Understanding and predicting complex environmental and hydrosystem processes is a central challenge in Earth system science. These systems are governed by interacting physical mechanisms across scales, are only partially observed, and are often characterized by limited data and substantial uncertainty. As a result, machine learning (ML) has emerged along two complementary development paths for environmental modeling.

In a first branch, physical process models remain the backbone of simulation, while ML is employed as a surrogate to approximate expensive numerical solvers. Surrogate modeling approaches based on Gaussian process emulators, polynomial chaos expansions, support vector regression, and related probabilistic representations are particularly well suited for data-poor settings. Neural networks are used more selectively in this context, as uncertainty-aware and sample-efficient methods are often preferred. In surrogate modeling, considerable effort is devoted to optimal sampling strategies, including active learning, which adaptively select informative simulations and help preserve scarce computational resources. These surrogate models enable efficient uncertainty quantification, sensitivity analysis, and Bayesian inference, while preserving physical interpretability.

A second, increasingly important branch emerges when physical models are incomplete, unavailable, or deliberately omitted, and ML models replace the governing equations altogether. This branch is most commonly based on neural network representations, but has recently also been explored using Gaussian processes and polynomial chaos–based learning concepts. In this setting, purely data-driven learning is insufficient, as unconstrained models tend to violate physical principles and extrapolate poorly. In this second branch, physical principles such as conservation laws or balance relations are embedded directly into learning architectures. Complementarily, constraint-driven learning strategies enforce physical laws, admissibility conditions, and structural consistency during training. By restricting the hypothesis space, these methods stabilize learning and support robust inference under incomplete physical knowledge.

Taken together, surrogate modeling for physics-based simulations and physics-aware ML for equation-free learning represent two coherent and complementary branches of modern environmental machine learning. We observe a growing convergence between these two branches, as physics-based surrogate modeling and equation-free machine learning increasingly borrow concepts from each other. This convergence is not accidental, but a direct response to fundamental model limitations and the challenge of making reliable predictions under scarce data and knowledge constraints. By integrating physics, probabilistic reasoning and constraints, emerging approaches increasingly focus on robustness and interpretability rather than unconstrained expressive power.

How to cite: Oladyshkin, S.: Physics-Aware Learning for Environmental Systems: Surrogate Modeling and Constraint-Driven Machine Intelligence, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3804, https://doi.org/10.5194/egusphere-egu26-3804, 2026.

Physics-Informed Neural Networks (PINNs) have emerged as a promising paradigm for solving problems governed by partial differential equations (PDEs) using the flexibility and generalization capability of deep learning. By embedding the governing physical laws directly into the training process, PINNs can approximate complex physical systems even when limited or no observational data are available. However, their performance and convergence can deteriorate significantly in domains characterized by high heterogeneity or discontinuities in material properties. In particular, standard PINN formulations tend to enforce implicit continuity in the hydraulic conductivity field, which can lead to inaccurate representations of physical processes in heterogeneous porous media.

This study introduces a novel and robust PINN framework for modelling transient fluid flow in heterogeneous porous media, with specific emphasis on accurately handling discontinuities in the hydraulic conductivity field. The proposed approach is based on a mixed formulation of the governing flow equations, in which the pressure and velocity fields are represented by independent neural networks. This structural separation eliminates the need to compute spatial derivatives of discontinuous or non-differentiable quantities during the evaluation of the loss function. As a result, the method achieves a more stable and accurate application of automatic differentiation while maintaining strong adherence to the underlying physical principles.

Furthermore, to address the high computational cost typically associated with training PINNs, a discrete-time mixed formulation is developed. By discretizing the temporal domain, this approach reduces the dimensionality of the problem, leading to substantial savings in both memory usage and training time. Despite these efficiency gains, the discrete-time PINN retains a high level of accuracy and fidelity in predicting transient flow dynamics in heterogeneous domains.

Comprehensive testing on various scenarios of unconfined aquifers demonstrate that the proposed implementation outperforms standard PINN approaches when applied to porous media with strong contrasts in hydraulic conductivity. The results obtained from the different PINNs techniques have been compared against the results from finite element software COMSOL to analyze their performance.

