Explore the vast range of titles available.

Accurately inferring realistic subsurface structures poses a considerable challenge due to the impact of morphology on flow and transport behaviors. Traditional subsurface characterization relies on two primary types of data: hard data, derived from direct subsurface measurements, and soft data, encompassing remotely sensed geophysical information and its interpretation. Existing deep‐learning‐based methodologies predominantly focus on the transition from multiple observations to subsurface structures. However, implicit non‐linear correlations among diverse data sources often remain underutilized, leading to potential bias and errors. In this study, we introduce a multiple‐condition fusion network (MCF‐Net) to characterize subsurface structures based on both hard and soft data. To harness the full potential of multiple‐source subsurface observations, two distinct neural networks extract implicit features from hard and soft data. The integration of these features is achieved through multiple‐condition fusion blocks, designed to capture representative characteristics. These blocks are also adept at reconstructing heterogeneous structures and facilitating hydrological parameterization. MCF‐Net exhibits accuracy in estimating subsurface structures across various types of subsurface observations. Experimental results underscore the utility and superiority of MCF‐Net in applications of hydrogeological modeling. Key Points A novel approach for describing complex subsurface structures using sparse observations (hard data) and auxiliary variables (soft data) The proposed deep learning network is able to establish the implicit relationship among multiple observations The proposed approach can be easily extended and widely used for reservoir characterization, hydrogeophysical modeling, and other fields

This research assesses the stability of sedimentary rock slopes in Teloi, Sik, Kedah, by focusing on the mechanical properties of the rock layers and their susceptibility to weathering. Key tests include the slake durability index (SDI), elastic modulus of knocking ball (Ekb), and electrical resistivity tomography (ERT). The incorporation of electrical resistivity tomography (ERT) data through the virtual reality platform facilitates the visualization of subsurface conditions. The variability of strength characteristic of interbedded sedimentary rocks leads to the differential weathering of rock layers, which causes deterioration on the slope structure. The testing revealed significant variability in rock strength, with sandstone displaying higher durability (Id > 17.1%) and elasticity (Ekb: 0.97 to 29.31 GPa) compared to shale and siltstone, which exhibited lower durability and elasticity (Id < 2.2%, Ekb: 0.2 to 2.2 GPa). Utilizing the Wenner array setup, three distinct electrical resistivity lines were established to evaluate subsurface anomalies. The ERT profiles revealed variations in electrical resistivity among different rock types, identifying areas of weaker material, which are siltstone and shale, while high resistivity areas indicate sandstone. Kinematic analysis through the stereonet process revealed direct toppling as the primary failure mechanism, driven by the critical orientations of joint sets J1, J2, and J3. This aligns with on-site observations of hanging sandstone blocks prone to toppling failure. The findings of this research show that the slake durability index (SDI) and the elastic modulus of the knocking ball (Ekb) enhance the assessment of mechanical properties and weathering resistance of interbedded sedimentary rocks. The virtual reality platform was particularly helpful in analyzing and visualizing the sub-surface conditions and enhancing the evaluation of complex geological data. As a conclusion, this integrated method was helpful in the comprehensive geotechnical evaluation of the slopes, enabling the selection of effective stabilization measures by assessing the differential weathering of interbedded sedimentary rock and identifying potential failure zones.

We propose a self-supervised deep learning methodology based on a one-dimensional adaptation of the Convolutional Visual Transformer (CVT) model for characterizing the Earth’s subsurface using well log data. Our foundation model is pre-trained on unlabeled multivariate sensor data across diverse basins and then fine-tuned for geological formation identification in the Williston Basin. The model achieves an average F1 score of 0.94 across six key formations, demonstrating faster convergence, increased robustness to missing inputs, and improved accuracy compared to baseline models, including U-Net, XGBoost, SVM, and KNN. We validate the model in geologically distinct basins, including the Groningen gas field, confirming its generalization to other fields. The framework applies to hydrocarbon exploration, carbon storage, geothermal, and groundwater reservoir characterization, supporting scalable and transferable AI solutions for the energy transition.

