Subsurface Model Research Articles

Minimum acceptance criteria for subsurface scenario-based uncertainty models from single image generative adversarial networks (SinGAN)

Minimum acceptance criteria for subsurface scenario-based uncertainty models from single image generative adversarial networks (SinGAN)

MODELING OF THREE-DIMENSIONAL SUBSURFACE STRUCTURES BASED ON GRAVITY ANOMALY IN SOUTHWEST SUMBA INDONESIA

Modeling subsurface conditions using gravity anomaly data, focusing on density contrasts, provides critical insights into subsurface structures and supports identifying rock types. This study aims to define residual gravity anomalies in the Sumba region and utilize them to develop a three-dimensional subsurface model of Southwest Sumba, characterizing density contrasts and associated rock formations. Gravity data from the TOPEX dataset were employed in this research. The Airy isostasy model was applied to separate regional and residual anomalies, followed by a three-dimensional inversion using the Generalized Cross-Validation (GCV) method. The results reveal residual gravity anomalies range from -170 mGal to 211 mGal, with the Java Trench exhibiting the highest anomaly. The 3D modeling shows a relatively homogeneous density contrast at shallow depths, transitioning to more erratic variations at greater depths, extending to 15 km beneath Southwest Sumba Island. Furthermore, the calculated densities are consistent with the region's known geological background. The Java Trench, located south of Sumba, notably demonstrates a consistently high-density contrast from shallow to deeper depths, highlighting its tectonic complexity.

Soil Texture‐Based Parameterisation and Hydrological Insights of a Fully Coupled Surface and Subsurface Model at the Hubbard Brook Experimental Forest, USA

Soil Texture‐Based Parameterisation and Hydrological Insights of a Fully Coupled Surface and Subsurface Model at the Hubbard Brook Experimental Forest, <scp>USA</scp>

The role of subsurface models to evaluate geo-containment for safe storage of CO 2 – A case study of the Porthos CCS project in the Netherlands

3D static and dynamic models are the recommended standard to assess uncertainties in the subsurface. When focusing on CO 2 storage in depleted hydrocarbon fields, uncertainties from the potential storage resources are reasonably constrained, but uncertainties can be substantial regarding injectivity and, most importantly for safe operations, the evaluation of storage containment and fault stability. Traditional modelling workflows from the oil and gas industry can be followed to evaluate storage resources and injectivity. However, CO 2 project specific risks require different modelling workflows with additional interfaces between disciplines like geomechanics and flow assurance, which need to be linked logically to capture the subsurface process. It has been demonstrated that the operational objectives for long-term CO 2 storage can be safely achieved. The structural setting in the area of interest is well understood, and sealing integrity of the overburden sections and risk of seismic events has been assessed. Mechanisms and impact of potential CO 2 migration have been studied in detail and safeguards have been assessed. The Porthos case has proven that the applied workflow is a suitable way to get the full picture from multiple disciplines, highlighting the need for a solid initial framing and a risk-based modelling approach.

Wavenumber-aware diffusion sampling to regularize multi-parameter elastic full waveform inversion

Summary Multi-parameter elastic full waveform inversion (EFWI) provides a more realistic depiction of the subsurface models than the standard acoustic approximation. In practice, however, the significant additional cost and interdependency between the unknown parameters (cross-talks) hinder the application of such algorithms. Diffusion model-based regularization can be used to improve the inversion results while simultaneously injecting prior information into the solution. The main challenge here is how to inject such priors into the EFWI iterations that can better complement the solution’s evolution. To address this challenge, we incorporate a model wavenumber continuation process into a diffusion model-based regularization contribution to multi-parameter EFWI. To do so, we promote a sampling strategy such that at the early iteration, the proposed regularization updates account for the low wavenumber component more and increase progressively with the iteration. We first train the diffusion model on elastic moduli images in an unsupervised manner and incorporate the trained model during the EFWI inversion. We deliberately use single-component measurements, which is the most common acquisition scenario, during the inversion to demonstrate the effectiveness of our regularization. At the inference stage, the proposed framework provides more accurate solutions with negligible additional computational cost compared to several conventional regularization algorithms.