Overall, the mixed formulation PINN frameworks are computationally more efficient, and produce results with improved accuracy compared to the standard PINNs technique for simulating fluid flow in complex porous media systems, representing a significant step forward in the application of deep learning to subsurface modelling.

How to cite: Virupaksha, A. G., Fahs, M., Nagel, T., Lehmann, F., and Hotiet, H.: Physics informed neural networks based on mixed pressure-velocity formulation for flow in heterogeneous aquifers, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1394, https://doi.org/10.5194/egusphere-egu26-1394, 2026.

Reactive nitrate transport in groundwater is governed by coupled advection–dispersion–reaction (ADR) dynamics and kinetically limited redox processes, including donor limitation and competition among electron acceptors. We compare two surrogate modeling approaches for reactive nitrate transport. The first is a physics-audited, data-driven approach based on a categorial boosting algorithm, with physical admissibility (e.g., non-negativity and ADR-consistent behavior) assessed via post-hoc diagnostics. The second is a physics-informed neural network (PINN) surrogate that embeds the ADR equation, boundary conditions, non-negativity, and a redox-ordering constraint directly into the training objective to promote mechanistic consistency. Both surrogates are trained and tested on the same one-dimensional PHREEQC benchmark suite spanning increasing hydrogeochemical complexity: linear denitrification, dual-linear nitrate–Fe(III) competition, dual-substrate Monod kinetics, and fully coupled dual-Monod redox systems. Predictive uncertainty is quantified to provide calibrated confidence bounds and identify regions of elevated sensitivity.

Results show that while both surrogates can interpolate reactive nitrate dynamics within the training domain, the PINN surrogate consistently provides superior physical consistency and robustness under increasing kinetic nonlinearity. Uncertainty estimates from the PINN are well calibrated, with prediction-interval widths increasing systematically near migrating reactive fronts where nonlinear redox competition amplifies model sensitivity. The results demonstrate that embedding governing physics directly into the learning process yields a more reliable and interpretable surrogate for uncertainty-aware reactive transport modeling, particularly in regimes dominated by nonlinear kinetics and competing redox pathways.

How to cite: Arab, A., Scheytt, T., Nagel, T., and Taherdangkoo, R.: From Data-Driven to Physics-Informed Surrogate Models for Reactive Nitrate Transport, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11359, https://doi.org/10.5194/egusphere-egu26-11359, 2026.

Generative super-resolution (SR) reconstruction models are widely applied in digital rock research to balance the 
trade-off between image resolution and the scanning device’s field of view. Existing methods often enhance 
visual details or structural fidelity separately. However, they fail to balance these goals effectively. This failure 
frequently leads to artifacts that distort porosity and permeability measurements. This paper proposes the Sta
tionary and Discrete Wavelet-Enhanced Generative Adversarial Network (SDWGAN). The model is a hybrid SR 
approach that integrates two wavelet decomposition methods. This integration addresses the challenge effec
tively. By integrating multi-scale frequency constraints from wavelet decomposition with adversarial training 
focused on high-frequency components, our method effectively distinguishes rock boundary details from imaging 
artifacts. The proposed model adopts a global-local feature integration architecture to preserve fine-grained 
textures and macroscopic structures. Experimental results on the DeepRock-SR dataset (carbonate, sandstone, 
coal) demonstrate SDWGAN’s enhancements: 0.63–2.12 dB PSNR and 0.01–0.11 SSIM improvements in fidelity, 
alongside 0.001–0.005 LPIPS and 0.62 NIQE gains in perceptual quality over RGB-domain loss-based models. 
Simulated seepage results indicate that SDWGAN estimates porosity and permeability with 98 % similarity to the 
reference images. In conclusion, the proposed model manages the perception-distortion trade-off via frequency 
domain optimization, ensuring petrophysical consistency between SR results and benchmarks. This approach 
offers a novel and reliable method for reservoir characterization in the field of petroleum geology.