Quantification of groundwater (GW) and surface water (SW) interactions is crucial for effective water resource allocation and management. Immense progress has been made in the past few decades to address the different aspects of GW–SW exchanges. These have resulted in a large volume of literature. This work reviews in detail the mechanism of interaction and the applications of different field and modelling techniques. The review of flux quantification methods identifies the streambed and the aquifer beneath as two major components affecting the interactions. It is observed that the streambed is highly idealised in modelling studies, and the significance of aquifer properties in the flux quantification is found to be less emphasised. Therefore, attempts are made to highlight the potential significance of both streambed and the aquifer properties through a 2D numerical experiment. Using a superimposed GW–SW system and appropriately grouping the system parameters (as hydraulic and geometric), the experiment shows that the aquifer properties can dominate exchanging flux under certain conditions, e.g., at higher streambed conductance. The work provides suggestions to modify the widely used Darcy’s approach to include aquifer properties.

Subsurface characterization is essential for a successful infrastructure design and construction. This paper demonstrates the use of an instrumented cone penetrometer (ICP) for a dense layer characterization at two sites. The ICP consists of a cone tip and rods equipped with an accelerometer and four strain gauges, which allow dynamic driving, in addition to quasi-static pushing of the cone. The force and velocity of the cone are measured using the ICP instrumentation and compared with the N value, dynamic cone penetration index, and static cone resistance. A strong correlation has been observed between the total cone resistance estimated from the ICP and the dynamic cone penetration index and static cone resistance. After the correction of the dynamic cone resistance effect, the static component of the total cone resistance can be used as an alternative to a static cone resistance. This novel approach of soil resistance estimation using the ICP may be useful for dense layer characterization.

We develop a novel method of parameterization for spatial hydraulic property characterization to mitigate the challenges associated with the nonlinear inverse problem of subsurface flow model calibration. The parameterization is performed by the projection of the estimable hydraulic property field onto an orthonormal basis derived from the grid connectivity structure. The basis functions represent the modal shapes or harmonics of the grid, are defined by a modal frequency, and converge to special cases of the discrete Fourier series under certain grid geometries and boundary assumptions; therefore, hydraulic property updates are performed in the spectral domain and merge with Fourier analysis in ideal cases. Dependence on the grid alone implies that the basis may characterize any grid geometry, including corner point and unstructured, is model independent, and is constructed off‐line and only once prior to flow data assimilation. We apply the parameterization in an adaptive multiscale model calibration workflow for three subsurface flow models. Several different grid geometries are considered. In each case the prior hydraulic property model is updated using a parameterized multiplier field that is superimposed onto the grid and assigned an initial value of unity at each cell. The special case corresponding to a constant multiplier is always applied through the constant basis function. Higher modes are adaptively employed during minimization of data misfit to resolve multiscale heterogeneity in the geomodel. The parameterization demonstrates selective updating of heterogeneity at locations and spatial scales sensitive to the available data, otherwise leaving the prior model unchanged as desired. Key Points Novel heterogeneity parameterization for groundwater model calibration Parameterization dependence upon grid connectivity information alone Adaptive multiscale parameter estimation balancing model and data resolution

Tracer technologies based on naturally occurring substances or intentionally introduced compounds have a broad spectrum of applications in hydrogeological research and subsurface resource management. DNA (deoxyribonucleic acid)-based tracers, with unlimited unique variations and exceptional specificity, could potentially map the complex intricacies of subsurface flow networks in greater detail than traditional tracer methods. Here, we review recent advances in DNA-based tracer research involving modern culture-independent (i.e., molecular) measurement techniques for subsurface/flowpath characterization purposes. The two broad categories of DNA-based tracers, i.e., synthetic and naturally occurring, are further classified into four specific types: “naked DNA”, “encapsulated DNA”, “barcoding microbial communities”, and “indicator microbial communities”. We summarize and compare the basic methodological workflows for each type of DNA-based tracer and provide an overview of research developments in the past two decades, covering both laboratory/field-scale experiments and data interpretation methods. Finally, we highlight remaining questions and challenges for each type of DNA-based tracer in terms of practicality. Future research directions are also identified, including the application of emerging DNA tracer methods to a wider range of geological formations. Fundamental characteristics of these novel tracers need to be better understood, and their applicability under a broader range of engineering scenarios requires further validation.

