Seismic Forward Modeling Research Articles

Seismic forward modeling of acoustic surface-related order-separated multiples

Seismic surface-related multiples have become a hot topic of great significance due to the buried geological information provided by broader illumination areas than primaries. In recent years, researchers attempt to extract the hidden hint of multiples rather than treating them as noise and eliminating them directly. The elimination methods, e.g., the surface-related multiple elimination (SRME) and the inverse scattering series free-surface multiple elimination (ISS-FSME), may be affected by the overlapping or proximity of primaries and multiples. Typical imaging methods, e.g., the reverse time migration (RTM) and the least-square reverse time migration (LSRTM), suffer severe crosstalk artifacts from multiples of inappropriate order and smooth migration velocities. To study the characteristics of primaries and surface-related multiples, whether for elimination or imaging, we propose a forward modeling method of acoustic surface-related order-separated multiples established on the areal/virtual source assumption. The free surface is replaced with an absorbing surface under the dipole source approximation and the ghost creation approach. We present two reflection operators to approximate the reflection at the free surface and apply them to the areal source to obtain ideal results. Numerical experiments on three models prove the effectiveness of the proposed forward modeling method of acoustic surface-related order-separated multiples.

P-wave prediction method under multi-source spatiotemporal feature fusion and physics-informed neural network

Compressional wave velocity (P-wave) plays an important role in seismic amplitude versus offset (AVO) forward and inversion modeling and reservoir prediction. However, due to the high cost and difficulty of P-wave logging, the actual acquired P-wave data is limited, which generally does not suffice for practical applications. To provide reliable data support for oil and gas reservoir exploration, a P-wave prediction method under multi-source spatiotemporal feature fusion and physics-informed neural network (MSTFPIN) is proposed. First, a spatiotemporal hybrid model (STHN) is constructed by combining convolutional neural network (CNN) and bidirectional long short-term memory (BLSTM) neural network to accurately express the spatial and temporal characteristics of the logging curves, which utilizes the local perception ability of CNN and long short-term memory function of BLSTM to mine their local shape and depth trends. Then, seismic data are introduced as an additional input of multi-source information, which fills the gap of inter-well geological information, and feature fusion is used to fully exploit the information from multiple sources to enhance the available data. Finally, a physics-informed neural network (PINN) is constructed based on the idea of seismic forward modeling in petroleum engineering to guide the training process of the prediction model, which ensures that the prediction results are consistent with geological laws. P-wave prediction experiments are conducted using real logging and seismic data to verify the effectiveness and generalization ability of the proposed method compared with other methods. The experimental results show that the proposed MSTFPIN model exhibit better prediction performance, and the correlation coefficient and coefficient of determination reach 0.9619 and 0.9234, respectively. Compared with traditional empirical and regression models, the proposed MSTFPIN exhibited higher accuracy and generalization. Overall, our results demonstrate the proposed method can be implemented in real oil fields to improve the accuracy of reservoir prediction.

Physics-driven self-supervised learning system for seismic velocity inversion

Seismic velocity inversion plays a vital role in various applied seismology processes. A series of deep learning methods have been developed that rely purely on manually provided labels for supervision; however, their performances depend heavily on using large training data sets with corresponding velocity models. Because no physical laws are used in the training phase, it is usually challenging to generalize trained neural networks to a new data domain. To mitigate these issues, we have embedded a seismic forward modeling step at the end of a network to remap the inversion result back to seismic data and thus train the neural network through self-supervised loss, i.e., the misfit between the network input and output. As a result, we eliminate the need for many labeled velocity models, and physical laws are introduced when back-propagating gradients through the seismic forward modeling step. We verify the effectiveness of our approach through comprehensive experiments on synthetic data sets, where self-supervised learning outperforms the fully supervised approach, which accesses much more labeled data. The superior performance is even more significant when compared with a new data domain that has velocity models with faults and more geologic layers. Finally, in case of unknown and more complex data types, we develop a network-constrained full-waveform inversion (FWI) method. This method refines the initial prediction of the network by iteratively optimizing network parameters other than the velocity model, as found with the conventional FWI method, and demonstrates clear advantages in terms of interface and velocity accuracy. With these measures (self-supervised learning and network-constrained FWI), our physics-driven self-supervised learning system successfully mitigates issues such as the dependence on large labeled data sets, the absence of physical laws, and the difficulty in adapting to new data domains.

