impacts on the P-wave impedance, S-wave impedance, and density. The template combined with Bayesian inverse theory is used to quantify gas saturation, porosity, clay content, and their corresponding uncertainties from elastic parameters. The application of well-log and seismic data demonstrates that our 3D rock physics template-based probabilistic inversion approach performs well in predicting the spatial distribution of high-quality tight sandstone reservoirs in southwestern China. [ABSTRACT FROM AUTHOR]

Seismic inversion holds significant potential for providing crucial lithostratigraphic information in hydrocarbon reservoir characterisation and in identifying new traps. However, one of the major challenges in achieving reliable reservoir models in Algeria stems from the inherent uncertainties associated with seismic inversion algorithms and the non-linear relationship between petrophysical measurements. Due to their usefulness, several Artificial Neural Network algorithms have been developed and employed for seismic inversion and reservoir characterisation in the last few years. Nevertheless, only few researchers have addressed this issue in terms of optimisation of Multilayer Feed-Forward Neural Network (MLFN) architecture. In this case study, the use of an MLFN to address these challenges is proposed. The primary contribution of this research lies in the optimisation of the MLFN architecture based on trial and error procedures. The goal is to ensure that the computational demands are manageable within the constraints of available computing resources and that the process is time-efficient for geo-modellers. This practical approach is particularly valuable when applied at the reservoir scale. MLFN supervised training is conducted using logging data, where measured log curves serve as inputs, and core porosity, obtained from laboratory analysis, serves as target output. Moreover, coloured inversion is employed to generate a 3D seismic acoustic impedance cube, which, in turn, serves as input for a model-based inversion method designed to calculate porosity volume using the trained network. Furthermore, the usefulness of the resulting density cube is demonstrated through the correlation with density logs and core density values at wells 1, 2, and 3. Thence, the obtained correlation ranges validate the reliability of the obtained porosity volume in enhancing the characterisation of the targeted reservoir within the Algerian Saharan field. [ABSTRACT FROM AUTHOR]

On 14 November 2021, a doublet earthquake, each event of which had an Mw of 6.1, struck near Fin in the Simply Folded Belt (SFB) in southern Iran. The first quake occurred at 12:07:04 UTC, followed by a second one just a minute and a half later. The SFB is known for its blind thrust faults, typically not associated with surface ruptures. These earthquakes are usually linked to the middle and lower layers of the sedimentary cover. Identifying the faults that trigger earthquakes in the region remains a significant challenge and is subject to high uncertainty. This study aims to identify and determine the fault(s) that may have caused the doublet earthquake. To achieve this goal, we utilized the DInSAR method using Sentinel-1 to detect deformation, followed by finite-fault inversion and magnetic interpretation to determine the location, geometry, and slip distribution of the fault(s). Bayesian probabilistic joint inversion was used to model the earthquake sources and derive the geometric parameters of potential fault planes. The study presents two potential fault solutions—one dipping to the north and the other to the south. Both solutions showed no significant difference in strike and fault location, suggesting a single fault. Based on the results of the seismic inversion, it appears that a north-dipping fault with a strike, dip, and rake of 257°, 74°, and 77°, respectively, is more consistent with the geological setting of the area. The fault plane has a width of roughly 3.6 km, a length of 13.4 km, and a depth of 5.6 km. Our results revealed maximum displacements along the radar line of sight reaching values of up to −360 mm in the ascending orbit, indicating an unknown fault with horizontal displacements at the surface ranging from −144 to 170 mm and maximum vertical displacements between −204 and 415 mm. Aeromagnetic data for Iran were utilized with an average flight-line spacing of 7.5 km. The middle of the data observation period was considered to apply the RTP filter, and the DRTP method was used. We calculated the gradient of the residual anomaly in the N-S direction due to the direction of the existing faults and folds. The gradient map identified the fault and potential extension of the observed anomalies related to a fault with an ENE-WSW strike, which could extend to the ~ E-W. We suggest that earthquakes occur in the sedimentary cover of the SFB where subsurface faulting is involved, with Hormuz salt acting as an important barrier to rupture. The multidisciplinary approach used in this study, including InSAR and magnetic data, underscores the importance of accurate fault characterization. These findings provide valuable insights into the seismic hazard of the area. [ABSTRACT FROM AUTHOR]

Nowadays, the most successful applications of full‐waveform inversion (FWI) involve marine seismic data under acoustic approximations. Elastic FWI of land seismic data is still challenging in theory and practice. Here, we propose a full dispersion spectrum inversion method and apply it to seismic data acquired in West Antarctica. Inspired by the conventional surface wave dispersion curve inversion method, we propose to invert the surface wave dispersion spectrum instead of the complicated waveforms. We compare the frequency‐velocity, frequency‐slowness, and frequency‐wavenumber spectra in terms of their ability to resolve dispersion modes and the feasibility of their adjoint updates and conclude that the frequency‐slowness spectrum is the best for our inversion objectives. We test four objective functions, subtraction, zero‐lag crosscorrelation, optimal transport, and the local‐crosscorrelation to quantify the spectrum mismatch and provide the corresponding adjoint source. We then theoretically analyze the convexity of the proposed objective functions and examine their convergence behavior using numerical examples. We also compare the proposed method with the classic FWI method and the traditional surface wave dispersion curve inversion method and discuss the strengths and weaknesses of each method. This technique is employed to evaluate the shallow velocity structures beneath a seismic array stationed in West Antarctica. Our proposed inversion scheme is also useful for more general applications such as imaging the shallow subsurface of the critical zones, like geothermal reservoirs, and CO2 storage sites. Plain Language Summary: Seismic full‐waveform inversion (FWI) is a cutting‐edge inversion method used for uncovering the Earth's subsurface structure. With the growing interest in clean energy and CO2 sequestration, exploring the subsurface in land is becoming crucial. However, there are only a few success stories of seismic FWI applied to land data mainly because of the complexity of the near‐surface and the increased nonlinearity of the problem. Here, we propose a full dispersion spectrum inversion method that seeks optimal velocity models in the subsurface by matching the seismic dispersion spectra. Dispersion spectra are the skeleton of seismic surface waves, which are simpler to quantify yet retain the key dispersion information of surface waves. It is generally easier for humans, as well as algorithms, to match simplified representations of the observed and simulated data, such as the dispersion spectrum, instead of the seismic waveforms themselves. The proposed method expands conventional 1D dispersion curve inversion to multiple dimensions and is accomplished under the framework of FWI. The proposed inversion method applies to seismic imaging applications in exploration and global seismology. Key Points: We propose a full dispersion spectrum inversion method for imaging the near‐surface and apply it to seismic array data collected in West AntarcticaThe surface wave frequency‐slowness spectrum is preferred over the frequency‐velocity and the frequency‐wavenumber spectra to measure the misfit in the proposed inversionThe optimal transport objective function is immune to cycle skipping but has a lower model resolution [ABSTRACT FROM AUTHOR]

Seismic inversion holds significant importance across various domains of geoscience and engineering, including the characterization of energy resource reservoirs, the assessment of polluted sites, and CO2 storage. It is a process of estimating rock properties from seismic data that is inherently uncertain, nonlinear, non‐unique, and highly challenging. Using multiple seismic attributes increases the size of the data, requiring considerable processing resources and time. However, deep learning can accurately fit quantities of nonlinear variables, making it an excellent method for predicting spatially distributed subsurface properties. We trained some multi‐output regression neural networks to carry out porosity inversion from seismic data. We initially computed a series of seismic attributes and generated the corresponding porosity using interpreted horizons, well logs, and seismic data. Subsequently, we proposed a technique to identify the most relevant seismic attributes for porosity inversion. Because our networks work as stochastic modeling entities, we created a weight‐averaging ensemble approach to build a strong model with the highest level of accuracy. We combined realizations from baseline entities, considering their respective performance levels. Using the statistics between these realizations and the robust model, we determined the degree of uncertainty associated with the outcome. We found an R2 of 0.993 and an MAE of 0.00112 in the F3 block offshore the Netherlands, proving the method's effectiveness. The mean porosity was 0.175193, compared to 0.175626 from a reference model, and the mean uncertainty was ±0.0008998. Plain Language Summary: Understanding subsurface properties, such as porosity, is crucial for various applications, such as identifying energy resources and determining suitable locations for CO2 storage. Typically, obtaining the latter involves using the inversion of seismic records, which can be a challenging process with multiple possible outcomes. Using many seismic measures leads to data size expansion and challenges the processing, necessitating additional time and resources. Deep learning is highly effective at accurately measuring subsurface features when analyzing large amounts of random data. As a result, we performed calculations on seismic measurements and used horizons and borehole records to collect porosity samples. We have put forward a technique for identifying the most pertinent metrics. Afterward, we conducted training on multiple networks to make predictions about porosity. We developed several models and created a "weight‐averaging ensemble" to account for the random nature of the networks. This considers the power of baseline models when incorporating realizations, resulting in a strong, more accurate model. We used statistical analysis to determine the level of uncertainty in the approach. The proposed method for measuring porosity in the F3‐Netherlands offshore block is faster than traditional approaches and produces results similar to a reference model. Key Points: This study developed an approach to identify pertinent seismic attributes for porosity inversion with uncertainty estimatesThese attributes are used to train an array of distinct networks used to generate multiple realizationsA grid search is developed to find the best weights assigned to these realizations, which are combined to achieve the most precise result [ABSTRACT FROM AUTHOR]