Combining crosshole and reflection borehole ground-penetrating radar (GPR) for imaging controlled freezing in shallow aquifers

Abstract. During test operation of a geological latent heat storage system as a potential option in the context of heat supply for heating and cooling demand, part of a shallow Quaternary glacial aquifer was frozen at the TestUM test site. In order to evaluate the current thermal state in the subsurface, the dimension of the frozen volume has to be known. As the target is too deep for high-resolution imaging from the surface, the use of borehole ground-penetrating radar (GPR) is being investigated. For imaging and monitoring of a vertical freeze–thaw boundary, crosshole zero-offset and reflection borehole GPR measurements are applied. The freezing can be imaged in the zero-offset profiles (ZOPs), but the determination of ice body size is ambiguous because of the lack of velocity information in the frozen sediment. Reflection borehole GPR measurements are able to accurately image the position of the freezing boundary through repeated measurements of ±0.1 m, relying on the velocity information from ZOPs. We have found that the complementary use of ZOPs and reflection measurements provides a fast and simple method to image freezing in geological latent heat storage systems. The presence of superimposed reflections from other observation wells and the low signal-to-noise ratio are problematic. The use in multiple observation wells allows an estimation of ice body size. A velocity model derived from multiple ZOPs enabled us to extrapolate geological information from direct-push-based logging and sediment cores to a refined subsurface model.

ANALYSIS OF 2D SEISMIC EXPLORATION WORKS PERFORMED IN THE GEOLOGICALLY COMPLEX STRUCTURES OF THE BAKU ARCHIPELAGO

The article focuses on the assessment of the geologically complex structures of Baku archipelago, mainly Sangachal-deniz, Duvanni-deniz, and Khara-Zira anticlines which are located in the northern part of Baku archipelago. These complex geological structures were influenced by tectonic activities, including faults and mud volcanoes which were revealed by conducted seismic exploration from the 1950s to the present. Seismic attribute analysis, specifically Sweetness and Variance attributes were applied and the results were analyzed in order to interpret these features. These attributes allowed us to characterize the lithofacial changes and to identify fault systems and chaotic seismic patterns caused by volcanic activities. The geological evaluation of the 2D seismic exploration works carried out in the Sangachal-deniz-Duvanni-deniz-Khara-Zira area was performed using "IHS Kingdom Suite" and "Petrel" software packages. The integration of 2D seismic data with well and core data provides valuable insights into the reservoir properties, highlighting the potential for hydrocarbon exploration in this geologically complex area. The study underscores the importance of continued seismic exploration to refine subsurface models and identify hydrocarbon reservoirs. Keywords: 2D seismic, geological assessment, seismic attributes, seismic processing, anticline, fault detection

Efficient and generalizable nested Fourier-DeepONet for three-dimensional geological carbon sequestration

Geological carbon sequestration (GCS) involves injecting CO 2 into subsurface geological formations for permanent storage. Numerical simulations could guide decisions in GCS projects by predicting CO 2 migration pathways and the pressure distribution in storage formation. However, these simulations are often computationally expensive due to highly coupled physics and large spatial-temporal simulation domains. Surrogate modelling with data-driven machine learning has become a promising alternative to accelerate physics-based simulations. Among these, the Fourier neural operator (FNO) has been applied to three-dimensional synthetic subsurface models. Despite its good accuracy in simulating CO 2 plume migration, it requires large computational resources in training and also lacks generalizability. Here, to further improve performance, we have developed a nested Fourier-DeepONet by combining the expressiveness of the FNO with the modularity of a deep operator network (DeepONet). This new framework is twice as efficient as a nested FNO for training and has at least 80% lower GPU memory requirement due to its flexibility to treat temporal coordinates separately. These performance improvements are achieved without compromising prediction accuracy. In addition, the generalization and extrapolation ability of nested Fourier-DeepONet beyond the training range has been thoroughly evaluated. Nested Fourier-DeepONet outperformed the nested FNO for extrapolation in time with more than 50% reduced error. It also exhibited good extrapolation accuracy beyond the training range in terms of reservoir properties, number of wells, and injection rate.

Integrating geomechanical proxy models with data assimilation for energy transition applications

This study presents a method to address the significant uncertainties in subsurface modeling that impact the efficiency of energy transition applications such as geothermal energy extraction and CO2 geological storage. The approach combines a physics-based geomechanical proxy model with an ensemble smoother with multiple data assimilation (ES-MDA), aimed at enhancing uncertainty quantification through the integration of vertical displacement measurements from fluid production and injection. The data from wells is limited in spatial coverage, while these measurements offer extensive spatial information, improving the understanding of subsurface behavior by reflecting changes in reservoir pressure and temperature. The open-DARTS simulator for fluid flow and a geomechanical proxy are used to perform data assimilation with ES-MDA. By generating an ensemble of model realizations with varied permeability, calculating vertical displacements at the surface, and applying ES-MDA, we effectively identify the probability distribution of the vertical displacement of the model conditioned to observed subsidence data. Entropy is used as a statistical measure to quantify the reduction of uncertainty of subsurface models based on observations. Our approach was tested on a 2D conceptual and 3D realistic datasets, demonstrating its capability to provide data assimilation. This workflow represents an advancement in subsurface modeling, supporting informed decision-making in geothermal energy production and CO2 sequestration by offering an improved alternative for data assimilation and enhancing tools for uncertainty quantification.