All-in-one proxy to replace 4D seismic forward modeling with machine learning algorithms

Time-lapse seismic data provides valuable information to assist reservoir monitoring, model-based reservoir management, and development activities. 4D seismic information can be used quantitatively to estimate saturation-pressure changes or update reservoir simulation models through a data assimilation process, which traditionally requires a 4D seismic forward model that includes two steps: (1) petro-elastic model to relate rock and fluid properties to elastic attributes and (2) seismic model to depict wave propagation in the reservoir. The traditional forward model brings some complications in quantitative applications, such as time-lapse seismic history matching. Its multidisciplinary nature is a significant bottleneck to update simulation models, once observed 4D seismic data is used with production data in the assimilation process. Furthermore, it is time-consuming, especially in three-dimensional simulation models within data assimilation with several iterations. Here, the traditional forward model needs to be run hundreds or even thousands of times. Aside from being time-consuming, it requires substantial computer memory for complex geology, where more sophisticated seismic models are needed. In 4D seismic quantitative applications, the timely use of the information and its full benefits fade away due to the computational cost and time of the 4D seismic forward model. This research proposes a proxy (we call it, S4D-Proxy) to replace the traditional approach. The S4D-Proxy leverages machine learning models to detect hidden patterns and learn relations between input features (e.g., porosity and saturation-pressure changes) and the target (time-lapse difference of seismic amplitude). Training data are prepared and fed to the machine learning algorithms to relate the inputs and the desired output. The performances of the machine learning models are evaluated based on a numerical measure (coefficient of determination) and visual comparison of the results with the traditional forward model. Our results show that the average coefficient of determination for the test dataset is in an acceptable range. Moreover, the visual comparison of the proxy predictions with those from the traditional approach shows high similarity. Most hardening and softening 4D signals are reproduced with the proxy models. The main advantage of our approach is that it lowers the computational cost and time. The application of the proxy in data assimilation process could offer a significant speed advantage with faster 4D seismic forward modeling for data assimilation iterations. Unlike the traditional forward model, which is one step (petro-elastic model) then another (seismic model), our approach performs the forward modeling at once (all-in-one proxy model). It also offers an alternative 4D seismic forward model for time-lapse seismic feasibility study. All these advantages make the S4D-Proxy valuable and practical in permanent reservoir monitoring (PRM), where the reduced turnaround time in the forward model indicates the timely use of the 4D seismic information.

Compression of seismic forward modeling wavefield using TuckerMPI

With the increasing use of seismic waveform inversion and imaging, the scale of numerical simulations of seismic waves has been increasing. Concomitantly, the resulting extremely large amounts of seismic data being produced are causing storage, reading, and processing difficulties. Data compression, especially lossy compression, can reduce the data storage requirement. In this study, we apply TuckerMPI, a recently developed lossy large-scale data compression tool, in seismic forward modeling, and compare TuckerMPI with ZFP, SZ2, SZ3, DCTZ, and TTHRESH on compression ratio, accuracy, compression and decompression times. Using TuckerMPI to accelerate seismic waveform inversion has also been studied. Instead of using the waveform difference with/without compression as an acceptable criterion, we focus on the influence of compression on the sensitivity kernels calculated using the 4D wavefields. Especially, we quantitatively investigate the allowed relative error level that does not significantly affect the results of the waveform inversion. Numerical experiments show that TuckerMPI can reduce the file size by 1000+ fold while the inversion result is similar to that using the raw fields.

How petrophysical properties influence the seismic signature of carbonate fault damage zone: Insights from forward-seismic modelling