Seismic inversion, complemented by machine learning algorithms, significantly improves the accuracy and efficiency of subsurface parameter estimation from seismic data. In this comprehensive study, a comparative analysis of machine learning techniques is conducted to predict subsurface parameters within the inter-well region. The objective involves employing three separate machine learning algorithms namely Probabilistic Neural Network (PNN), multilayer feedforward neural network (MLFNN), and Radial Basis Function Neural Network (RBFNN). The study commences by generating synthetic data, which is then subjected to machine learning techniques for inversion into subsurface parameters. The results unveil exceptionally detailed subsurface information across various methods. Subsequently, these algorithms are applied to real data from the Blackfoot field in Canada to predict porosity, density, and P-wave velocity within the inter-well region. The inverted results exhibit a remarkable alignment with well-log parameters, achieving an average correlation of 0.75, 0.77, and 0.86 for MLFNN, RBFNN, and PNN algorithms, respectively. The inverted volumes portray a consistent pattern of impedance variations spanning 7000–18000 m/s*g/cc, porosity ranging from 5 to 20%, and density within the range of 1.9–2.9 g/cc across the region. Importantly, all these methods yield mutually corroborative results, with PNN displaying a slight edge in estimation precision. Additionally, the interpretation of the inverted findings highlights anomalous zones characterized by low impedance, low density, and high porosity, seamlessly aligning with well-log data and being identified as sand channel. This study underscores the potential for seismic inversion, driven by machine learning techniques, to swiftly and cost-effectively determine critical subsurface parameters like acoustic impedance and porosity. [ABSTRACT FROM AUTHOR]

Understanding subterranean reservoirs, geological characteristics, fluid composition, and hydrocarbon potential strongly relies on precise reservoir characterization. Seismic inversion is a key method in reservoir characterization to approximate the acoustic impedance and porosity of underlying rock formations using seismic and well-log data. A sparse layer reflectivity (SLR) post-stack inversion method approach is used in this study to make thin layers more visible. To generate an impedance volume, it uses a predetermined wavelet library, an objective function, and a regularization parameter, the regularization parameter is a tunable parameter used to control the balance between fitting the data closely (minimizing the misfit) and ensuring a smooth and stable model for and sparseness computed coefficients. This study uses Blackfoot data to estimate the density, velocity, impedance, and porosity of a particular region using the SLR and Radial Basis Function Neural Network (RBFNN). According to the interpretation of the impedance section, a low impedance anomaly zone with an impedance range of (8500–9000) m/s*g/cc is present at a time of (1040–1065) ms. The low impedance zone is classified as a clastic glauconitic sand channel (reservoir zone) based on the correlation between seismic and borehole data. Further, a Radial Basis Function Neural Network (RBFNN) has been applied to the data to estimate porosity volume and to conduct a more thorough examination of the reservoir zone and cross-validate inverted results. The research demonstrates that the high porosity zone, low velocity, and density zone are discovered by the RBFNN technique, and the low impedance zone interpreted in inversion findings are correlating, which confirms the existence of the glauconitic sand channel. This research is crucial for understanding how well SLR, RBFNN, and multi-attribute analysis work to define sand channels. [ABSTRACT FROM AUTHOR]

Reservoir characterization entails mapping properties such as lithology, fluid type, porosity and permeability, to describe reservoir quality and identify areas for the location of new wells for optimum production. In this study, we test the reliability of rock-physics modelling and model-based post-stack seismic inversion for characterizing reservoir rocks located within the 'Rhoda' field, onshore Niger Delta Basin, Nigeria. The aim is to accurately discriminate lithology and estimate reservoir properties for accurate prediction of new prospective areas within the field of interest. Unlike some conventional inversion and geomodelling algorithms, which are probabilistic, model-based post-stack seismic inversion is favoured for its sensitivity in resolving thin-bed tuning for increased stratigraphic resolution and reservoir property prediction. More so, it gives satisfactory results, even with limited well control and fair to poor seismic quality, as will be shown in this study. On the basis of gamma-ray and resistivity log signatures, two reservoir sands, dubbed D6000 and D7000, were identified in the field of study. Along the eastern direction, well correlation revealed a significant net sand reduction. Within the reservoirs of interest, petrophysical parameters such as water saturation (Sw), porosity (Ø), and shale volume (Vsh) were evaluated. With high Ø and low Sw and Vsh values, these reservoirs were classified as good quality reservoirs. Crossplots of well-log properties such as density vs acoustic impedance, velocity ratio vs acoustic impedance, gamma-ray vs density, and gamma-ray vs acoustic impedance showed lithologies and pore fluids such as hydrocarbon sand, brine-sand, and shale. The model-based post-stack seismic inversion results revealed a lateral distribution of hydrocarbon and brine sands characterized by low and moderate acoustic impedance values, respectively. The low values of acoustic impedance slice and seismic horizon attributes (instantaneous phase and frequency) identified away from the drilled locations imply the presence of hydrocarbon-charged sands, which are potential prospects for future exploitation. This study therefore, shows the effectiveness of rockphysics-based modelling and model-based post-stack seismic inversion in the characterization of reservoir sands within the Niger Delta Basin and can be applied in other provinces around the world. [ABSTRACT FROM AUTHOR]

Estimation of reservoir properties from seismic data suffers from non-unique solutions. A workflow based on the numerical reformulation of a laboratory-based rock physics model may reduce the non-uniqueness. This study attempts to integrate seismic and well log data of relatively unexplored parts of Upper Assam (UA) basin using inversion driven by laboratory based rock physics model. The laboratory-based rock physics model was developed based on experimental measurements conducted on rock cores from Tipam and Barail formations of the basin. Seismic inversion analysis was performed in OpendTect, an open-source software on post-stack seismic data to derive the acoustic impedance (AI) using coloured inversion. A multilayered feed-forward neural network was developed to spatially populate different petrophysical properties. Laboratory-based correlations between AI, density, porosity were utilised for the AI model from which velocity was computed using multivariate rock physics equation. This derived velocity value was transformed to AI and subsequently trained with well log to populate density (2.23-2.73 gm/cc) and porosity (7-28%) for the entire survey area. A reasonable to high correlation is obtained between bulk density and porosity derived by NN using well log and that derived by laboratory-based model (r = 0.78, 0.91 for Barail and 0.95, 0.94 for Sylhet formation). Thus, integrating datasets of different scale from seismic to core with well log data using neural network helps to derive more realistic models that helps in quantitative decision analysis. [ABSTRACT FROM AUTHOR]

Seismic inversion is one of the fundamental techniques for solving geophysics problems. To obtain the elastic parameters or petrophysical parameters, it is necessary to establish a direct or indirect mapping that is usually nonlinear between the observed data and the inversion parameters in seismic inversion. The traditional model-based inversion method, which is calculating complex and expensive, relies on strict physics theories. Neural networks are an excellent artificial intelligence method for establishing nonlinear mapping. Theoretically, any linear and nonlinear functions can be fitted with neural networks. Compared with model-based inversion methods, the inversion efficiency and accuracy can be improved remarkably using neural networks with their powerful ability to discover and extract features from big data. By reviewing the application of fully-connected neural networks, probabilistic neural networks, convolutional neural networks, recurrent neural networks, generative neural networks, and physics-based neural networks in seismic inversion, we provide a comprehensive overview of neural network methods to seismic inversion, including the basic principles of different neural networks, types of seismic inversion, seismic datasets, and the general framework of neural networks for seismic inversion. In addition, the future trends of seismic inversion based on neural networks are also discussed, including the application of image segmentation networks and generative adversarial networks in seismic inversion. [ABSTRACT FROM AUTHOR]

Tight sandstones produce an increasing amount of natural gas worldwide. Apart from identifying the gas enrichment, the predictions of lithology and permeable zones are crucial for the prediction of tight gas sandstones. In the present study, a seismic inversion method is developed based on rock physical modeling, by which it is possible to directly predict the lithology and pore structure in tight formations. The double-porosity model is used as a modeling tool in considering complex pore structures. Based on the model, the microfracture porosity is then predicted using logging data, which are used as a factor to estimate microfractures. Parameters representing the lithology and pore structure are proposed and estimated using logging data analyses and rock physical modeling based on the framework of the Poisson impedance. Thereafter, a new AVO equation is established and extended to the form of an elastic impedance for a direct prediction of the lithology and pore structure parameters. Real data applications show that the indicators of lithology and permeable zones are consistent with the production status. They agree with the petrophysical properties measured in wellbores, thereby proving the applicability of the proposed method for the effective characterization of tight gas sandstones. [ABSTRACT FROM AUTHOR]