Numerical experiments on hydrogen storage characterization in porous media using elastic full-waveform inversion

In recent years, underground hydrogen storage (UHS) has garnered significant attention as a concept for large-scale energy storage. Developing hydrogen characterization and monitoring methods specific to each target geologic formation (e.g., salt caverns, reservoirs in depleted oil/gas fields, and saline aquifers in porous media) is crucial for the safe operation of hydrogen storage. However, due to the limited number of actual operations and laboratory experiments (e.g., rock physics for hydrogen) related to UHS in porous media, the specifications required for seismic monitoring (e.g., P-/S-wave velocity and density changes generated by hydrogen injection) are not yet fully understood. A recent study showed the results of laboratory experiments on cyclic UHS in porous media. The experiments demonstrated that P-wave velocity decreases nonlinearly with increasing hydrogen saturation in sandstone specimens, with a reduction of up to 3.5% observed at 36% hydrogen saturation. This study presents hydrogen characterization in porous media with conditions similar to the experiment using elastic full-waveform inversion (FWI) for time-lapse surface seismic data. The shot data are generated by seismic forward modeling for several synthetic subsurface models that incorporate a hydrogen plume. The hydrogen plume assumes four velocity-density scenarios, partially based on the laboratory study results. Numerical experiments are conducted to explore the validity and challenges of hydrogen characterization using elastic FWI.

Subsurface Modelling of the Progo River Area Using Euler Deconvolution of Topex Gravity Data

Subsurface Modelling of the Progo River Area Using Euler Deconvolution of Topex Gravity Data

Enhancing aquifer protective capacity prediction over Ibeator and environ, Southeastern Nigeria using artificial neural networks and multivariate linear regression analysis

Aquifer protection is essential for securing a sustainable supply of clean water. This study integrates an artificial neural network (ANN) model, identifying non-linear connections, with multivariate linear regression (MLR) analysis to improve predictions of aquifer protective capacity and assess vulnerability. Twelve vertical electrical soundings (VES) were conducted with a maximum electrode spacing of 250 m. Aquifer parameters derived from the VES dataset were analyzed using ANN to capture complex patterns. The ANN model, trained on historical data, learned the relationship between input variables and protective capacity. MLR analysis identified influential factors affecting vulnerability. Results reveal varying aquifer depths, with Umudime being the deepest and western parts having the shallowest depths. The resistivity map shows high values around Okorobi and Uhuala and low values in eastern to northeastern parts. Hydraulic conductivity and 3D subsurface models exhibit an inverse relationship with resistivity. Transmissivity and storativity maps exhibit similar patterns. MLR outperforms ANN in predicting resistivity, transmissivity, and storability, indicating high forecasting accuracy for aquifer protective capacity. Input parameters' contribution levels follow a specific order for different aquifer properties. R2 Value 0.0869, indicating a weak correlation between the predicted and actual values in ANN model while R2 Value 0.9775 in MLR model shows a strong correlation and much better performance than the ANN model. The results of the modeling suggest that both the ANN and MLR models have shown promising effectiveness and accuracy in predicting aquifer parameters, aiding decision-makers in implementing targeted protection measures, predicting aquifer parameters, providing insights for effective management strategies.

Comprehensive seismic refraction analysis and 2-D geological modeling in Ini Local Government Area, Akwa Ibom State, Nigeria

Seismic refraction methods are vital in geophysical investigations to reveal subsurface geological structures. This study focuses on the Ini Local Government Area, Akwa Ibom State, Nigeria, analyzing geological features beneath the surface, with emphasis on seismic velocities and layer thicknesses. The primary aim was to determine subsurface composition, identify geological disruptions, and develop 2-D subsurface models using seismic refraction data. Data were collected from forty locations using a Geometric ES 3000 12-Channel Seismograph, a Vulcan Sledge Hammer as the energy source, and geophones as detectors. The data were processed into Two-Way Time (TWT) graphs, and models were generated using strata modeling software. Three primary layers were identified: Layer 1 (topsoil) with lower seismic velocities (209-500m/s) and shallow depths (1-3m), Layer 2 (consolidated subsoil) with higher velocities (221-1210m/s) and greater depths (4-9m), and Layer 3 (bedrock) with the highest velocities (510-1700m/s) and depths (4.5-14.5m). Variations in seismic velocities and thicknesses highlighted geological instabilities. The VP/VS and Poisson’s ratios provided insights into lithological variations. 2-D sections showed lateral variations and models revealed disruptions like faults. The study delineated three seismic layers, highlighting areas with suitable and unsuitable conditions for construction. This research’s novelty lies in integrating seismic refraction data with advanced 2-D modeling, offering a comprehensive subsurface analysis crucial for construction planning and geological hazard assessment in Ini Local Government Area (LGA).

