Orals: Wed, 6 May, 10:45–12:30 | Room 3.16/17
Fractured porous media and karst systems are two heavily studied families of geological formations. Both types of systems tend to display highly heterogeneous hydrogeological properties over a wide range of pore-to-field length scales, and they both typically display irregular or non-uniform flow and transport behaviors in both space and time. While these features suggest commonalities in the mechanisms that control flow and transport dynamics, several key features distinguish these two systems and demand different perspectives on assessment and quantification. More specifically, for example, quantifying “just” the volumetric water fluxes in fractured systems – for which the host rock permeability can vary over orders of magnitude, e.g., considering sandstone to carbonate to granitic formations – first faces the obstacle of obtaining a realistic, site-specific delineation of the (3D) fracture geometry. In addition to this difficulty, karst systems can contain large cave-like features (which are not “fractures”) that strongly influence flow and even induce turbulent flow; and at the catchment scale, a critical problem is to assess total system response subject to time-dependent input (precipitation, flooding) conditions and/or output (pumping) conditions. Assessment of chemical and radionuclide transport in both fractured and karst systems, and then chemical reactions and interactions such as precipitation and dissolution, requires even more complex considerations and increased degrees of uncertainty and variability over pore-to-field spatial and temporal scales. Addressing relevant questions requires field measurements, laboratory experiments, and appropriate model formulation. Significantly, data acquisition at field scales is highly restricted and extensive field studies conducted already three to four decades ago in fractured formations – notably at Aspö and Stripa (Sweden), Yucca Mountain (USA), and Chalk River (Canada) – illustrated the severe limitations to detailed characterization, which lead to serious limitations in realistic mimicking and prediction of flow and transport behaviors. Moreover, while laboratory studies provide important insights at pore, column (generally 1D) and flow cell (generally 2D) scales, they are limited in terms of spatial and temporal scales, and realistic representation of fracture network and karst complexities. In terms of model conceptualization and development, the question of how to model flow, transport, and reactive chemical transport should thus be based, first and foremost, on the specific phenomenon or problem of interest (e.g., flow, transport, chemical reactions, local dynamics or overall system behavior, spatial and temporal scale). This then directs conceptualization and the choice of a model that focuses, principally, either on “exact” dynamics or on phenomenological or overall ”functional” dynamics. We will address both similarities and differences in fractured porous media and karst systems, suggesting methods of characterization and quantification that can be common to both, and other methods that are tailored to address the distinct features and real hydrogeological relevance between these systems.
How to cite: Berkowitz, B.: Fractured Porous Media and Karst Systems – Shared and Distinguishing Features in Conceptualization and Assessment of Flow and Transport Processes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4236, https://doi.org/10.5194/egusphere-egu26-4236, 2026.
How to cite: Lenci, A., Méheust, Y., Klepikova, M., Di Federico, V., and Tartakovsky, D.: Unraveling Anomalous Thermal Transport in Fracture–Matrix Systems: Interplay Between Advective Transport and Matrix Conduction, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-869, https://doi.org/10.5194/egusphere-egu26-869, 2026.
Diffusion in natural porous media, e.g., in soils, rocks and geological formations, is a widely observed phenomenon and is critical to many subsurface applications, such as deep nuclear waste disposal and contaminated aquifer remediation. In much of the existing literature, diffusion is considered to be effectively Fickian. However, recent experimental studies have shown that diffusion can exhibit non-Fickian behavior. To explain this behavior, and in the spirit of percolation theory, we hypothesize that non-Fickian diffusion arises from the low-connectivity nature of pore networks, even when percolating channels exist. Based on a systematic study involving a large number of particle tracking simulations in two- and three-dimensional domains, with low and high connectivity, we demonstrate that non-Fickian diffusion appears in domains nearer the percolation threshold, while it approaches Fickian behavior in high-connectivity domains. Low-connectivity domains contain primary diffusive channels as well as dead ends and even isolated pore clusters that can trap diffusive plumes over extremely long times. This leads to diffusion occurring with power-law transition time behavior. This study highlights the limitations of using purely Fickian models to characterize diffusion behavior in geological settings, as structural features such as pore network connectivity can have a significant influence.
How to cite: Yin, T. and Berkowitz, B.: Structural Controls on Solute Diffusion in Porous Media, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1936, https://doi.org/10.5194/egusphere-egu26-1936, 2026.