The rock mechanical properties influence the selection of drilling parameters, optimized drilling trajectory, and appropriate hydraulic fracturing intervals. Estimating the 3-D spatial distribution of these geomechanical properties at the reservoir scale is a challenging task, especially in the case of limited data only at the well locations. Advanced geostatistical techniques can be utilized to represent a reservoir's inherent spatial variation more realistically. In this study, we investigate the spatial variability of rock mechanical properties, including Young's modulus, shear modulus, bulk modulus, and Poisson's ratio, as major constituents of the reservoir geomechanical model. The data are extracted from a hydrocarbon field located southwest of Iran and consist of forty wells. We first build a 1-D model of rock elastic moduli at each well by integrating petrophysical and core-based laboratory measurements and then develop a corresponding 3-D model using geostatistical simulation techniques. Thereafter, 3-D seismic data are employed to optimize the geomechanical model. Results show that the integration of well logs with seismic data increases the accuracy of field-wise 3-D elastic moduli models. Furthermore, we used various co-simulation techniques to demonstrate the improving effect of complementary data in constructing a more realistic reservoir geomechanical model. [ABSTRACT FROM AUTHOR]

The unreliable prediction of the low-frequency components from inverted acoustic impedance causes uncertainty in quantitative seismic interpretation. To address this issue, we first calculate various seismic and geological attributes that contain low-frequency information, such as relative geological age, interval velocity, and integrated instantaneous amplitude. Then, we develop a method to predict the low-frequency content of seismic data using these attributes, their high-frequency components, and recurrent neural networks. Next, we test how to predict the low-frequency components using stacking velocity obtained from velocity analysis. Using all the attributes and seismic data, we propose a supervised deep learning method to predict the low-frequency components of the inverted acoustic impedance. The results obtained in both synthetic and real data cases show that the proposed method can improve the prediction accuracy of the low-frequency components of the inverted acoustic impedance, with the best improvement in a real data example of 57.7% compared with the impedance predicted using well-log interpolation. [ABSTRACT FROM AUTHOR]

Transforming seismic data from the time domain to the depth domain is a very important step when using 3D seismic exploration to guide the exploration and development of coalbed methane (CBM). However, the conventional time-depth conversion method has difficulty meeting the control accuracy requirements of CBM development based on horizontal well technology when the 3D seismic data in a mining area are old. Therefore, a precise time-depth conversion method was found to improving the accuracy of time-depth conversion, which is based on the splicing of seismic inversion velocity and poststack migration velocity. The first step of this method is obtaining the standard layers in the time domain by precise interpretation of seismic data. Then, the inversion velocity and poststack migration velocity are spliced to obtain the complete interval velocity volume of the study area, and the results are corrected. The next step is the prediction of the coal seam floor elevation based on the spliced velocity, and the predicted coal seam floor elevation is corrected by borehole data. Finally, the mesh is between standard layers in the depth domain to obtain the 3D data volume in the depth domain. The method was applied to the time-depth conversion of 3D seismic data in the Yangquan X study area. The results show that the relative error between the predicted results and the borehole data of No. 3, No. 8 and No. 15 coal seam is only 0.72% through the validation of the reserved boreholes, indicating that the method is effective. This study provides a precise method of time-depth conversion for seismic data when there is only poststack seismic data in the mining area, which can not only improve the interpretation accuracy of standard layers but can also improve the prediction accuracy of other layers between standard layers, which can better guide the well location arrangement of coalfield and CBM. [ABSTRACT FROM AUTHOR]

Recent years have seen deep learning (DL) architectures being leveraged for learning the nonlinear relationships across the parameters in seismic inversion problems in order to better analyse the subsurface, such as improved velocity model building (VMB). In this study, we focus on deep-learning-based inversion (DLI) for velocity model building, leveraging on a conditional generative adversarial network (PIX2PIX) with ResNet-9 as generator, as well as a comprehensive mathematical methodology for generating samples of multi-stratified heterogeneous velocity models for training the DLI architecture. We demonstrate that the proposed architecture can achieve state-of-the-art performance in reconstructing velocity models using only one seismic shot, thus reducing cost and computational complexity. We also demonstrate that the proposed solution is generalisable across linear multi-layer models, curved or folded structures, structures with salt bodies as well as higher-resolution structures built from geological images through quantitative and qualitative evaluation. [ABSTRACT FROM AUTHOR]

Failure at steep slopes and areas covered by a landslide can be controlled by laterally extended weak layers within the slope stratigraphy. Seismic data are essential in slope stability studies for identifying and tracking failure planes and the spatial extent of weak layers since weak layers often develop in decimeters and are buried together under 10–100 m sediment and water. Although seismic reflection profiles must be used to characterize weak layers at depths, it is challenging to visually characterize thin weak layers accurately due to the low vertical resolution. This study presents a method of merging 2D seismic data with different source resolutions on the same track to generate high-resolution broadband seismic data and perform seismic inversion to facilitate the tracing of thin weak layers. The resultant profile is a high-resolution acoustic impedance profile, which can track weak layers whose attributes are difficult to characterize using conventional methods. We also confirm the extension of the weak layer based on the porosity profile derived from the acoustic impedance of borehole 19ESDP-103 located on the seismic line used in the present study. Glide-planes, previously considered single or double layers due to their limited resolution, were further subdivided, tracked, and characterized in terms of lateral changes. [ABSTRACT FROM AUTHOR]

In this study, seismic inversion is carried out using a genetic algorithm (GA) as well as a hybrid genetic algorithm (HGA) approach to optimize the objective function designed for the inversion. An HGA is a two steps coupled process, where a local optimization algorithm is applied to the best model obtained from each generation of the GA. The study aims to compare the qualitative as well as the quantitative performance of both methods to delineate the reservoir zone from the non-reservoir zone. Initially, the developed algorithm is tested on synthetic data followed by its application to real data. It is found that the HGA for synthetic data is providing more accurate and high-resolution subsurface information as compared with the conventional GA although the time taken later is less as compared with the former methods. The application to real data also shows very high-resolution subsurface acoustic impedance information. The interpretation of the impedance section shows a low impedance anomaly zone at (1055–1070) ms time interval with impedance ranging from (7500 to 9500) m s−1*g cc−1. The correlation between seismic and well data shows that the low impedance zone is characterized as a clastic glauconitic sand channel (reservoir zone). In seismic inversion using an HGA, one can delineate the areal extent of the reservoir zone from the non-reservoir zone more specifically as compared to the GA-derived impedance. The convergence time of HGA is 4.4 per cent more than GA and can be even more for larger seismic reflection data sets. Further, for a more detailed analysis of the reservoir zone and to cross-validate inverted results, an artificial neural network (ANN) is applied to data, and porosity volume is predicted. The analysis shows that the low impedance zone interpreted in inversion results are correlating with the high porosity zone found in ANN methods and confirm the presence of the glauconitic sand channel. This study is important in the aspect of qualitative as well as quantitative comparison of the performance of the GA and HGA to delineate sand channels. [ABSTRACT FROM AUTHOR]

Full waveform inversion (FWI) is arguably the current state‐of‐the‐art amongst methodologies for imaging subsurface structures and physical parameters with seismic data; however, important challenges are faced in its implementation and use. Keys amongst these are (a) building a suitable initial model, from which a local minimum is unlikely to be reached, and (b) availability of tools for evaluation of uncertainty. An algorithm we refer to as implicit full waveform inversion (IFWI), designed using continuously and implicitly defined deep neural representations, appears in principle to address both of these issues. We observe in IFWI, with its random initialization and deep learning optimization, improved convergence relative to standard FWI model initialization and optimization. Models close to the global minimum, capturing relatively high‐resolution subsurface structures, are obtained. In addition, uncertainty analysis, though not solved in IFWI, is meaningfully addressed by approximating Bayesian inference with the addition of dropout neurons. Numerical experimentation with a range of 2D geological models is suggestive that IFWI exhibits a strong capacity for generalization, and is likely well‐suited for multi‐scale joint geophysical inversion. Plain Language Summary: We propose implicit full waveform inversion (IFWI) by replacing the grid‐based subsurface parameters with a continuous neural network representation. Compared to the conventional full waveform inversion (FWI), this simple reparameterization allows IFWI to start from a random initial model by benefiting from the frequency principle of deep learning optimization. The uncertainty of inversion results using IFWI can be easily performed by adopting the Bayesian neural network, or adding dropout neurons into the Multilayer Perceptron. In addition, one can use a single network to represent physical parameters at different scales, thus accommodating multi‐scale and multi‐physics problems. Synthetic examples demonstrate that IFWI is able to produce a high‐resolution image of subsurface with fine structures, and has strong generalization ability and a certain degree of robustness. Key Points: We propose an implicit full waveform inversion (IFWI) algorithm using continuously and implicitly defined deep neural representationsIFWI has the ability to capture high‐resolution heterogeneities of subsurface with random initializationIFWI allows uncertainty analysis without additional calculations and exhibits a strong capacity for generalization, robustness [ABSTRACT FROM AUTHOR]