Seismic imaging is the most common tool to detect subsurface structures in continental and marine settings. Despite technological advancement, seismic analysis of carbonates is still challenging because they are strongly influenced by petrophysical heterogeneities, being this even more difficult when further heterogeneities are added by the presence of faults. In this work, unmigrated forward-seismic models are developed to understand the seismic response changes related to carbonate-bearing fault systems and their deformation behaviour. We focus on the carbonate ramp of the Majella Massif (central Italy), that is characterised by the presence of porous and faulted carbonate lithologies. This carbonate system represents an analogue of carbonate reservoirs worldwide. Field and laboratory data of fault rocks sampled at increasing distances from the slip planes highlight a damage zone/fault core architecture with a decreasing porosity approaching fault planes. This observation is also confirmed by the thin sections analysis and the calculation of the shear modulus. Seismic images of fault rocks presenting lower porosity than the host rock show weak diffraction hyperbolas while diffraction hyperbolas are more evident in the forward-seismic models of fault rocks with greater porosity than the host rock. Moreover, factors such as the variation in the dip angle of the fault or the thickness of the damage zone can enhance or reduce the hyperbolas. Performing the migration process to stacked sections would not provide evidence of increased porosity in the damage zone due to the suppression of the diffractive component. The presence of weak hyperbolas in unmigrated seismic images is thus interpreted as evidence of a lower porosity damage zone with respect to the host rock. This can be related to deformation mechanisms leading to porosity loss in the damage zone for porous rocks as observed in the study area. However, this could also be related to the confining stress or fracture filling that counteract the fracture-related porosity increase when faults are hosted by tight rocks.

Smart Seismic Modeling - Artificial Intelligence in the Petroleum Industry

Vertical seismic profile (VSP) data have many known advantages, if being compared to seismic data such as broader bandwidth, less attenuation, clearer under well true depth (TD) image, multiple noise differentiation, shear wave information, direct attenuation and anisotropy measurements, higher fidelity amplitude versus offset (AVO) effect, etc. In this paper, we share the bottom-up approach that aims to maximize the information extracting from borehole seismic, in conjunction with seismic and well log data, to generate a geological model by integrating full wave-field modeling (FWM) to deep learning techniques, called smart seismic modeling (SSM) project. The study starts using data at well locations, with vertical seismic profile (VSP) forward modeling process. The 2D rock physics model is thus built from the prior well logs and the major interpreted seismic horizons and faults. By altering rock-physics properties values and thickness of each layer, the model has been multiplied thousand times. The rock-physics models and their simulated seismic traces are fed, as labels and inputs, respectively, into a 2D deep learning network. The network then extracts the non-linear relationships between inputs and labels so that thickness and rock-physics properties can simultaneously be regressed with least squared error loss function. Optimal trained weight set (artificial intelligence engine) is finally used to predict wellbore’s surrounding geologic structures and rock properties for all the wells in the study area. To make sure machine learning can provide a good work, similar training and prediction process is carried out from surface seismic forward modeling. Not only is it used for rock property prediction at the area apart from wells, but also for cross checking the prediction at well location. All 2D predicted rock-property models are merged by multi-point statistics (MPS) method using hard data from wells’ prediction and trends from actual VSP processing, such as anisotropy, 2D Qp, Qs, for a complete 3D geologic model. By doing with two different modeling processes, i.e., VSP forward modeling and surface seismic forward to cross check the results, results can obtain from this study are: 1)high quality static model for field development; 2)high quality dynamic model for field monitoring; 3)robust artificial intelligence engines for variety types of geology. In this study, we introduce the SMM approach and share promising successes in applying deep learning in rock property inversion process. The successful showcase has proved that deep learning technique can be used to extract any kind of rock properties from the big data of VSP, well log and of seismic.

Hydrocarbon accumulation conditions and exploration targets of the Sinian Dengying Formation in the Southeastern Sichuan Basin

No breakthrough has been made in petroleum exploration of the Sinian Dengying Formation in the southeastern Sichuan Basin as the major factors controlling the hydrocarbon accumulation of the Dengying Formation have not been well established, including the development period and distribution of the platform margin, the effectiveness of the source-reservoir combination and the reliability of the Cambrian pre-salt structural trap. Based on the interpretation of seismic data and hydrocarbon-source correlation, the distribution of the platform margin of the Sinian Dengying Formation in the southeastern Sichuan Basin was mapped. Moreover, the gas source rocks of the Dengying Formation were clarified according to geochemical characteristics of fluid inclusions, and the pre-salt structures were further confirmed by seismic forward modeling and experiments of prestack depth migration. Results showed that the sedimentary facies of the 4th member of Dengying Formation in the southeastern Sichuan Basin was platform margin of gentle slope type. The platform margin reservoir of the Dengying Formation and the Lower Cambrian source rocks constitute a favorable source - reservoir combination. The carbon isotope characteristics of gas in fluid inclusions indicate that some of the oil/gas in the Dengying Formation were contributed by the Doushantuo Formation source rocks. As the difference in seismic velocities between carbonate rocks and gypsum/salt beds may cause a reduction in the amplitude of pre-salt structures, the low-amplitude structures identified on current seismic profiles are underestimated instead of overestimated. Accordingly, the Sinian structural traps are confirmed to develop in the deep pre-salt sequences in the low-steep anticline belt, southeastern Sichuan Basin. On the basis of the above new understanding, we propose two types of hydrocarbon migration modes for the Sinian Dengying Formation in southeastern Sichuan Basin, including lateral migration of hydrocarbons generated from the Cambrian Qiongzusi Formation and vertical migration of hydrocarbons derived from the Sinian Doushantuo Formation. Lithologic traps at the platform margin and pre-salt structural traps within the platform are favorable targets for natural gas exploration.