Fractured crystalline rocks form highly heterogeneous subsurface systems in which fluid flow is controlled by the connectivity and aperture of fracture networks. These hydraulic properties are controlled by in-situ stress fields, which, in the shallow crust, results from the combined effects of regional tectonic stress, gravitational forces, and topographic relief. Although stress-dependent fracture behavior has been extensively studied at the fracture scale, the role of spatially variable stress tensors in controlling hydraulic conductivity at the catchment scale remains poorly constrained.
In this study, we investigate how topography and tectonic stresses interact with fracture network geometry to produce heterogeneity and anisotropy in hydraulic conductivity across a high mountain catchment. A three-dimensional geomechanical model is used to compute spatially variable stress tensors for both synthetic and real topographic surfaces. The resulting stress tensors are extracted cell by cell and applied to discrete fracture networks with different orientation statistics. Subsequently, for each fracture within every cell, the normal stress is calculated and used to determine stress-dependent hydraulic apertures through an exponential closure law. This is followed by the computation of fracture transmissivities using the parallel-plate cubic law. Finally, directional hydraulic conductivities are then obtained by numerical flow simulations in the principal directions (Kx, Ky, Kz).
The results show strong spatial heterogeneity and directional dependence of hydraulic conductivity across the catchment. At first order, the anisotropy of Kx, Ky and Kz is controlled by fracture network geometry, with low conductivities occurring in directions poorly aligned with dominant fracture orientations (e.g., low Kx for predominantly N–S–oriented vertical fractures or low Kz for horizontal fracture sets). Together with this geometric control, the relative orientation between fractures and the local stress tensor exerts a strong mechanical influence, fractures oriented perpendicular to the principal compressive stress experience increased normal stress, reduced aperture, and consequently lower hydraulic conductivity. In addition, spatial variations in topography introduce local perturbations in the stress tensor that produce zones of relatively higher hydraulic conductivity, where fractures remain less compressed. These combined effects lead to pronounced hydraulic anisotropy and preferential pathways at the catchment scale, highlighting the importance of explicitly accounting for spatially variable stress fields when modeling flow in fractured catchments.
How to cite: Figueroa, R., Halloran, L., Valley, B., Davy, P., Le Goc, R., Darcel, C., and Roques, C.: Influence of Fracture Orientation and Stress Fields on Hydraulic Conductivity Under Strong Topographic Relief, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14150, https://doi.org/10.5194/egusphere-egu26-14150, 2026.
Karst features are often linked to fractures and faults, considered the main pathways for water circulation and dissolution. Our study shows this view is incomplete, highlighting the overlooked role of stylolites, serrated seams formed under compressive stress, as starting points for karst. Fieldwork in Apulia (Southern Italy), combined with laboratory analyses, reveals that stylolites can initiate porosityguide fluid flow, then acting as proto-karst structures. They concentrate insoluble minerals such as clays, micas, and oxides that, under the right conditions, promote localized dissolution. Microscopic analyses of 15 samples show higher porosity within stylolites than in the host rock, with pores often clustering at stylolite–matrix boundaries. This suggests that what begins as a sealing surface can later evolve into a porous interface once stress regimes change or insoluble residues are removed. Mineral re-precipitation and localized dolomitization confirm past fluid circulation along stylolites. Field surveys reinforce these findings: in Apulia, over 80% of stylolites display dissolution features, compared with far fewer in faults or fractures. Orientation data from caves such as Grave Rotolo show main passages often align with stylolites rather than other tectonic structures. We propose a three-stage model of stylolite evolution: (1) sealed seams enriched in insoluble residues, (2) microporosity nucleation under favorable conditions, and (3) growth of interconnected pores into early conduits. This model shows stylolites can shift from barriers to fluid pathways. Recognizing them as proto-karst has practical implications for hydrogeology and carbonate reservoir management.
How to cite: Magni, S., Gratier, J.-P., and Šmuc, A. Š.: The role of Stylolites as Proto-Karsts in Speleogenesis, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-651, https://doi.org/10.5194/egusphere-egu26-651, 2026.