How to cite: Ma, G. and Wang, Z.: Multiscale wavelet-adversarial learning eliminates imaging artifacts in digital rock analysis for reliable reservoir evaluation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-576, https://doi.org/10.5194/egusphere-egu26-576, 2026.

This study evaluates the field applicability of a multi-resolution Convolutional Long Short-Term Memory (ConvLSTM) framework for predicting time-series retaining wall deformations during staged excavation using field measurements from various excavation sites in South Korea. The proposed framework integrates three ConvLSTM models trained with different temporal input resolutions to capture deformation characteristics at multiple time scales. Their multi-step predictions are subsequently refined using a stacking ensemble strategy with a neural network–based meta-learner, which mitigates error accumulation commonly observed in recursive long-horizon forecasting and enhances overall prediction stability and accuracy.

To generate a comprehensive training database, numerical analyses were conducted on a wide range of excavation cross-sections with varying final excavation depths, wall tip restraint conditions, and initial groundwater levels, reflecting diverse geotechnical and structural configurations. The geotechnical and structural properties were defined probabilistically to account for inherent uncertainties in ground conditions and structural stiffness. In total, 4,000 numerical analysis cases were generated and further augmented into 16,000 training datasets through Gaussian noise injection to improve model generalization ability.

For field validation, 34 time-series displacement measurements collected from 11 excavation sites in South Korea were employed to assess the predictive performance of the proposed framework under real construction conditions. When lateral displacement data obtained from earlier excavation stages were provided as inputs, the model predicted retaining wall deformation induced by an additional excavation depth of 5.0 m, achieving an average coefficient of determination (R²) of 0.85 and a mean absolute error (MAE) of 5 mm. Furthermore, the framework demonstrated an average inference time of 0.92 s, confirming its suitability for near–real-time prediction and potential integration with field monitoring systems. These results indicate that the proposed multi-resolution ensemble framework is practically applicable to real-world excavation projects and offers a robust tool for predictive decision-making in excavation safety management.

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2023R1A2C1007635).

How to cite: Kim, J. and Youn, H.: AI-Driven Time-Series Prediction of Retaining Wall Deformation: A Case Study in Korea, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3675, https://doi.org/10.5194/egusphere-egu26-3675, 2026.

Geological carbon storage is a key strategy for mitigating global CO₂ emissions, and reliable monitoring of subsurface CO₂ migration is critical for storage safety. Time-lapse seismic provides valuable insights into CO₂ plume evolution. However, accurately predicting high-resolution CO₂ saturation from seismic data remains a major challenge. In this study, we propose a novel physics-constrained deep learning framework that treats time-lapse seismic data as video sequences and leverages the Video Masked Autoencoder (VideoMAE) architecture to capture spatial and temporal dependencies. The approach consists of two stages: self-supervised pretraining on seismic data and supervised fine-tuning for CO₂ saturation prediction. During pretraining, masked reconstruction enables the model to extract rich spatiotemporal feature representations from seismic videos. In fine-tuning, the pretrained model is adapted to predict future CO₂ saturation from historical time-lapse seismic data without requiring seismic data from the target year. A physical constraint based on Fick’s law of diffusion is incorporated into the loss function to regularize the temporal evolution of CO₂ saturation during fine-tuning. Results on the Kimberlina synthetic multiphysics dataset demonstrate that the physics-constrained VideoMAE framework consistently outperforms baseline models in both prediction accuracy and spatial consistency. These findings highlight the effectiveness of combining video-based self-supervised learning with physical constraints for time-lapse seismic monitoring and provide a promising physics-informed approach for CO₂ storage surveillance.

How to cite: Tang, M., Zong, Z., and Jiang, D.: CO2 Saturation Prediction from Historical Time-Lapse Seismic Data Using Physics-Constrained VideoMAE, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6206, https://doi.org/10.5194/egusphere-egu26-6206, 2026.