Reservoir multiparameter prediction method based on deep learning for CO2 geologic storage

Time-lapse seismic difference refers to the comprehensive response of fluid saturation, pore pressure, and porosity. However, the contribution of different parameters to the seismic response is difficult to distinguish. The high-precision prediction of these reservoir parameters is of great significance in CO2 geologic storage and oil and gas development. Therefore, a simultaneous time-lapse reservoir multiparameter prediction method based on a multitask learning network is proposed. Combined with CO2 geologic storage monitoring, the process of generating training data is described, involving numerical simulation, petrophysical models, and seismic forward modeling. Moreover, the Hertz-Mindlin formula, which considers pressure changes, is used to establish the relationship between formation elasticity and physical parameters in CO2 storage. The effects of fluid saturation, pressure, and porosity on P- and S-wave velocities and densities are analyzed, and the amplitude-variation-with-offset response characteristics of fluid saturation, pressure, and porosity changes are discussed. In total, 4700 and 300 sets of reservoir parameters and seismic angle gather data are used for network training and testing, respectively. The prediction results of synthetic and field data find that the time-lapse reservoir multiparameter prediction method based on multitask learning can effectively distinguish changes in each parameter and simultaneously obtain high-precision prediction results of fluid saturation, pressure, and porosity. Once the network is constructed, the prediction will take only a few seconds, which will promote the further development of the CO2 geologic storage theory and technology.

Lithosphere Structure of the Southern Dinarides and Continuity of the Adriatic Lithosphere Slab Beneath the Northern Dinarides Unravelled by Seismic Modelling

Currently recognised lithospheric models hypothesise the non-existence of a lithospheric slab (a so-called slab gap) in the area of the Northern Dinarides, and the possible existence of a shallow slab in the Southern Dinarides. These geological models are mostly based on previous regional and global 3D velocity models obtained from teleseismic tomography. Recent local tomographic models providing a good resolution have regularly shown the existence of a fast anomaly underneath the entire Dinarides, directly indicating the existence of a lithospheric slab. To avoid interpretation pitfalls and increase reliability, forward seismic modelling based on new tomographic models was performed. Seismic modelling indicates a continuous lithospheric slab along the entire Dinarides in the shallow mantle, but it is not continuous vertically. In the Northern Dinarides, the shallow lithospheric slab extends at least to a depth of 150 km. In the Southern and Central Dinarides, there is a deep fast anomaly that can be interpreted in two ways due to the weak vertical resolution of teleseismic tomography. The first model suggests a steeply dipping continuous Adriatic lithospheric slab whereas the second model shows that the slab consists of two separate blocks, meaning that the deeper block was formed by delamination of the Adriatic lithospheric slab. Due to a similar correlation between the inverse velocity models for the synthetic model and the observed data, preference is not given to any model. The second model could indicate two independent Dinaridic subduction phases, as opposed to viewing subduction as a single long process during the geological past.