Estimation of reservoir parameters (reservoir characterization) including lithology, porosity and water saturation is crucial in the oil and gas industry. These parameters can be calculated in the wells and/or laboratories but these methods are time-consuming, costly, and can cover only a small part of the reservoir. Therefore, integration of seismic and well data for reservoir characterization has received special attention in the oil and gas industry since it provides information from the entire volume of the reservoir. In this study, to establish a relationship between the seismic attributes and the petrophysical parameters of a reservoir at the location of the wells, seismic inversion was performed using model-based, linear programming sparse spike, and maximum likelihood sparse spike algorithms, and the acoustic impedance were calculated accordingly. The model-based inversion method has provided a better answer than the other two methods by providing 99% correlation between the actual and estimated acoustic impedance. Therefore, this method was used to calculate the acoustic impedance in the space between the wells. In the next step, porosity, water saturation, and lithology (quartz and dolomite volume) were estimated from different seismic attributes. In this paper, the multi-attributes regression (MAR) method and artificial neural network (ANN) were used to estimate each of the petrophysical parameters of the reservoir, and we found out that the ANN provided a more accurate estimate than the MAR method for our given dataset. The correlation between the actual and estimated values using the ANN method for porosity, water saturation, quartz volume and dolomite volume is 86, 93, 93 and 95% respectively in the training data and 75, 78, 79 and 82% in the validation data. [ABSTRACT FROM AUTHOR]

Although a mixed carbonate–siliciclastic system of the Strawn Group on the Eastern Shelf in King County, Texas, USA provides excellent hydrocarbon reservoirs, facies variability and reservoir properties within such systems are not well understood. We conducted prestack, simultaneous seismic inversion, and high-level petrophysical analysis to derive elastic properties of rocks to facilitate lithology identification and determination and distribution of the different carbonate facies. Our results show that (1) the Strawn Group in King County is dominated mostly by carbonates and (2) given the ratio of P- and S-wave velocity (Vp/Vs ratio), the carbonates can be separated into three facies: (a) high-Vp/Vs-ratio shelf-edge reef carbonates, in which the Vp/Vs ratio decreases linearly as porosity increases and the Vp/Vs ratio varies from ~2.1 to ≤2.6; (b) moderately low-Vp/Vs-ratio shelf (platform) carbonates, in which the Vp/Vs ratio also decreases as porosity increases and in which the Vp/Vs ratio ranges from ~1.75 to ≤2.15; (c) extremely low-Vp/Vs-ratio slope and basin carbonates, in which the Vp/Vs ratio, although appearing to be almost constant for a wide range of porosity, increases as porosity increases, and in which most Vp/Vs-ratio values appear to range from ~1.5 to ≤2. Results of a through c can be summarized thusly: the Vp/Vs ratio of reef carbonates >the Vp/Vs ratio of platform carbonates and >the Vp/Vs ratio of slope and basin carbonates in the study area. [ABSTRACT FROM AUTHOR]

As an irreplaceable quantitative interpretation method, prestack seismic inversion enables the effective estimation of subsurface elastic parameters for reservoir prediction. However, for the model-driven prestack seismic inversion, the band-limited characteristics and noise interference of observed seismic data result in its high dependence on the initial models. This suggests that reasonable initial models act as a supplement to reliable variation trends in formation and can reduce the non-uniqueness of inversion results. In this article, we introduce a well-log interpolation method with a feature map-guided non-local means algorithm, which is for establishing high-fidelity initial models used for prestack seismic inversion. First, we briefly review the basic theory of general model-driven prestack seismic inversion. Then, we use dictionary learning to split the poststack seismic record into patches, and represent them with sparse vectors, instead of directly using seismic record. The advantage of dictionary learning is that it can adaptively extract useful signals from noisy observed data and provide fine structures by sparse reconstruction. Therefore, the proposed feature extraction method can improve the noise immunity and reliability of the well-log interpolation. More accurate initial models are pre-constructed efficiently by our feature extraction method, which improves the reliability of prestack seismic inversion results. Two kinds of observed seismic data are used, including the poststack seismic record for well-log interpolation and prestack seismic data used for inversion. Synthetic and field data tests both demonstrate the favorable performance of the proposed well-log interpolation method. In summary, a novel and convenient initial model building approach is provided, which contributes to seismic exploration and geologic modeling. [ABSTRACT FROM AUTHOR]

Inverting seismic data to build 3D geological structures is a challenging task due to the overwhelming amount of acquired seismic data, and the very-high computational load due to iterative numerical solutions of the wave equation, as required by industry-standard tools such as Full Waveform Inversion (FWI). For example, in an area with surface dimensions of 4.5 km × 4.5 km, hundreds of seismic shot-gather cubes are required for 3D model reconstruction, leading to Terabytes of recorded data. This paper presents a deep learning solution for the reconstruction of realistic 3D models in the presence of field noise recorded in seismic surveys. We implement and analyze a convolutional encoder–decoder architecture that efficiently processes the entire collection of hundreds of seismic shot-gather cubes. The proposed solution demonstrates that realistic 3D models can be reconstructed with a structural similarity index measure (SSIM) of 0.9143 (out of 1.0) in the presence of field noise at 10 dB signal-to-noise ratio. [ABSTRACT FROM AUTHOR]

Time-lapse 3D Seismic, often abridged as '4D-seismic', is an advanced surveillance tool for monitoring fluid saturation and other property changes in reservoirs due to hydrocarbon production or water/gas injection. 4D seismic and well log data were utilized in monitoring reservoirs in 'X' Field using rock attributes extracted from time-lapse 3D seismic data. Two reservoir markers, identified as HD1000 and HD2000, picked from well logs were marked as HD2 and HD2_Version2 horizons on the base and monitor seismic volumes. Horizon slices of acoustic impedance (AI), Poisson ratio (s), water saturation, density (ρ), Lambda rho (λρ) and porosity were extracted from impedance volumes for the base and monitor seismic using neural network technique. The acoustic impedance slices of the base and monitor horizons show no distinct change, suggesting that this attribute is not very sensitive to changes in the type of fluid in the reservoir. Poisson ratio slices indicate an observable change for the base and monitor horizons. Its value increased in horizon HD2 of the monitor slice. Hydrocarbons have a lower Poisson ratio than water. Therefore, as hydrocarbon is withdrawn from the reservoir during production and replaced with brine, higher values of Poisson ratio predominates. Water saturation slices show low water saturation in the base slices, which implies high hydrocarbon saturation. In the monitor slices, an increase in water saturation was observed due to the replacement of hydrocarbon with brine. The monitor horizon slices for ρ and ρ exhibited a significant increase in the attributes compared to the base slices. No appreciable difference was observed in porosity slices for both horizons. These findings are indicative of pressure depletion and hydrocarbon extraction, and justifies that '4D-seismic' can be used to monitor changes in the reservoir due to hydrocarbon production. This is invaluable in alleviating the risk involved in reservoir management decisions. [ABSTRACT FROM AUTHOR]

Seismic impedance inversion is one of the most commonly used techniques for reservoir characterization. High accuracy and high resolution seismic impedance is a prerequisite for subsequent reservoir interpretation. The data-driven approach offers the opportunity for accurate impedance prediction by establishing a nonlinear mapping between seismic data and impedance. However, existing data-driven methods take the raw seismic data directly as input, making it difficult for the network to learn high frequency weak signal information and resulting in low resolution inversion results. In order to mitigate the above issues, a data-driven seismic impedance inversion method based on multi-scale strategy is proposed. The method first obtains seismic data at different scales using frequency division techniques and do normalization on the extracted multi-scale data to ensure the consistency of the seismic signal energy in different frequency bands. The multi-scale seismic data will then be fed into the network, which helps the network to learn the high frequency information features more easily, and ultimately obtain higher resolution inversion results. We use the most commonly used convolutional neural network (CNN) as an example to demonstrate that the proposed multi-scale data-driven seismic impedance inversion method can improve the resolution of the inversion results. In addition, since the above seismic impedance inversion method is executed trace-by-trace, the f-x prediction filtering technique is introduced to improve the lateral continuity of the inversion results and obtain more geologically reliable impedance profiles. The validity of the proposed method is verified in the application of synthetic model data as well as an actual data set. [ABSTRACT FROM AUTHOR]