Accurate prediction of the hydraulic conductivity of compacted bentonite is critical for assessing the long-term safety of high-level radioactive waste repositories, where barrier efficiency depends on coupled processes. This study develops a data-driven machine learning model to predict saturated hydraulic conductivity and a physics-based machine learning model to predict unsaturated hydraulic conductivity of compacted bentonite. For the saturated hydraulic conductivity prediction, a dataset of 215 experimental measurements was compiled, incorporating key soil properties such as montmorillonite content, specific gravity, plasticity index, initial water content, dry density, and temperature as input. To predict unsaturated hydraulic conductivity, the study considers experimental data, synthetic data generated using the Van Genuchten model, and outputs from the machine learning model developed for saturated hydraulic conductivity. The input dataset includes specific gravity, montmorillonite content, initial dry density, initial water content, initial void ratio, plasticity index, and suction. The AdaBoost, CatBoost, and XGBoost algorithms were used to train the machine learning models, and the whale optimization algorithm was used for hyperparameter tuning. The trained machine learning models demonstrate good predictive performance for both saturated and unsaturated hydraulic conductivity of compacted bentonite, showing close agreement with experimental measurements.

How to cite: Shehab, M., Taherdangkoo, R., and Christoph, B.: Physics-based and data-driven machine learning modeling of saturated and unsaturated hydraulic conductivity of bentonite, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5372, https://doi.org/10.5194/egusphere-egu26-5372, 2026.

Shallow geothermal energy represents a significant opportunity to reduce energy waste in the heating and cooling sectors. Geothermal maps are a valuable tool for enhancing the exploitation of geothermal resources on a national scale. In this work, we produced maps of Italian shallow geothermal potential for different saturation and groundwater velocity scenarios. For this purpose, we developed a method for quickly computing geothermal potential, based on lithological data and the simplified application of the Moving Infinite Line Source model for heat dispersion. Our approach follows the G.POT methodology proposed by Casasso and Sethi in 2016, but it also incorporates the contribution of groundwater flow, which was not considered in the original G.POT computation. This method enables the computation of geothermal potential from a large amount of input data, given the geological asset, the thermal and hydrogeological properties of the materials that form the subsoil, the initial underground temperature, and the required thermal load. Using this approach, we estimated the geothermal potential related to more than 28.000 sites for which stratigraphic data are available. We gather the stratigraphic logs of every site and compute the geothermal potential for each lithological layer encountered in each log. The derived values have been averaged to obtain the mean potential of the shallow subsoil at a reference depth of approximately 70 m. The final maps are the result of interpolating the point estimates. The different scenarios explore the variability of the geothermal field as it is intrinsically linked to the geological and hydrogeological context. From completely unsaturated to completely saturated conditions, the geothermal potential can increase by a factor that ranges from 4 to 10, depending on the groundwater flow velocity. The regions showing larger increments related to groundwater action are those characterized by sandy or gravelly subsoils, such as Emilia-Romagna, Piedmont, Lombardy, Friuli-Venezia Giulia, and Veneto. The high permeability of these sediments strongly influences their geothermal potential. On the other hand, areas where consolidated rock prevails are less susceptible to variation due to the presence of water in the underground soils, as observed in some regions of Sardinia, Sicily, and Apulia. Both the final maps and selected intermediate results have been published on open-access data platforms managed by Ricerca sul Sistema Energetico - RSE S.p.A, which also host a wide range of other energy-related information to support territorial energy planning.

How to cite: Squarzoni, G., Colucci, F., and Aiello, M.: Mapping the geothermal potential of Italy’s shallow subsoil: a streamlined MILS model approach for extensive datasets, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3529, https://doi.org/10.5194/egusphere-egu26-3529, 2026.

The geothermal sector generates a vast amount of knowledge—from research project data to scientific publications, technical reports, patents, and open datasets—produced by scientists, operators, consultants, public authorities, and funding agencies. However, this wealth of information is often scattered across multiple repositories and platforms, which hampers effective access, integration, and utilization. EGRISE 2.0, developed within the EU-funded Geotherm-FORA project, addresses this challenge as the largest thematic repository for geothermal research and innovation in Europe. The platform aggregates information from EU-funded projects, open access publications, scientific journals, and public datasets hosted on repositories such as Zenodo and Pangaea. Each research product is indexed with detailed metadata, enabling users to search, filter, and explore thousands of documents—currently over 11,000—by criteria such as publication type, funder, country, year, language, or resource access.