Seismic forward modelling of the Kvalhovden outcrop, Spitsbergen, Norway

With the advent of seismic forward modeling, it is now possible to investigate and bridge the resolution gap between outcrop, reservoir, and seismic-scale observations. Seismic processing is central to seismic technology and has grave consequences on the work of a seismic interpreter, reserve estimation, and economic decisions. To investigate the strength and inadequacies of contemporary processing algorithms used for seismic simulations, this work provides several modeled seismic sections from the Kvalhovden outcrop, Spitsbergen, Norway. We emphasize the differences in migrated and inverted seismic sections and their ability to accurately render the geological model. Our results show that Reverse Time Migration (RTM) significantly improves the geological model compared to the Kirchhoff migration and that the elastic RTM migration is excessive. In the RTM case, a smooth velocity model was sufficient to achieve realistic results. However, all migrated images have a false continuity, structures, apparent truncations, and false low-angle terminations. Introducing colors to the modeled sections create illusionary details, exaggerated the stratification, and reinforced the occurrence of seismic artifacts. In the inverted images, sandstone blocks and the underlying collapse structures are better imaged than on the migrated sections and are with a reduced number of artifacts. Importantly, our work shows the relevance of inverted sections for general structural and stratigraphic interpretation, albeit low resolution compared to the migrated data. Migrated and inverted sections should be used jointly to improve seismic interpretation quality.

Fast iterative method for seismic wave travel time calculation under undulating surface conditions

Seismic wave travel time is an important seismic attribute information and is widely used in various seismic forward and inversion methods, including seismic migration imaging, seismic tomography, seismic wave forward modeling and other core seismic data processing methods. The accuracy and efficiency of the travel time algorithm are important for the above methods. In the practical application of seismic exploration, it is often carried out under the condition of undulating surface, which has a significant impact on the travel time calculation. Therefore, it is of great significance to study an efficient and high-precision travel time calculation method that can adapt to undulating surface condition. In this paper, Fast iterative method (FIM) is modified to a topography travel time calculation method. The method employs an iterative method to solve the equation of function to obtain seismic wave travel time by maintaining a narrow band called the active list, and the algorithm can update all grid nodes in the active list at a time. We will verify the travel time computing power of the FIM method through several different velocity models.

Deep carbonate reservoir and gas prediction based on multicomponent seismic amplitude attributes — A case study

Multicomponent seismic exploration provides more valuable information for the prediction of underground structures, lithology, fluids, and fractures. Most studies on lithology estimation and fluid description using PP- and PS-wave seismic data mainly focus on relatively shallow targets with high porosity. Focusing on deep carbonate (dolomite) reservoirs with low porosity (generally less than 5%), we first analyze the characteristics of the seismic response difference between PP and PS waves, and then we combine logging data and seismic forward modeling to implement the identification of fluids using the amplitude difference between PP and PS waves. Data sets are acquired over grain beach dolomite reservoirs in the Cambrian Longwangmiao Formation, the central Sichuan Basin, Western China. The reservoirs are filled with different types of fluids (e.g., gas, water, bitumen, etc.). Based on comparisons between seismic responses of PP and PS waves, we observe that, at the location of the gas-bearing layer, PP-wave amplitude exhibits strong peaks and PS-wave amplitude finds relatively weak peaks, which is different from that obtained for the water-saturated layer. However, at the location of bitumen-filled reservoirs, PP- and PS-wave amplitudes exhibit strong peaks, and the reflection amplitude of PP waves is stronger than that of PS waves. Therefore, the amplitude difference of PP and PS waves provides valuable information and feasibility for fluid identification. We verify that the maximum peak amplitude attribute of PS waves may better characterize the distribution of porous dolomite reservoirs than that of PP waves, and using the attribute of maximum peak amplitude difference between PP and PS waves, we may distinguish gas-bearing and water-saturated layers. Comparing with actual drilling results, we conclude that our PP/PS interpretation technique is feasible for fluid identification in deep carbonate reservoirs.

Adaptive parameter-related implementation of the free surface in elastic anisotropic full-waveform modeling

The near-surface anisotropy is one of the factors that lead to errors in static corrections, seismic imaging, and inaccurate surface velocity building. Therefore, it is important to study the propagation features of seismic waves in the near-surface with anisotropy. The finite-difference method is commonly used in seismic wave forward modeling, however, the numerical implementation of the free surface is nontrivial in this method, in particular to its implementation in the anisotropic media which is less studied. Here, we propose an adaptive parameter-related free surface implementation. This method is developed from the physical viewpoint with the average medium theory and limit theorem, in which the stress-free condition is implemented by modifications of the original constitutive relation and density at the free-surface points without additional computational requirements. Numerical experiments with the increasing complexity of anisotropy in both 2D and 3D demonstrate that this method can produce precise simulation results with a spatial sampling of 10 grid points per minimum wavelength. In addition, a time-space discontinues grid strategy can be introduced in this implementation to allow the usage of the coarse grid in the deeper part where the oversampling for the surface waves is not required, which increases the flexibility and the computational efficiency of the proposed method.