The primary aim of this research is to enhance seismic data interpretation and CO 2 monitoring by utilizing seismic inversion techniques based on the simulated annealing method. Simulated annealing is a global optimization technique employed for inverting seismic data and provides better results as compared with local optimization-based inversion. This methodology is implemented in the Utsira Formation, located at a depth of 1000 m within the Sleipner Field, Norway. The study encompasses the analysis of three sets of time-lapse seismic data, first from 1994 (pre-injection), followed by surveys in 1999 and 2001, corresponding to the injection of 2.35 million tonnes and 4.26 million tonnes of CO 2 , respectively. Firstly, synthetic data is used to check the reliability of the algorithm followed by real data application. This process starts by performing the inversion analysis on the synthetic data which shows a decrease in the impedance values observed at the injection site whereas the seismic amplitude increases. The qualitative as well as quantitative analysis depicts that the algorithm works satisfactorily. The same process is applied to the real data from the Sleipner field. Acoustic impedances are calculated using a simulated annealing-based inversion scheme for the pre-injection case in 1994 and post-injection scenarios in 1999 and 2001. Because of the presence of injected CO 2 in the years 1999 and 2001, a low impedance zone that ranged from 2000 m/s*g/cc to 2400 m/s*g/cc appeared at the time interval of 0.85–1.10sec. The interpretation of the inverted impedance section and seismic attribute analysis show no signature of CO 2 leakage. The results indicated that the inverted section which is derived from the SA optimization technique shows very clear CO 2 information offering a more realistic representation with enhanced resolution of the CO 2 plume and its migratory paths. • The study developed a methodology for enhanced CO 2 monitoring of time-lapse seismic data. • The methodology uses seismic inversion based on simulated annealing, and compared with traditional methods. • The developed method provides very high-resolution subsurface information and hence CO2 movement. • Finally, the methodology is applied to the Sleipner data and CO 2 leakage has been detected with a very high confidence level. [ABSTRACT FROM AUTHOR]

Shales represent strongly intrinsic vertical transverse isotropy (VTI) property or polar anisotropy. The presence of vertically aligned fractures makes shale exhibit orthogonal anisotropy and the seismic responses have azimuthal variation. Conventional linearized PP-wave reflectivity expression for orthorhombic (ORT) media includes at least 8 parameters to be inverted, and the amplitude inversion based on such equations can be highly ill-posed. In this paper, we derive a modified PP-wave approximation of reflection coefficients in ORT media and propose a two-step strategy based on the Bayesian-framework inversion method. The new equation represents the seismic response of polar anisotropy and azimuthal anisotropy separately, and it consists of only 5 model parameters to be inverted. Three of these model parameters are azimuthal independent: A—acoustic impedance, B—anisotropic shear modulus and C—horizontal P-wave phase velocity (along the fracture strike), and two of them are azimuthal dependent: D—azimuthal anisotropic gradient and E—horizontal P-wave anisotropy parameter. Accuracy analysis demonstrates that the derived approximation has similar accuracy to the existing approximation. A stepwise inversion strategy is proposed to invert azimuthal-independent parameters and azimuthal-dependent parameters separately. Synthetic tests show that the proposed strategy is more stable and better conditioned than the existing multi-parameter simultaneous inversion. Field seismic data set of a fractured shale-gas reservoir is also used to demonstrate the stability and accuracy of the proposed inversion method. Article Highlights: PP-wave reflection coefficient is derived in terms of only 5 model parameters for orthorhombic media. Three of them are azimuthal independent: A—acoustic impedance, B—anisotropic shear modulus, C—horizontal P-wave phase velocity (along the fracture strike), and the others are azimuthal dependent: D—azimuthal anisotropic gradient, and E—horizontal P-wave anisotropy parameter A stepwise inversion strategy for inverting azimuthal-independent parameters and azimuthal-dependent parameters is proposed Tests on synthetic and field seismic datasets imply that the proposed inversion method has the potential to generate reliable results of the model parameters for reservoir characterization [ABSTRACT FROM AUTHOR]

Hydraulic fracturing (HF) has been used for years to enhance oil and gas production from conventional and unconventional reservoirs. HF in enhanced geothermal systems (EGS) has become increasingly common in recent years. In EGS, hydraulic fracturing creates a geothermal collector in impermeable or low-permeable hot dry rocks. Artificial fracture networks in the collector allow for a continuous flow of fluid in a loop connecting at least two wells (injector and producer). However, it is challenging to assess the fracability of geothermal reservoirs for EGS. Consequently, it is necessary to design a method that considers multiple parameters when evaluating the potential of geothermal development. This study proposes an improved fracability index model (FI) based on the influences of fracability-related geomechanical and petrophysical properties. These include brittle minerals composition, fracture toughness, minimum horizontal in-situ stress, a brittleness index model, and temperature effect to quantify the rock's fracability. The hierarchical analytic framework was designed based on the correlation between the influencing factors and rock fracability. The results of the qualitative and quantitative approaches were integrated into a mathematical evaluation model. The improved fracability index model's reliability was evaluated using well logs and 3D seismic data on low-permeable carbonate geothermal reservoirs and shale gas horizontal wells. The results reveal that the improved FI model effectively demonstrates brittle regions in the low-permeable carbonate geothermal reservoir and long horizontal section of shale reservoir. We divide the rock fracability into three levels: FI > 0.59 (the rock fracability is good); 0.59 > FI > 0.32 (the rock fracability is medium); and FI < 0.32, (the rock fracability is poor). The improved FI model can assist in resolving the uncertainties associated with fracability interpretation in determining the optimum location of perforation clusters for hydraulic fracture initiation and propagation in enhanced geothermal systems. [ABSTRACT FROM AUTHOR]

Time‐lapse seismic (four‐dimensional seismic) data play a preeminent role in closed‐loop reservoir management by providing a full‐field image of dynamic reservoir behaviour during production. Nonetheless, the multidisciplinary nature of four‐dimensional closed‐loop approaches demands more quantitative and fast methods to integrate rock physics models, reservoir flow simulation models and four‐dimensional seismic analysis. In this work, we tackle this time‐consuming and expensive process and develop a data‐driven quantitative approach to leverage the inherent physics between four‐dimensional seismic and reservoir property changes. We propose an inversion flow method using machine learning strategies to estimate changes in reservoir properties directly from a fast‐derived four‐dimensional seismic attribute. This study was carried out in a Brazilian deep‐water field where production started in 2013 with 3 years of production and injection history. For this reservoir, the estimation of fluid saturation maps is a critical objective to assist engineers with data assimilation procedures. We also generated millions of training data samples using 200 simulation models from the field mentioned above (before history matching) to highlight the benefits of restricting training samples to proper fluid flow consistent combinations. Results demonstrate high prediction accuracy for the targeted reservoir property changes. Additionally, it provides insights for the detection of sweet spots and positioning of infill wells. We significantly simplify the four‐dimensional seismic integration process, allowing initial engagements of reservoir engineers with decision‐making processes and data assimilation applications. [ABSTRACT FROM AUTHOR]

Petrophysical properties are the least explored parameters in the Mahanadi basin in the eastern Indian margin. We have used 2D seismic reflection data along a profile AC with available well logs of the Expedition-01 of the Indian National Gas Hydrates Program (NGHP) for the evaluation of lateral variation of porosity – a petrophysical property. The seismic line passes through the Sites NGHP-01-19B in the AB section, and NGHP-01-09A in the BC section. The porosity is estimated using two different approaches based on acoustic impedance (AI) transformation and direct seismic inversion. The AI transformation approach fails to decipher consistent porosity distributions whereas the direct seismic inversion approach provides satisfactory results along seismic line AC (AB and BC) in the Mahanadi Offshore region. The AI transformed porosity section illustrates erroneous porosity values within the target zone, irrespective of a strong coefficient of determination (R 2 ) of 0.82 and 0.85 between AI and porosity at sites NGHP-01-19B and NGHP-01-9A, respectively. The accuracy of the porosity results obtained from the direct inversion approach depends on the best linear regression fit established between acoustic and porosity reflectivity-based models with a coefficient of determination as 0.91 at Site NGHP-01-19B and 0.94 at Site NGHP-01-9A. The porosity distributions are also corroborated with the density-porosity log values, and it is observed that porosity from direct seismic inversion is consistent with the porosity log information at respective sites. The porosity results at the two sites signify the overall good performance of the direct seismic inversion method and also indicate that the probability of free-gas occurrences in shallow sediments is higher compared to that of gas-hydrates. [ABSTRACT FROM AUTHOR]

This study is a comparative analysis of three types of post-stack inversion techniques, namely, Band-Limited (BLI), Linear Programming Sparse-Spike (LP-SSI), along with Model-Based (MBI), in addition to a neural network in the Sufyan oil field of the Muglad rift basin, Sudan. It investigates whether a combination of multi-attribute analysis and a neural network could bring about a better definition of unresolved thin reservoir layers. Applied inversion techniques produce accurate and reliable results, while the MBI method brings a higher correlation coefficient (0.988) and higher amplitude spectrum correlation (0.999). Hence, it is better for the Sufyan seismic data. The resulting resolution and precision of multi-attribute rock properties prediction is achieved using multi-layer feed forward neural (MLFN), probabilistic neural (PNN), along with radial basis function neural (RBFN) networks to predict porosity. The combination of PNN and LP-SSI has the highest training correlation of 0.9534 and validation correlation of 0.7998. However, MLFN shows that LP-SSI, when used as an external attribute, produces high-resolution images compared to those estimated with other combinations with training correlation of 0.9400 and validation correlation of 0.7111. The resulting resolution and precision in terms of porosities are higher when the combination of LP-SSI and MLFN approach is used compared to that estimated by other methods. Thus, LP-SSI is more accurate than other techniques to predict petrophysical parameters in the Sufyan oil field. [ABSTRACT FROM AUTHOR]