By consolidating this vast body of knowledge and facilitating its exploration, EGRISE 2.0 allows stakeholders to precisely map the state of R&D in the geothermal industry. Researchers can spot emerging trends, identify gaps, and recognize key contributors, while funding agencies and policymakers can evaluate technological maturity and set priorities for future research and investment. Additionally, the platform facilitates the preparation of innovative project proposals by offering instant access to scientific publications, datasets, and project deliverables.

A set of integrated charts further enhances the platform’s value, offering insights such as publication trends, openness over time, and data FAIRness.

EGRISE is an open tool available at https://egrise.openaire.eu/. It is powered by OpenAIRE CONNECT, a service to build customizable search portals on top of the OpenAIRE Graph, one of the largest open scientific knowledge graph.

In this way, EGRISE 2.0 not only consolidates knowledge but actively empowers innovation, collaboration, and strategic decision-making leveraging on open research information, establishing itself as an indispensable tool for the European geothermal community.

How to cite: Trumpy, E., Bardi, A., and Manzella, A.: EGRISE 2.0: the knowledge at the fingertips to power Research and Innovation in Geothermal Energy, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11715, https://doi.org/10.5194/egusphere-egu26-11715, 2026.

Former oil and gas fields offer a repository of historical geophysical well logs that can help support geothermal exploration across large areas. Lithology classification from logging data is a fundamental task in subsurface geological interpretation. Existing deep learning approaches typically formulate this problem as a point-wise or sequence-wise classification task, where logging curves are treated as one-dimensional depth-dependent signals. Although such methods have demonstrated promising performance, they usually rely on large-scale labelled datasets for training. Moreover, logging datasets commonly exhibit severe class imbalance due to complex geological environments and strong heterogeneity, which further degrades the performance and robustness of data-hungry deep learning models.

To address these challenges, we propose a novel lithology segmentation framework, in which we reformulate lithology classification as a semantic segmentation task, where different lithological units are characterized by continuous intervals separated by distinct boundaries along the depth dimension. Based on this formulation, we develop a lithology segmentation framework that leverages large-scale vision foundation models, enabling effective learning under data-scarce and class-imbalanced conditions. Our core motivation is to transfer the strong image representation and generalization capabilities learned by large pretrained models on massive image data to the geological logging domain.

Specifically, well logging curves are transformed into two-dimensional pseudo-images by a structured multi-scale channel combination along the depth dimension. The repetition factor k controls how many times each logging curve is duplicated in the pseudo-image, enabling Vision Transformer (ViT) with fix-sized patches to encode logging patterns at multiple effective scales. For each scale k, a composite representation X^(K)∈ R^H×W_K  is formed by repeating selected logging curves with scale-dependent repetition factors, where H is the number of depth samples. Accordingly, the width of the pseudo-image at scale k is defined as W_k = k·N, where N is the number of logging curves. The final input representation X is obtained by concatenating all scale-specific representations: X = Concat(X⁽¹⁾, X⁽²⁾, X⁽²⁾, ..., X^(K)).

Building upon the pretrained Segment Anything Model (SAM), we retain the image encoder to extract high-level visual features, while a task-specific decoder is initialized and trained from scratch for lithology segmentation. The encoder weights are initially frozen and gradually unfrozen during training, and fine-tuned jointly with the decoder to adapt the feature space to the geological patterns of the specific domain. This staged training strategy stabilizes the optimization process, reduces overfitting with limited data, and effectively transfers knowledge from natural images to well logging images. Furthermore, by using a weighted loss function at the segmentation level to address class imbalance, it ensures that a minority of lithological classes contribute sufficiently to model updates.

Overall, the proposed framework demonstrates a new workflow for lithology interpretation by integrating foundation models with geological data analysis. It provides a data-efficient solution for lithology segmentation under realistic constraints of limited and imbalanced well logging datasets.