Interpretation method of tight sandstone based on seismic forward modeling: a case study of Permian in Southeast Ordos Basin

Based on Permian data and the reflection coefficient convolved 0° and 90° phase ricker wavelets form different phases of synthetic seismic sections. This paper theoretically discusses three aspects: one is Permian sandstone thickness and seismic amplitude, the second is tight sandstone (tuning) thickness and seismic wave frequency, and the last is tight sandstone and seismic wave phase. The result shows that when sandstone thickness is greater than seismic wavelength λ/4, the sandstone thickness can be determined through the time difference between the sandstone’s top and bottom Peaks and troughs. When sandstone thickness is less than λ/4, only the seismic reflection amplitude of the top surface of sandstone was used to determine the thickness of sandstone. In the 0° phase seismic model, peaks were located on the thin layer’s upper part and trough in the lower half, with no correspondence between the polar and lithology. In the 90° phase synthesis record, the thin sandstone layer roughly corresponds to the seismic reflection trough. 0° and 90° phase synthesis have the exact vertical resolution; the frequency variation can reflect sand body distribution. The optimal frequency makes the thickest layer of sandstone reach amplitude-frequency tuning (tuning frequency).

Analysis of AVAZ Seismic Forward Modeling of Fracture-Cavity Reservoirs of the Dengying Formation, Central Sichuan Basin

For the purpose of clarifying the seismic response characteristics of fractured-cavity reservoirs of Dengying Formation in the central Sichuan Basin, the paper first intends to establish three geological models of fracture cave reservoirs based on drilling, logging, and core data of the Dengying Formation in the central Sichuan Basin. Then, the formation reflection is calculated with reference to anisotropic Horizontal Transverse Isotropy (HTI) medium. Finally, further research on Amplitude Variation with Azimuth (AVAZ) seismic forward modeling has been conducted to clarify the seismic response characteristics of different reservoir types in the study area. The results suggest that: Seismic response characteristics of fractured-cavity reservoirs are controlled by incident angle and azimuth angle of seismic waves in different types of reservoirs. The incident angle of the seismic wave controls the difference in amplitude caused by different micro-fracture densities, and the azimuth angle controls the identification ability of the micro-fracture direction. The increase in incident angle brings about a gradual decline in amplitude. The magnitude reaches the highest when the azimuth is parallel to the normal direction of the fracture surface; however, it’ll come down to the lowest as the azimuth is perpendicular to the normal direction of the fracture surface. The fracture density fails to affect the amplitude as long as the azimuth angle is parallel to the direction of the fracture. However, the decreased amplitude reflects the increasing fracture density as the azimuth angle is identical to the normal direction of the fracture surface. The comparison between the theoretical model of three different types of fractured-cavity reservoirs and the actual uphole trace shows that the model has high accuracy. The prospect of seismic identification of fractured-cavern reservoirs, based on the results, can provide us with feasible and applicable evidence for future research on seismic identification of reservoirs and prediction of fracture distribution in the Dengying Formation of central Sichuan.

The roles and seismic expressions of turbidites and mass transport deposits using stratigraphic forward modeling and seismic forward modeling