Integrated Subsurface Hydrologic Modeling for Agricultural Management Using HYDRUS and UZF Package Coupled with MODFLOW

The present work aims to compare two different subsurface hydrological models, namely HYDRUS and MODFLOW UZF package, in terms of groundwater recharge; thus, both models were coupled with MODFLOW. The study area is an experimental kiwifruit orchard located in the Arta plain in the Epirus region of Greece. A novel conceptual framework is introduced in order to (i) use in situ and laboratory measurements to estimate parameter values for both sub-surface flow models; (ii) couple the developed models with MODFLOW to estimate groundwater recharge; and (iii) compare and evaluate the performance of both approaches, with differences stemming from the distinctive equations describing the flow in the unsaturated zone. Detailed soil investigation was conducted in two soil horizons in the research field to identify soil texture zones, along with infiltration experiments implementing both double-ring and single-ring infiltrometers. The results of the field measurements indicate that fine-textured soils are predominant within the field, affecting several hydrological processes, such as infiltration, drainage, and root water uptake. Field measurements were incorporated in unsaturated zone flow modeling and the infiltration fluxes were simulated with the application of both the UZF package of MODFLOW and the HYDRUS code. The two codes presented acceptable agreement between the simulated and observed hydraulic head values with a similar performance in terms of statistics; however, they produced different results regarding recharge rates in the aquifer as simulated by MODFLOW. HYDRUS produced higher hydraulic head values in the aquifer throughout the simulation, related to higher recharge rates arising from the root water uptake and the capillary effects that are computed by HYDRUS but neglected by the UZF package of MODFLOW.

Regularization by double complementary priors for full waveform inversion

Full-waveform inversion (FWI) represents an advanced geophysical imaging technique focused on intricately depicting subsurface physical properties by iteratively minimizing the differences between the simulated and observed seismograms. Unfortunately, the conventional FWI utilizing a least-squares loss function suffers from various drawbacks, including the challenge of local minima and the necessity for human intervention in parameter fine-tuning. It is particularly problematic when handling noisy data and inadequate initial models. Recent works have exhibited promising performance in two-dimensional FWI by integrating structural sparse representation to procure adaptive dictionaries. Drawing inspiration from the competitiveness of structural sparse representation, we introduce a paradigm of group sparse residuals that integrates two types of complementary prior information by harnessing both the internal and external subsurface media models. The proposed algorithm is based on an alternate minimization algorithm to guarantee workflow flexibility and efficient optimization capabilities. We experimentally validate our method for two baseline geological models, and a comparison of the results demonstrates that the proposed algorithm faithfully recovers the velocity models and consistently outperforms other traditional or learning-based algorithms. A further benefit from the group sparse coding used in this method is that it reduces the sensitivity to data noise.

Development of a 3D Geologic Model Used in the Seismic Hazard and Liquefaction Hazard Analysis of Madison County, Tennessee

ABSTRACT Madison County of western Tennessee is approximately 100 km southeast of the New Madrid seismic zone and, thus, is subject to potentially dangerous seismic shaking and liquefaction events. Holocene floodplain alluvium and associated Pliocene/Pleistocene terraces of the Forked Deer and Hatchie Rivers cover a large part of Madison County. Underlying these unconsolidated sediments are approximately ∼650 m of poorly consolidated Paleogene and Cretaceous terrestrial and marine sediments, which are underlain by Paleozoic limestone bedrock. Uniform regional geology used in the 2014 United States Geological Survey seismic hazard maps assumes a homogenous seismic velocity structure above bedrock. Three-dimensional geologic mapping of the ∼700 m of Quaternary, Neogene, Paleogene, and Cretaceous sediments conducted in this study provides a more detailed velocity structure, resulting in a more accurate and detailed map of the expected ground motions in Madison County during future earthquakes. In addition, the surface geologic map reveals regions that are susceptible to liquefaction. This paper discusses the methods used to construct the surface geology map, subsurface structure contour maps, isopach maps, and 3D geologic model used in the seismic hazard and liquefaction hazard analyses of Madison County, Tennessee, and summaries of the results of these analyses are presented.