In seismic data analysis, acoustic impedance may be used for predicting the porosity of a sedimentary layer with known lithology, where the influence on impedance from pore fluid is known. This method typically assumes that the stiffness of the rock frame is solely a function of porosity and that there is no variation in stiffness between rocks with the same composition and porosity. However, contact cementation can cause an increase in rock frame stiffness with no significant reduction in porosity. This means that rocks with similar porosities and varying degrees of cementation would have significantly different elastic moduli; thus, attempting to derive porosity from seismic impedance alone would incur errors. The primary goal of this study is to assess how significant this error is for porosity prediction. The secondary goal is to calculate the dry rock frame stiffness across the field and evaluate to what extent the time of hydrocarbon placement affects the lithification of the rock. To fulfill these objectives, we studied petrophysical logs and core data from the Dan field and derived seismic impedance, porosity and stiffness expressed in terms of Biot's coefficient for 14 wells across the field. Trends defined between acoustic impedance, porosity, and Biot's coefficient in the Maastrichtian unit of the Dan field showed that, for a given porosity and water saturation in the oil zone, the chalk located at the crest of the reservoir is softer than the chalk located at the flanks of the field. We also found that chalk is stiffer in the down‐faulted north‐western part of the field for a given porosity and water saturation in the oil zone. In the water zone, no difference was found. We speculate that contact cementation downflank in the northwest block caused an increase in the chalk stiffness without reducing the porosity, while earlier hydrocarbon filling of the south‐eastern block and the crest of the structure delayed contact cementation between the chalk particles. This difference in contact cementation between rocks of the same porosity causes porosity prediction from acoustic impedance to vary by up to six porosity units. [ABSTRACT FROM AUTHOR]

The concept of infinitesimal elastic deformation and the theory of elastic seismic waves was formed in the first part of the 19th century and was based mainly on the Fermat, Huygens and Snell developments in the theory of optics. At the same time, seismic wave propagation (utilized in geophysical prospecting) and optic wave propagation through defined media are based on the same physical-mathematical principles, making it possible to transfer nonconventional procedures developed in the first domain to the second one and back. In this investigation, we propose transferring advanced methodologies established in seismic prospecting to practical optics. We selected two advanced approaches with the following aims: (a) homeomorphic imaging; (b) novel description of boundary conditions. The first approach is established with the utilization of the revealed local theoretical relationship between the geometrical features of two fundamental beams and the geometrical properties of hidden geological targets of the media under study. The employed geometrical characteristics of the fundamental beams are spreading functions and curvatures of the singular wavefronts. The second approach is based on a novel description of the boundary conditions. It enables the determination of a faultless seismic (optical) system with the preassigned focusing and imaging assets when any aberrations are absent. An optimal optical system is usually determined as some arrangement agreeing to some perfect system with acceptable correctness. Employment of the developed procedures in the optical design will permit the application of a description of the optical surface using: (1) parametric functions, (2) differential equations, and (3) mixed (parametric-differential). On this basis, optical systems with a minimal number of optical features with complicated shapes can be promptly computed. Another important application field of the suggested methods is the design of optical systems with diffractive elements. [ABSTRACT FROM AUTHOR]

The current research deals with studying the petrophysical properties represented by the porosity and its distribution on the level of all units of the top and bottom of the Kirkuk Formation Group. The study area is located in Maysan province in the southeastern part of Iraq in the Amara field. The Kirkuk Group was deposited in the Tertiary Age. The post-stack method using seismic inversion and creating a relationship between seismic data was accomplished using Hampson-Russel software at well Am-1 and Seismic lines Ama 20 and 30. The research results indicate high porosity values on top of the formation with a decrease in acoustic impedance (Z) and, therefore, a reduction in the density. At the same time, low porosity values were indicated for the bottom of the formation with an increase in both acoustic impedance and density. The top of the formation contains clastic sediments with oil-sand of high porosity, which is considered hydrocarbon indication and low porosity carbonate materials at the bottom of the formation. [ABSTRACT FROM AUTHOR]

We implement a machine-learning inversion approach to infer petrophysical rock properties from pre-stack data that combines a convolutional neural network and a discrete cosine transform of data and model spaces. This transformation is used for model and data compression. The network learns the inverse mapping between the compressed seismic and the compressed petrophysical domain. A theoretical rockphysics model relates elastic and petrophysical properties, while the exact Zoeppritz equations map the elastic properties onto the seismic domain. Training and validation examples are generated under the assumption of a Gaussian variogram model and a non-parametric prior. A Monte Carlo simulation strategy is employed for uncertainty assessment. We present synthetic inversions on a realistic subsurface model and the outcomes of the proposed approach are compared with those achieved by a standard linearised inversion. The network predictions are assessed in case of errors in the calibrated rock-physic model, in the estimated source wavelet, and in the assumed noise statistics. We also demonstrate that transfer learning avoids retraining the network from scratch when the target and training properties differ. Our experiments confirm that the implemented inversion successfully solves the petrophysical seismic inversion, opening the possibility to get instantaneous predictions of reservoir properties and related uncertainties. [ABSTRACT FROM AUTHOR]

Elastic impedance (EI) inversion for partial angle stack seismic data is a key technology in seismic reservoir prediction within the oil and gas industry. EI inversion provides a consistent framework to invert partial angle stack seismic data, just as the AI inversion does for post-stack data. The commonly used EI inversion process is angle by angle. Hence, the inverted EI for different angles may be nonconforming, especially for the seismic data with a low signal-to-noise ratio. This paper proposes to simultaneously invert multiple partial angle stack seismic data to obtain EI for different angles at once. To obtain conformable EI, we used the joint sparse constraint on the reflection coefficients for different angles. Then, the objective function for simultaneous EI inversion was constructed. Next, synthetic seismic data profiles with three different angles were used to show the superiority of the proposed EI inversion method compared to the conventional method. At last, a real seismic data line was used to test the feasibility of the proposed method in practice. The inversion results of synthetic data and real data showed that it provides an effective new alternative method to estimate EI from partial stack seismic data. [ABSTRACT FROM AUTHOR]

In this paper, the recurrent neural network structure of a bidirectional long short-term memory network (Bi-LSTM) with special memory cells that store information is used to characterize the deep features of the variation pattern between logging and seismic data. A mapping relationship model between high-frequency logging data and low-frequency seismic data is established via nonlinear mapping. The seismic waveform is infinitely approximated using the logging curve in the low-frequency band to obtain a nonlinear mapping model of this scale, which then stepwise approach the logging curve in the high-frequency band. Finally, a seismic-inversion method of nonlinear mapping multilevel well-seismic matching based on the Bi-LSTM network is developed. The characteristic of this method is that by applying the multilevel well-seismic matching process, the seismic data are stepwise matched to the scale range that is consistent with the logging curve. Further, the matching operator at each level can be stably obtained to effectively overcome the problems that occur in the well-seismic matching process, such as the inconsistency in the scale of two types of data, accuracy in extracting the seismic wavelet of the well-side seismic traces, and multiplicity of solutions. Model test and practical application demonstrate that this method improves the vertical resolution of inversion results, and at the same time, the boundary and the lateral characteristics of the sand body are well maintained to improve the accuracy of thin-layer sand body prediction and achieve an improved practical application effect. [ABSTRACT FROM AUTHOR]

The Brazilian pre-salt carbonate province is a scientific frontier that has been subject of extensive research and academic discussion. Santos basin carbonates show that reservoir quality is strongly conditioned by the characteristics of the depositional environment and its associated diagenetic history and alterations. Dolomitization and silicification are two of the most common diagenetic processes in Brazilian carbonates, and the combination of sedimentary and diagenetic complexity leads to a much more heterogeneous reservoir, especially complicated for geoscientists trying to mimic these geological features in terms of static and dynamic modelling. This study proposes a methodology for seismic derived diagenetic properties calculation through seismic quantitative interpretation and classification of inverted P- and S-impedance volumes applied to Brazilian Pre-Salt carbonate reservoirs. Our proposition is to establish relations between petroelastic properties and diagenetic features, first in the well-log domain and, later, in the seismic domain through different quantitative probabilistic techniques. Using the available seismic data and probabilistic approaches through stochastic inversion and Bayesian facies classification, we can interpret the occurrence of diagenetic silica and dolomite in 3D volumes. The volumetric analysis generated through the presented methodology can provide much more insights into the sedimentary history of the field, especially for static and dynamic porosity and permeability modelling. • Diagenetic features are a major concern for porosity modelling in pre-salt carbonates. • Most diagenetic studies are thin section limited and simply extrapolate facies away from wells without lateral constraints. • We calculated dolomitization and silicification volumes that are laterally mappable. • 3D diagenetic volumes can provide valuable inputs for petrophysical and geological modelling. [ABSTRACT FROM AUTHOR]