• Mass transport is more efficient than turbidity current in forming submarine canyons . • Turbidites are inside the river-fed systems as jigsaw puzzle geometries vertically. • MTDs are not constrained by river-fed systems and tend to occur in the bottom. • MTDs have chaotic or transparent seismic facies with irregular, massive thick shapes. • Turbidites have high amplitude seismic facies with sheet, mound or channel shapes. Turbidity currents and mass transport are two principal processes in deepwater settings. However, their roles in shaping deepwater depositional systems and interpreting their deposits in seismic profiles have not been fully settled due to the lack of extensive well data and high-quality seismic data in comparison with onshore oil fields. Therefore, this study integrated stratigraphic forward modeling (SFM) and seismic forward modeling to differentiate between the contributions of turbidity currents and mass transport as well as their seismic expressions. The workflow firstly compared three single-scenario SFM models, Model A (mass transport active), Model B (turbidity currents active), Model C (both active) to explore their contributions and interplay. Secondly, a multi-process and multi-scenario SFM model, Model D, is discussed with special emphasis on canyons. Thirdly, synthetic seismic profiles are generated via seismic forward modeling using the outcomes of the second step, then compared with actual seismic facies to verify the reliability. Through the abovementioned three steps, this study reveals that mass transport plays a major role in initiating canyons whereas turbidity currents act mainly as the reworking process. Turbidites are usually confined to shelf-incised canyons and adjacent canyons fed by terrestrial systems. Mass transport deposits (MTDs) are mainly distributed on the bottom of canyons and deep basins, forming the base boundary of each depositional cycle. A total of 14 synthetic seismic facies are identified. The comparison with actual seismic data shows that modeling results are similar to equivalent actual seismic features. This study helps to predict the spatial emplacement of MTDs and turbidites and interpret deepwater seismic data.

3D outcrop geologic modeling and seismic forward modeling of mound-beach complexes

3D outcrop geologic modeling and seismic forward modeling of mound-beach complexes

Two-stage semi-supervised learning inversion for reservoir physical parameters

Reservoir physical parameters are crucial indexes for reservoir description. Deep learning is a data-driven algorithm, which may avoid some issues in conventional deterministic and stochastic inversion methods, such as cumulative and approximate errors. We develop a two-stage semi-supervised learning inversion method to predict porosity, shale content, and water saturation. Firstly, an inversion network is designed to predict physical parameters from pre-stack seismic data and initial models. The Spatio-temporal characteristics of seismic data and the low-frequency structural information of the initial model can be utilized efficiently by the network. In addition, considering the high production cost of label data in exploration geophysics, we construct a semi-supervised learning inversion workflow to relieve the dependence on the label in training. The rock physics model and seismic convolution forward model are invoked as a forward process in the workflow. Finally, we propose a two-stage inversion strategy to deal with the inversion of physical parameters with different data characteristics. The proposed method is applied to a field survey successfully. A high-resolution and high-accuracy inversion result can be obtained with our method, and potential locations of the gas reservoir can be indicated by prediction results. Meanwhile, the comparisons of the two stages demonstrate that illusions in predicted water saturation can be eliminated effectively.

Rock Physical Modeling of Tight Sandstones Based on Digital Rocks and Reservoir Porosity Prediction From Seismic Data

Digital rock physics (DRP) has become an important tool to analyze the characteristics of pore structures and minerals and reveal the relationships between microscopic structures and the physical properties of reservoirs. However, it is greatly difficult to upscale the rock physical parameters, such as P-wave velocity, S-wave velocity, and elastic moduli, from DRP to large-scale boreholes and reservoirs. On the other hand, theoretical rock physical modeling can establish the internal relationship between the elastic properties and physical parameters of tight sandstones, which provides a theoretical basis for seismic inversion and seismic forward modeling. Therefore, the combination of digital rock physics and rock physical modeling can guide the identification and evaluation of the gas reservoir’s “sweet spot.” In this study, the CT images are used to analyze the mineral and pore characteristics. After that, the V-R-H model is used to calculate the equivalent elastic moduli of rocks containing only the mineral matrix, and then, the differential equivalent medium (DEM) model is used to obtain the elastic moduli of dry rocks containing minerals and pores. Subsequently, the homogeneous saturation model is used to fill the fluids in the pores and the Gassmann equation is used to calculate the equivalent elastic moduli of the saturated rock of tight sandstones. Rock physical modeling is calibrated, and the reliability of the rock physical model is verified by comparing those with the logging data. Afterward, the empirical relationship of rock porosity established from CT images and rock elastic moduli is obtained, and then, the elastic parameters obtained by seismic data inversion are converted into porosity parameters by using this empirical relationship. Finally, the porosity prediction of large-scale reservoirs in the study area is realized to figure out the distribution of gas reservoirs with high porosity. The results show that the H3b and H3c sections of the study area exhibit higher porosity than H3a. For the H3b reservoir, the northeast and middle areas of the gas field are potential targets since their porosity is larger than that of others, from 10% to 20%. Because of the effects of the provenance from the east direction, the southeast region of the H3c reservoir exhibits higher porosity than others.