Quantitative assessment of long-distance hydrothermal groundwater flow in crystalline rock is challenging, primarily due to low matrix permeability and extreme heterogeneity. However, sites exist worldwide where deep groundwater circulation occurs over long distances within crystalline rocks. For instance, the Idaho Batholith (USA) hosts dozens of hot springs within granitic formations, featuring discharges in the tens of L/s and recharge areas of at least 5 000 km². Moreover, such flow can be remarkably rapid, as observed in Karlovy Vary (Carlsbad) in the Czech Republic. The main spring, known as the Vřídlo hot spring, has a temperature of 73 °C and a discharge of 30 L/s. Considering the local terrestrial heat flow, a recharge area of at least hundreds of km² is required to heat the Vřídlo spring, assuming a circulation depth of ≈ 2.5 km.
This study evaluates over a century of research on the Vřídlo hot spring and addresses the question of why water is channelized from hundreds of square kilometers into a single discharge point. Radiometric dating of spring travertine accumulations reveals that the Vřídlo hot spring has been active for at least 230 000 years. Stable isotope analysis confirms the thermal water is of meteoric origin, having infiltrated after the last Ice age. Evidence of high-permeability fractures with large apertures exists both directly within Karlovy Vary and in the broader region. Extensive drilling campaigns have identified fracture apertures on the order of tens of centimeters, with these zones characterized by flow rates of up to 30 L/s. Tracer tests revealed high anisotropy in flow velocities both within the Vřídlo discharge conduit and in the surrounding granite massif, demonstrating the presence of relatively voluminous conduits. Notably, a well-logging probe inadvertently descended through the bottom of a 133 m deep borehole into the conduit feeding the Vřídlo hot spring, reaching a final depth of 370 m in the granite.
Highly permeable flow paths extend at least tens of km from Vřídlo within the granite. Hydraulic tests conducted 15 km from Karlovy Vary established a direct hydraulic connection with the Vřídlo hot spring. This repeatedly verified relationship manifests as a decrease in the Vřídlo discharge following increased abstraction at the distant site, with a time lag of approximately three months. Without elucidating the origin of highly permeable conduits in granite and understanding the effect of hot groundwater flow on porosity and permeability changes, the implementation of deep geological repositories (DGRs) for radioactive waste cannot be considered safe. This is particularly critical given that a considerable proportion of such facilities are planned within granitic formations.
Funded by the GAUK No. 356525: Character of the feeding vents of thermal springs.
How to cite: Landa, D.-A., Bruthans, J., Vylita, T., and Mareš, J.: Multi-scale evaluation of high-permeability hydrothermal conduits in crystalline rocks (Karlovy Vary Spa, Czech Republic) , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7053, https://doi.org/10.5194/egusphere-egu26-7053, 2026.
In groundwater flow modelling within mining environments, three main numerical approaches are commonly used: i) the Equivalent Porous Media (EPM), according to which the rock mass and its discontinuities are represented as a continuum; ii) the Discrete Fracture Network (DFN), where hydraulic properties are assigned to natural and/or anthropic discontinuities; iii) the hybrid approach, a combination of the above. To evaluate the advantages and limitations of each method, a selected mine located in the Metalliferous Hills (Colline Metallifere; GR), Tuscany Region, Italy, was selected as a case study. The mine is part of a five-unit complex, joined by a level at −200 m asl, that was active for pyrite extraction until 1981. It was selected due to the availability of hydrogeological and operational data covering a time span of almost 100 years, from around 1920 to 2024. Currently, two 3d transient numerical models have been developed with FEFLOW 10, following the hybrid and EPM approaches. In the former, 1-D anthropic discontinuities (tunnels, shafts, and wells) and 2-D natural discontinuities (major faults) have been represented as discrete features, while in the latter they are implemented as boundary conditions and zones with specific hydrogeological properties, respectively. Hence, the main difference between the two modelling approaches is the representation of geological and anthropic elements. Both models have been successfully used to simulate three managed flooding events that occurred in the period 1997-2014, during which groundwater levels in the mine were raised from −140 m asl to the current −95 m asl by controlling pumping rates of the dewatering system wells active 24/7. At this stage of the research project, the main objective is to evaluate which model better fits observations during managed flooding events. The best fitting model will be used to run predictive simulations of the planned dewatering systems shutdown, which will raise groundwater levels from the current −95 m asl to 70 m asl, where a drainage tunnel is located. In the next stage of the research project, both models will be compared with a DFN model, to evaluate which approach is the most representative of the hydrogeological conditions of the study area and suitable for applications of similar case studies.