Full waveform inversion (FWI), a powerful geophysical technique for subsurface imaging through seismic velocity-model construction, relies on numerical optimization, thus requiring the computation of derivatives for an objective function. This paper proposes a discrete development for accurate computation of the gradient and Hessian-vector product, providing second-order optimization benefits like higher convergence rates and improved resolution. The approach is a promising alternative for computing the gradient and Hessian action in time-domain FWI, applicable to various geophysical problems. Computational costs and memory requirements are comparable to the Adjoint-State Method and more avorable than Automatic Differentiation. While efficient automatic differentiation algorithms have transformed gradient computation in applications like FWI, challenges may arise in 3D due to unforeseen memory allocations. Our approach addresses this by exploring the reverse mode differentiation algorithm, mapping temporary memory allocations and computational complexity. By means of introducing auxiliary fields all involved wavefield evolutions can be carried out with the very same evolution scheme, in this way simplifying the implementation and focusing the performance improvement effort in a single routine thus reducing the maintenance cost of these algorithms, especially when using GPU implementations. [ABSTRACT FROM AUTHOR]

The fluid identification of fracture media is susceptible to factors such as observation data, fluid sensitivity parameters, and inversion algorithms. However, appropriate nonlinear algorithms can be applied to reduce errors generated during the inversion process, thus helping to improve the reliability of fluid identification. In this paper, analysis and comparison are made to various reflection coefficient approximate formulas, of which the applicability and accuracy are demonstrated in fluid-saturated poroelastic media, and the objective function is constructed further from Russell's reflection coefficient formula. In this sense, the genetic algorithm (GA) can achieve the pre-stack inversion of seismic waves. Based on the inversion results such as P-, S-wave velocity, and density, the Russell fluid indicator, Poisson's ratio, and Poisson impedance are combined to identify the water-, gas-, and oil-bearing fracture. Finally, the applicability and reliability of the method are demonstrated through numerical calculations in some regions of Marmousi2 model and the field data. The results show that the seismic data can be preliminarily analyzed by the inversion method, followed by combining the Russell fluid indicator, Poisson's ratio, and Poisson impedance, effectively reducing the multiplicity of the inversion results and improving the reliability of fluid identification in fractured media. • Proper fluid factors can effectively reduce the multiplicity of inversion results. • The objective function is formed with Russell's reflection coefficient. • Combining three parameters can improve the reliability of fluid identification in fractured media. [ABSTRACT FROM AUTHOR]

Seismic inversion is aimed at building a mapping from low-resolution seismic data to high-resolution impedance data. Most of the traditional methods have satisfactory interpretability, and most parameters tend to have specific physical definitions. On the other hand, deep learning-based methods present poor interpretability as their prediction performance is not always clearly explainable. One of the significant challenges of deep learning-based methods is to quantify the uncertainty of the model. The uncertainty includes aleatoric uncertainty and epistemic uncertainty, and epistemic uncertainty can be used to evaluate the predicted accuracy of the trained model. In this article, we propose a new deep learning model called uncertainty backpropagation network (UB-Net) to perform impedance inversion. The proposed UB-Net is based on a closed-loop framework and can predict the impedance and the epistemic uncertainty simultaneously. UB-Net has three closed-loop data flows, whereby the predicted uncertainty is utilized as the weight of loss functions to improve the inversion accuracy. Experimental analyses demonstrate that UB-Net presents advanced inversion accuracy on both synthetic and real examples. In particular, the mean absolute error (MAE) on synthetic examples drops by 40%, and the Pearson correlation coefficient (PCC) on real examples increases by 2%. Besides, compared with existing approaches, UB-Net presents superior spatial continuity and preserves more geological structures such as little faults in real examples. [ABSTRACT FROM AUTHOR]

Gas content and brittleness characteristics, which are mostly quantified using the fluid factor and brittleness index, are the key factors in the evaluation of coal‐measure gas reservoirs, which are similar to traditional unconventional gas reservoirs. However, most previous seismic inversion research and evaluations only focused on one of the parameters. Therefore, in this study, we performed a combined inversion of the fluid factor and brittleness index based on pre‐stack seismic records collected from a typical coal‐measure gas block in the Sichuan Basin. First, a new P–P wave reflection coefficient approximation based on both parameters was derived, and the precision and sensitivity of the inversion parameters were analysed. We then constructed a new inversion equation based on Bayes' theorem. The logging curves obtained for a typical coal‐measure gas well and through pre‐stack seismic profile were used to evaluate the inversion method. The results of the model test and profile inversion at the well location were in good agreement with the original logging values. Finally, we analysed and discussed the results of the inversion of both parameters for the evaluation of the gas content and brittleness characteristics of the target reservoirs. [ABSTRACT FROM AUTHOR]

The conventional approach in reservoir characterization uses an extrapolation of petrophysical properties calculated from well logs in determining the entire field volume. This technique fails to acknowledge the heterogeneity of a reservoir and consequently leads to errors (under or overestimation) in reservoir capacity quantification. This research employs a deterministic seismic inversion approach in determining the petrophysical properties of each reservoir sand. First, the reservoir sand was identified from intervals of low gamma-ray value with corresponding high resistivity log value. Then, a synthetic seismogram was generated to tie the well to the seismic data and identify events corresponding to the reservoir sands. This was done by convolving the developed reflectivity series, derived from approximated Zeoppritz reflectivity with a zero-phase wavelet generated using the seismic. Acoustic impedance from the well and seismic were analyzed and strongly correlated. The impedance was then constrained with determined petrophysical parameters from the well, and it was used in generating the petrophysics base model used in the inversion. Finally, deterministic seismic inversion was carried out to determine the petrophysical properties such as porosity, the volume of shale, and water saturation associated with the fluids from the seismic volume of Jay-field shallow offshore of Niger Delta. [ABSTRACT FROM AUTHOR]

The identification of gas content is very important for exploration and development of coalbed methane (CBM) reservoirs. As a kind of gas-bearing reservoirs in coal seam, CBM reservoirs usually show strong heterogeneity, which makes the gas content varies greatly in the strata. What's more, the thin interlayer that is common in coal-bearing formation makes it difficult to predict the favorable gas-bearing distribution based on conventional methods. In this study, a seismic waveform indication inversion method was applied to reveal the gas content of No. 8 coal seam reservoirs in the Upper Carboniferous Benxi Formation of the DJ area in the eastern of the Ordos Basin, China. The first step of this method is to calculate the p-wave impedance inversion volume of No. 8 coal seam. The second step is to build the correlation between elastic parameters and gas content in No. 8 coal seam. Through the statistical analysis based on velocity, density, p-wave impedance and measured gas content data of 16 wells, the fitting formula between p-wave impedance and gas content is obtained, with a highest correlation coefficient up to 96%. The third step is to calculate the gas content data volume from the p-wave impedance inversion volume by the above fitting formula, and then the quantitative plane distribution of gas content in No. 8 coal seam can be predicted. The prediction results indicate that the gas content of No. 8 coal seam can be divided into two Classes. To verify the reliability of the inversion results, the production data of well X11 was applied to verify the gas content which located in the Class I area. The application of seismic waveform indication inversion has provided a precise prediction for the spatial distribution of gas content in CBM reservoirs, serving as a basis for locating and designing wells for CBM development. [ABSTRACT FROM AUTHOR]

Geostatistical seismic rock physics amplitude-versus-angle (AVA) inversion allows the joint prediction of rock and fluid properties from seismic reflection data. In these seismic inversion methods, the model perturbation and update occur iteratively in the petrophysical domain. A facies-dependent precalibrated rock physics model is applied to the simulated rock properties to calculate elastic properties. Synthetic seismic reflection data are computed from these elastic models. The rock physics models are calibrated at the well locations and act as the link between the rock and the elastic domains, remaining unchanged during the inversion procedure: the convergence of the inversion and the geological plausibility of the inverted rock property models depend on the quality of the calibration and available well-log data. To overcome this limitation, account for calibration errors and geological scenarios not sampled at the well locations, we introduce an iterative geostatistical seismic AVA inversion where the elastic predictions from a facies-dependent precalibrated rock physics model are updated iteratively, based on the mismatch between observed and synthetic data. We assume an initial $a$ priori uncertainty in the rock physics model and update it iteratively based on the data mismatch. The proposed method is applied in a 1-D synthetic example to illustrate the concept and validate the method. Then, we apply the inversion method to a 3-D real data set. In this application, we use a blind well to assess locally the performance of the technique. The results show that the parameters of space exploration is not compromised by the rock physics model’s uncertainty. [ABSTRACT FROM AUTHOR]