Accurate subsurface characterization is essential in geotechnical surveys, yet traditional soil sampling techniques often suffer from sparse data coverage and limited spatial resolution. This review explores recent advancements in Intermediate Soil Properties (ISP) interpolation as a solution to these limitations. By integrating geoprocessing tools with Multidimensional Raster Data (MRD), the proposed approach enables high‐resolution, three‐dimensional modeling of soil property gradients across diverse geotechnical environments. Key findings highlight that ISP interpolation improves the fidelity of subsurface models, supports more accurate risk assessments, and enhances the safety and efficiency of construction planning. Additionally, the method reduces dependency on dense physical sampling, offering a more cost‐effective and scalable alternative. The review further discusses the comparative performance of traditional, hybrid, and machine learning‐based interpolation methods, providing insights into method selection based on data quality and project scale. It concludes with an evaluation of current limitations, including data sparsity and validation bias, and emphasizes the need for future integration with real‐time monitoring and remote sensing technologies. Enhancing geotechnical survey accuracy: Integrating intermediate soil properties (ISP) interpolation with multidimensional raster data (MRD) for comprehensive subsurface characterization and improved project efficiency.

Accurate prediction of compressional (P-wave) and shear (S-wave) velocity is vital for structural, geomechanical, and petrophysical analyses of subsurface formations. Since velocity is typically measured as its reciprocal, slowness is the standard parameter recorded in sonic logs and is the focus of this study. This work investigates the predictive performance of three machine learning (ML) models—Artificial Neural Network (ANN), Support Vector Regression (SVR), and Adaboost regressor—using well-log data of varying sizes. The dataset includes true resistivity (RT), density (RHOB), neutron porosity (NPHI), gamma-ray (GR), P-wave slowness (DTC), S-wave slowness (DTS), and photoelectric (PEF) logs from nineteen wells in the Sarvak Formation, a well-known carbonate reservoir in southern Iran. Results indicate that ANN consistently outperformed SVR and Adaboost in DTS prediction, achieving the highest accuracy with R 2 values of 0.8956–0.9192 and RMSE values of 2.752–2.352 µs/ft when using DTC, NPHI, GR, RHOB, and PEF as input features for a dataset containing 3654 data points from three wells. SVR demonstrated competitive performance for larger datasets (R 2  = 0.9185, RMSE = 2.497 µs/ft) but showed high sensitivity to dataset size, underperforming with limited data. Adaboost, while improving steadily with increasing data, remained the least accurate (R 2  = 0.875, RMSE = 3.0 µs/ft). For DTC prediction, ANN again showed superior accuracy, with R 2 values ranging from 0.798 to 0.9101 and RMSE decreasing from 2.82 µs/ft (2,896 data points) to 1.841 µs/ft (16,550 data points), demonstrating the positive impact of data availability on model accuracy. SVR exhibited competitive performance for larger datasets (R 2  = 0.798–0.9101, RMSE = 3.442–2.171 µs/ft), while Adaboost remained the least effective across all cases (R 2  < 0.88, RMSE = 3.546–2.461 µs/ft). A threshold of ~ 12,650 data points was identified, beyond which additional data yielded diminishing returns in model performance. Additionally, incorporating optimal input parameters, such as PEF and RT logs, significantly improved prediction accuracy, while less critical features (e.g., resistivity in DTS prediction) contributed minimally. These findings highlight the critical role of dataset size, input parameter selection, and model choice in optimizing ML applications for subsurface characterization, providing practical guidelines for future geophysical studies.