How to cite: Qian, N., Jagert, F., and Sester, M.: Lithology Segmentation from Well Logs for Geothermal Exploration using Vision Foundation Models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7749, https://doi.org/10.5194/egusphere-egu26-7749, 2026.

Hot dry rock (HDR) represents a promising form of clean and renewable geothermal energy, with substantial global potential to support the transition to low-carbon energy systems. Among the most prospective regions for HDR development in China is the Gonghe Basin in Qinghai Province. However, the basin's complex subsurface geological conditions present significant challenges for the accurate assessment of HDR resources.

This study proposes a multi-scale integrated geophysical framework for HDR characterization, combining gravity, magnetic, magnetotelluric (MT), ambient noise tomography, and time-frequency electromagnetic methods. Multi-source geophysical datasets were systematically processed, calibrated with available borehole data, and interpreted through inversion modeling to construct a three-dimensional geological-geophysical model of the study area.

The results reveal the spatial distribution, burial depth, and thermal-structural properties of HDR reservoirs, identifying a high-potential zone with reservoir temperatures exceeding 200 °C. The integrated approach effectively addresses the limitations of individual geophysical methods, significantly enhancing the accuracy of HDR reservoir identification and parameter estimation. This study demonstrates the feasibility and effectiveness of integrated geophysical techniques in HDR exploration, offering a robust technical basis for future development in the Gonghe Basin and similar geothermal environments worldwide.

How to cite: lian, S., cheng, Z., wei, Q., and zhang, L.: Integrated Geophysical Characterization of Hot Dry Rock Resources in the Gonghe Basin, Qinghai, China, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7708, https://doi.org/10.5194/egusphere-egu26-7708, 2026.

Climate change drives urgent action in decarbonisation, and carbon capture and storage (CCS) has emerged as a crucial technology in mitigating greenhouse gas emissions. Large-scale subsurface CO₂ injection carries the inherent risk of inducing fault reactivation and microseismic events, which could compromise the project. To optimise CCS projects while mitigating these geological hazards, passive acoustic emission (AE) monitoring offers a real-time method to detect initial fracture activity before failure.

In this study, triaxial compression experiments were conducted on reservoir-analogue sandstone sample plugs. Intact samples were axially loaded under an initial confining pressures (Pci) with continuous passive AE recording. A shear fracture was then induced in each sample, which was subsequently re-sheared under different confining pressure regimes (Pc) to mimic fault reactivation. Two porosity groups (~20% and 26%) were tested to evaluate deformation effects on AE response. Acoustic sensors at the sample ends captured the P-wave signals throughout each loading cycle, and the AE events were analysed in conjunction with the mechanical stress-strain data. From these mechanical data, failure envelopes were derived to assess the applicability of failure criteria. The results show that the Mohr–Coulomb criterion provides good agreement with all tests conducted and that fractured specimens may exhibit friction angles different from intact rock while retaining a non-zero cohesion, which should not be neglected when modelling fractured reservoirs for CCS.

The acoustic emission results reveal clear precursor patterns to fracture slip. For intact samples, axial loading triggered intense AE activity from the outset, reflecting micro-cracking and particle rearrangement. In contrast, samples with pre-existing fractures showed an initially low rate of emissions, increasing significantly just before the peak stress. Notably, higher-porosity samples generated roughly an order of magnitude more emissions than lower-porosity samples during both the initial fracturing and the reactivation phases, and consequently a much higher cumulative acoustic energy release.

Crucially, the cumulative AE record revealed a distinct acoustic precursor to failure. During re-shearing, the cumulative event count initially increased steadily, then underwent a sudden acceleration (an identifiable inflection point) shortly before the peak stress. This surge in event rate consistently occurred when the sample was still below its peak strength, signalling imminent failure. Such a signal could serve as an early warning. In a field injection scenario, detection of this acoustic inflection would allow operators to adjust injection rates or pressures before fault reactivation. Incorporating passive AE monitoring in this way could enhance CCS safety by optimising operations and preventing induced seismicity.