One of the main goals of pre‐stack seismic inversion is the estimation of elastic properties (i.e. P‐, S‐ wave velocities and density) and litho‐fluid classes in the investigated area. To this end, many inversion strategies have been proposed, but the most popular is based on a two‐step inversion approach: First, elastic properties are inferred from pre‐stack data, and then a classification algorithm is used to convert the outcomes of the first stage into litho‐fluid facies. In this work, we propose an alternative approach based on recurrent neural networks. We train two bidirectional long short‐term memory networks to predict the inverse mappings from pre‐stack seismic data to elastic properties, and litho‐fluid classes. In the elastic inversion, we also use a Monte Carlo simulation approach to properly propagate onto the model space both the uncertainties related to noise contamination in the data and to the modelling error introduced by the network approximation. One crucial aspect of any machine‐learning inversion strategy is the definition of an appropriate training set. In this case, the models forming the training and validation examples are drawn according to a previously defined elastic and facies prior models derived from actual well log recordings. In particular, we assume a Gaussian‐mixture elastic prior, and we also take into account the uncertainties affecting the estimation of the transition probabilities of facies. We invert each seismic gather independently, and in this context, the generation of the training set, and the learning process can be accomplished with a very limited computational effort on a common notebook. Once trained, the networks estimate the elastic properties, the litho‐fluid facies and the related uncertainties from the pre‐stack data in near real‐time. Synthetic and field data inversions are used to validate the proposed method. The network predictions are also benchmarked against the outcomes of a more standard two‐step approach that combines a linear elastic inversion and a subsequent point‐wise Bayesian classification. Our results demonstrate that the implemented algorithm guarantees more accurate elastic property estimations and facies predictions than the standard inversion strategy. In particular, the predictions provided by the long short‐term memory network are less affected by erroneous assumptions on the noise statistics and prior model and by errors in the estimated source wavelet. [ABSTRACT FROM AUTHOR]

Based on interpretation and inversion of high‐resolution seismic data, a previously undocumented occurrence of pore‐filled gas hydrates that are associated with a submarine canyon system was discovered in the shelf break of the north‐east Qiongdongnan Basin (QDNB) in the northern South China Sea (SCS). Continuous bottom simulating reflections (BSRs) were recognized in the submarine striking NW‐SE ridges along the shelf break. The two‐way travel times (TWTs) of the BSRs are about 200–250 ms below the sea floor and the burial depths are mapped at ~180–225 mbsf after time‐depth conversion. The distribution of the enhanced reflections (ERs) and acoustic blanking zone overlying the BSRs indicate possible pore‐filled gas hydrate enrichment. ERs have been commonly observed immediately below the BSRs, indicating the accumulation of free gas, which exhibits distinct amplitude versus offset (AVO) anomalies. The acoustic impedance inversion results show that the gas hydrate saturation is 20–50%, and the thickness of the gas hydrate‐bearing layers is usually ~10–40 m, with a maximum thickness of up to ~90 m in local regions. Large‐scale listric faults originating from the deep sag connect the hydrocarbon source region and the shallow gas hydrate stability zone (GHSZ) and act as migration pathways for biogenic gas and potentially thermogenic gas. In addition, hydrocarbons may also be transported into the GHSZ along the sidewalls of the canyon. Canyon incision is believed to cause the increase in the burial depth of the BSRs on the flank of the ridge toward the base of the channel and thereby to trap the hydrocarbons migrating along the sidewall in permeable beds due to the decrease in permeability caused by the gas hydrate accumulation, resulting in the occurrence of ERs immediately below the BSRs. It is speculated that in the channel‐levee system, the coarse‐grained sediments should be deposited within the GHSZ, and the reservoir's physical properties are conducive to the wide distribution and enrichment of pore‐filled gas hydrates in the shelf break of the northern QDNB. [ABSTRACT FROM AUTHOR]

Traditional optimization algorithms are usually applied to estimate the elastic parameters of the subsurface by using field seismic data. However, these optimization algorithms highly depend on prior knowledge (e.g., the initial model setup and sparsity), leading to serious inversion uncertainties. Nowadays, with the rapid development of neural networks, convolutional neural networks (CNNs) have been widely imposed on estimating elastic parameters from field data. However, the deficiency of labeled seismic data impedes the CNNs application in seismic inversion. Moreover, both the size and diversity of labeled datasets are also critical factors influencing the accuracy and resolution of predicted parameters when using the CNNs-based inversion techniques. In this work, taking the unconventional tight sandstone formation as an example, we develop a geological and geophysical model driven CNNs (GGCNNs), named as GGCNNs. The proposed GGCNNs allow us to take advantage of both the prior geological information and basic geophysical model from the generated synthetic labeled prestack seismic datasets, representing essential characteristics of the subsurface. Moreover, under the consideration of data diversity, the GGCNNs model enables us to make a tradeoff between the inversion accuracy and labeled data size. Applications on both synthetic and field data clearly demonstrate the effectiveness of the proposed GGCNNs model for predicting elastic parameters by using prestack seismic data, i.e., its predicted results are with high accuracy in the vertical profile and continuity and smooth in the horizon slice. [ABSTRACT FROM AUTHOR]

Increasing mining depths and intensities have resulted in frequent accidents during roadway excavation, thus increasing the necessity of preemptively determining the geological conditions of rock formations along roadways. Identifying the spatial distributions of lithotypes and the mechanical properties of non-marker bed intervals in coal measure strata are crucial for roadway horizon design, roof management, and efficient excavation. Therefore, based on petrophysical tests, well logging, and seismic data, a comprehensive method is proposed, which uses post-stack wave impedance inversion, petrophysical statistical analyses, and pre-stack simultaneous inversion to predict the lithology and elastic parameters of rocks surrounding roadways. The lithology of the target layer is first divided according to the wave impedances of the inversion body, and the sand lenses within the mudstone interval are described. Based on petrophysical test data, the shear-wave conversion equations of the sand lenses in this target interval and background mudstone are then obtained via marginal limit extrapolation. Finally, pre-stack simultaneous inversion, based on the seismic data and shear-wave conversion curves of different lithologies, is performed to obtain the elastic parameter profile of the target interval in the cross-section of the roadway, laying the foundation for further prediction of the mechanical properties of the surrounding rocks. The results agree with roadway drilling data by > 90%. [ABSTRACT FROM AUTHOR]

In this study, we develop a 3D geostatistical seismic inversion method based on turning bands simulation and co-simulation to estimate acoustic impedance (AI) from seismic and well data. The proposed method uses an iterative approach based on cross-over concept in genetic algorithm optimization to perform global stochastic inversion. The objective function of the optimization algorithm is the measure of correlation coefficients between the modeled and observed seismic data. In the first iteration of the proposed algorithm, the seismic cubes corresponding to the AI realizations generated by the turning bands simulation are compared to the observed seismic data. Subsequently, the local areas of AI models producing the highest correlation coefficients to the observed seismic data are merged to construct a new supplementary cube that is further used as the secondary variable in the turning bands co-simulation for the next iteration. Generation of the seismic cubes and picking the best AI cube via cross-over genetic algorithm approach is performed iteratively until convergence to a constant correlation coefficient between the modeled and observed seismic data. The proposed method is applied to a synthetic dataset that includes 20 wells with known AI logs and 3D stacked seismic data. According to the results, the algorithm converges to a constant correlation coefficient after a few iterations. In addition, it is observed that employing multi-attribute analysis outputs (meta-attributes) during turning bands co-simulation in the initialization step would improve the final global correlation coefficient from 0.774 to 0.815. [ABSTRACT FROM AUTHOR]

A 3D velocity model was created by using stacking velocity of 9 seismic lines and average velocity of 6 wells drilled in Iraq. The model was achieved by creating a time model to 25 surfaces with an interval time between each two successive surfaces of about 100 msec. The summation time of all surfaces reached about 2400 msec, that was adopted according to West Kifl-1 well, which penetrated to a depth of 6000 m, representing the deepest well in the study area. The seismic lines and well data were converted to build a 3D cube time model and the velocity was spread on the model. The seismic inversion modeling of the elastic properties of the horizon and well data was applied to achieve a corrected velocity cube. Then, the velocity cube was converted to a time model and, finally, a corrected 3D depth model was obtained. This model shows that the western side of the study area, which is a part of the stable shelf, is characterized by relatively low thickness and high velocity layers. While the eastern side of the study area, which is a part of the Mesopotamian, is characterized by high thickness and low velocity of the Cretaceous succession. The Abu Jir fault is considered as a boundary between the stable and unstable shelves in Iraq, situated at the extreme west part of the study area. The area of relatively high velocity gradient is considered as the limit of the western side of the Mesopotamian basin. This area extends from Najaf-Karbala axis in the west to the Euphrates River in the east. It is found that the 3D stacking velocity model can be used to obtain good results concerning the tectonic boundary. [ABSTRACT FROM AUTHOR]