Improving the Parameterization of Complex Subsurface Flow Properties With Style‐Based Generative Adversarial Network (StyleGAN)

AbstractRepresenting and preserving complex (non‐Gaussian) spatial patterns in aquifer flow properties during model calibration are challenging. Conventional parameterization methods that rely on linear/Gaussian assumptions are not suitable for representation of property maps with more complex spatial patterns. Deep learning techniques, such as Variational Autoencoders (VAE) and Generative Adversarial Networks (GAN), have recently been proposed to address this difficulty by learning complex spatial patterns from prior training images and synthesizing similar realizations using low‐dimensional latent variables with Gaussian distributions. The resulting Gaussian latent variables lend themselves to calibration with the ensemble Kalman filter‐based updating schemes that are suitable for parameters with Gaussian distribution. Despite their superior performance in generating complex spatial patterns, these generative models may not provide desirable properties that are needed for parameterization of model calibration problems, including robustness, smoothness in the latent domain, and reconstruction fidelity. This paper introduces the second generation of style‐based Generative Adversarial Networks (StyleGAN) for parameterization of complex subsurface flow properties and compares its model calibration properties and performance with those of the convolutional VAE and GAN architectures. Numerical experiments involving model calibration with the Ensemble Smoother with Multiple Data Assimilation (ES‐MDA) in single‐phase and two‐phase fluid flow examples are used to assess the capabilities and limitations of these methods. The results show that parameterization with StyleGANs provides superior performance in terms of reconstruction fidelity and flexibility, underscoring its potential for improving the representation and reconstruction of complex spatial patterns in subsurface flow model calibration problems.

Along-strike variation of structural style: Mansourabad Anticline in the Dezful Embayment, SW Iran

Understanding the governing factors that influence structural style and fault-related folding mechanisms is crucial in the Dezful Embayment, an area of ~ 60,000km2 which accounts for most of oil production from Iran. Such studies enable the subsurface kinematic modeling of structures and structural geological analysis of hydrocarbon traps. In this study, variations in geometry and folding mechanism along the strike of the Mansourabad anticline are studied through field data, 2D and 3D seismic lines interpretation and well data. The displacement-distance profile of the forelimb thrust fault indicates that the anticline is a fault propagation fold in its central and NW parts. In the SE part of the anticline, there is a north-verging detachment fold, which is opposite to the southward vergence at the NW part. Due to structural variations, the amount of slip along the NW-SE trending Behbahan Fault varies. This variation in structural style results from changes in slip along the Behbahan Fault's forelimb. This blind thrust, which trends NW-SE, extends along the entire length of the Mansourabad anticline. The variable thickness of the syn-folding sediments controlled the structural style of the anticline, which interacted with the migration of the Gachsaran Formation and the deformation of the competent rocks.

The impact of grid partitioning on the convergence efficiency of tree-based Bayesian inversion

Abstract Bayesian inversion commonly employs the rj-MCMC sampling algorithm for global search, but it is known for its low efficiency. To enhance sampling efficiency in multidimensional subspaces, advanced model parameterization methods are required. Hawkins proposed a tree-based model parameterization using wavelet basis functions, which leverages the tree structure to subdivide the subsurface model. This paper examines the impact of logarithmic interval subdivision of the subsurface and a combination of shallow uniform subdivision with deep logarithmic interval subdivision on sampling efficiency. The results indicate that the combined approach of shallow uniform subdivision and deep logarithmic interval subdivision leads to faster convergence in the inversion process.

ARTIFICIAL INTELLIGENCE FOR OPTIMIZED WELL CONTROL AND MANAGEMENT IN SUBSURFACE MODELS WITH UNPREDICTABLE GEOLOGY

A comprehensive control policy structure utilizing Artificial Intelligence (AI) is presented for closed-loop management in subsurface models. Conventional Closed-Loop Optimisation (CLO) approaches entail the iterative implementation of information assimilation, past synchronization details, and effective optimizing procedures. Information assimilation is more difficult when there is uncertainty in the geological approach and the specific model conclusions. Closed-Loop Reservoir Monitoring (CLRM) offers a control strategy that promptly correlates flow information obtained from wells, as typically accessible, to appropriate well stress configurations. The rule is characterized by time-based compression and gate-based converter sections. Learning is conducted during a preprocessing phase utilizing geological modeling derived from various geological settings. Illustrative instances of oil extraction using water insertion, utilizing both 2 and 3-dimensional geological designs, are shown. The AI-oriented technique demonstrates a 17.2% increase in Net Present Value (NPV) for 2D instances, an additional 31.5% for 3D cases compared to the effective optimization of previous models, and a 1-6.5% enhancement in NPV relative to standard CLRM. Based on the methods and variable configurations examined in this study, the controlling policy method yields a 71.34% reduction in processing expenses compared to conventional CLRM.