How to cite: Mesquita, R., Forbes Inskip, N., Naderloo, M., Barnhoorn, A., Doster, F., and Busch, A.: Early Warning of Fault Reactivation through Passive Acoustic Emission in Samples Analogous to Carbon Storage Reservoir, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1788, https://doi.org/10.5194/egusphere-egu26-1788, 2026.

The anion-accessible porosity fraction (f_a) is an important parameter controlling solute transport in claystones. A safe disposal of nuclear waste in such rocks relies on a comprehensive understanding of transport in clays. In stratigraphic sequences, established Cl or Br profiles provide insights into the paleo-hydrogeological evolution. Anion concentrations in the accessible porewater can be calculated from measured bulk porewater concentrations and the anion-accessible porosity fraction f_a. Since experimental data for f_a are scarce, and data extrapolation within a heterogeneous stratigraphy is challenging due to the f_a dependency on multiple parameters (e.g., pore/grain shapes and size distributions), detailed understanding of texture and its influence on macroscopic transport parameters is paramount.

In this study, imaging techniques (µCT and SEM) and other methods (e.g., N₂ adsorption) were combined to characterise texture and the pore network of rock samples from Opalinus Clay and confining units. The techniques unravel pore characteristics at different scales: N₂ adsorption from nanometers to a micrometer, SEM larger than 50 nm, and µCT larger than a few µm. Samples with different mineralogical compositions, lithologies, and experimentally determined f_a for Cl (f_Cl) were analysed.

Two sand/siltstone samples with different f_Cl but similar clay content show identical ratios of grains to porous clay regions, but different pore sizes in high-resolution SEM images. This can qualitatively explain the different f_Cl for these samples. However, SEM cannot resolve small pores (<50nm), and a structural model is additionally required to derive quantitative results.

The gained textural insights make clear that f_Cl does not necessarily correlate with the clay fraction. Moreover, extended correlations of f_Cl with quantified textural information allow a better prediction of f_Cl for formations where this parameter was not measured. The outcome of this study encourages further investigations for verifications such as transmission electron microscopy (TEM) imagery to explore the nanometric pore space within and around clay minerals.

[1] C. Zwahlen, T. Gimmi, A. Jenni, M. Kiczka, M. Mazurek, L. R. Van Loon, et al., "Chloride accessible porosity fractions across the Jurassic sedimentary rocks of northern Switzerland," Appl. Geochem., vol. 162, p. 105841, 2024. DOI: 10.1016/j.apgeochem.2023.105841

How to cite: Zwahlen, C., Gimmi, T., Jenni, A., and Wüst, R.: Influence of texture on anion-accessible porosity fraction explored by µCT, SEM & N2 adsorption data , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6683, https://doi.org/10.5194/egusphere-egu26-6683, 2026.

Clays and clay rocks are relevant materials in many natural or engineered systems. Particles and pores in these materials are very small, which results in very low permeabilities. Accordingly, clays or clay rocks are considered as sealing materials or as host rocks for the safe disposal of hazardous waste in the underground.

The architecture of the pore space, i.e., the pore size distribution and the pore connectivity, are fundamental characteristics that define macroscopic properties of these materials, such as water retention function, hydraulic conductivity, diffusion coefficients, or the mechanical behavior. Unfortunately, the resolution of imaging techniques is often insufficient for a direct visualization of all pores in clays, and mostly indirect methods have to be used. Moreover, porewater close to charged clay surfaces may be partly bound, and this can also affect hydraulic conductivities.

We applied a range of different methods (Hg injection, N₂ and H₂O ad-/desorption, simultaneous thermal analysis coupled with evolved gas analysis STA-EGA) to characterize the pore space architecture and physical properties of porewater of a set of twelve very different rocks. We addressed the following questions: (1) Is porewater close to solid surfaces more strongly bound compared to porewater far from surfaces? (2) Are physical porewater properties related to basic properties of the clay rocks, such as clay-mineral content, cation exchange capacity, or the pore solution composition?

When comparing pore size distributions derived from the above methods and from NMR cryoporometry (Fleury et al., 2022), we see that mostly similar size ranges are obtained, but specific peaks should not be overinterpreted. Only NMR cryoporometry allows measurements at the original saturation state (except for high salinity solutions), which minimizes potential artefacts from drying. During STA, mainly water was released up to ~200°C (heating rate 10°C/min) in all samples. Vaporization enthalpy distributions derived from the STA data – indicators of water binding states – are unimodal in many cases, meaning that no clearly distinct water populations exist. However, the width of the distributions varied considerably among the samples. Comparably narrow distributions with a main peak in the region of bulk water vaporization enthalpies were seen for samples with relatively large pores, and wider or very wide distributions for samples with small pores, complex pore networks, higher surface charge concentration per volume of pore water, or high salinity pore solutions. The latter demonstrates that the derived vaporization enthalpies do not only reflect surface interactions, but also interactions with solutes. Finally, the partly large differences in the energetic state of the porewater should be considered as a relevant pore-scale feature when trying to derive macroscopic hydraulic parameters.

Fleury, M., T. Gimmi, M. Mazurek (2022). Porewater content, pore structure and water mobility in clays and shales from NMR methods. Clays Clay Miner. 70, 417–437, https://doi.org/10.1007/s42860-022-00195-4

How to cite: Gimmi, T., Mazurek, M., and Emmerich, K.: Pore space architecture and water binding state in clay-rich rocks, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19147, https://doi.org/10.5194/egusphere-egu26-19147, 2026.

Various types of porous media (both unconsolidated and consolidated geological bodies and engineering materials, etc.) and fluids (water, gas, oil, supercritical carbon dioxide, etc.) are closely intertwined with multiple fields such as the environment, geology, and geotechnical engineering, involving soil contamination and groundwater remediation, high-level nuclear waste disposal, carbon dioxide storage, shale oil and gas extraction, hydrogen energy storage, and geothermal utilization. Nano-petrophysical studies focus on rock properties, fluid properties, and the interaction between rocks and fluids, especially for low-permeability geological and engineering media with a large number of nano-scale pores, as their microscopic pore structure (pore size distribution, pore shape and connectivity) controls the macroscopic fluid-rock interaction and the efficient development or preservation of various energy fluids. Such a subsurface system involves a wide range of nm-μm scale pore sizes, various pore connectivity and wettability, in addition to the coupled thermal-hydraulic-mechanical-chemical (THMC) processes of deep earth environments. This work showcases the development and application of an integrated and complementary suite of nano-petrophysical characterization approaches, including pycnometry (liquid and gas), porosimetry (mercury intrusion, gas physisorption), imaging (Wood’s metal impregnation followed with field emission-scanning electron microscopy), scattering (ultra- and small-angle neutron and X-ray), and the utility of both hydrophilic and hydrophobic fluids as well as fluid invasion tests (imbibition, diffusion, vacuum saturation) followed by laser ablation-inductively coupled plasma-mass spectrometry imaging of different nm-sized tracers in porous materials. These methodologies have been extended into coupled THMC processes under reservoir-relevant setting, such as the small-angle scattering (SAS) method developed and utilized for the direct observation of rock deformation behavior at a spatial resolution of 1 nm with stresses up to 164 MPa using a self-developed high-pressure cell for mechanistic studies of fluid-solid coupling. 

Acknowledgement: This work was supported by the Basic Science Center Program of the National Natural Science Foundation of China (NSFC) (Type A; No. 42302145) and the International Cooperation Project of PetroChina (2023DQ0422).

How to cite: Hu, Q., Hao, F., Xiao, Y., Liu, K., Zhang, T., Ke, Y., Cheng, H., Li, X., Wang, Q., Zhao, C., and Yang, S.: Microscopic pore structure and macroscopic fluid flow-chemical transport with the coupled thermal-hydraulic-mechanical-chemical processes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15580, https://doi.org/10.5194/egusphere-egu26-15580, 2026.