How to cite: Tonucci, R., Ghirotti, M., Canova, F., Castellani, S., Giaramida, G., Bonfedi, G., and Piccinini, L.: Comparison of numerical approaches for groundwater flow modelling within mining environment, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4188, https://doi.org/10.5194/egusphere-egu26-4188, 2026.
Karst aquifers are highly dynamic and heterogeneous systems, where solutes can be transported rapidly through conduit networks while parts of the contaminant mass temporarily reside in immobile zones, resulting in complex transport behavior and prolonged tailing in breakthrough curves. Accurately predicting solute dispersion in these systems remains challenging due to their multiscale structure and highly variable flow dynamics. To investigate these processes, flow in synthetic karst networks and in the real conduit system of the Seefeldhöhle (Switzerland) is simulated using a graph-based Laplacian solver capable of capturing both laminar and turbulent conditions. Solute transport is modeled using a Time-Domain Random Walk (TDRW) particle tracking approach. Particular attention is given to the role of mixing at conduit intersections, where both complete-mixing and streamline-routing rules are implemented to assess their influence on longitudinal and transverse dispersion. Transport behavior is characterized through first passage time distributions, particle visitation maps, and spatial moments. To further analyze the structure of particle velocities along trajectories, Lagrangian speed series are derived and their dependence structure is quantified using speed copulas. This analysis reveals distinct forms of correlation and intermittency that govern spreading, breakthrough tailing, and the sensitivity to network heterogeneity. While specific mixing processes at intersections affect local dispersion patterns, bulk metrics such as breakthrough curves and spatial moments remain comparatively insensitive. Building on these findings, the framework will be extended toward a multiscale upscaling approach based on a Continuous-Time Random Walk (CTRW), aiming to link conduit-scale velocity statistics with emergent network-scale transport dynamics.
How to cite: Grundhöfer, T., Dentz, M., and Kordilla, J.: Dispersion on laminar and turbulent network flow, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1801, https://doi.org/10.5194/egusphere-egu26-1801, 2026.
Karst aquifers commonly exhibit non-linear responses to rainfall and recharge due to the activation of preferential flowpaths and time-variable exchange between conduit, epikarst, and fractured matrix domains. Discharge, electrical conductivity, and turbidity are widely used to gain insight into these processes and to inform conceptual and numerical models; however, they often do not provide comprehensive information on internal connectivity and the mobilisation of stored water during transient recharge events. This contribution explores the use of flow cytometry (FCM) as a complementary observational approach for characterising karst aquifer system behaviours.
FCM has previously been applied in karst hydrogeology for purposes such as microbial contamination fingerprinting (Vucinic et al., 2022; 2023) and the detection of injected artificial microbial tracers such as yeast (Vucinic et al., 2024). These measurements can be interpreted in terms of changes in cell concentration, forward and side light-scatter distributions (used here as proxies for relative particle size and structural heterogeneity), and physiological state. Taken together, FCM observations can provide useful information on transport conditions and mixing processes within the aquifer, rather than solely reflecting water quality or pollution impacts. Hence, we argue that FCM signals can be used as indicators of hydrodynamic behaviour and relative mobilisation from fast and slower flow domains, capturing changes associated with rapid recharge, threshold behaviour, and recession-driven storage mobilisation.
The potential influence of microbial and chemical pollution on FCM signals is considered within this framework. While water quality conditions may affect baseline cytometric characteristics, event-driven deviations and response timing should remain informative for hydrological/hydrogeological interpretation. Emphasis is, therefore, placed on relative changes and event-phase behaviour rather than on absolute cytometric values.
The approach has implications for karst modelling, where FCM may provide additional constraints on connectivity changes, conduit–matrix exchange, and storage release during recharge events. When used in combination with standard hydrological/hydrogeological observations and measurements, FCM data may help refine conceptual understanding and support the parameterisation and evaluation of models describing transient karst aquifer system dynamics.
REFERENCES
Vucinic, L., O’Connell, D., Teixeira, R., Coxon, C., Gill, L. (2022). Flow cytometry and fecal indicator bacteria analyses for fingerprinting microbial pollution in karst aquifer systems. Water Resources Research, vol. 58, no. 5, e2021WR029840. https://doi.org/10.1029/2021WR029840
Vucinic, L., O'Connell, D., Dubber, D., Coxon, C., Gill, L. (2023). Multiple fluorescence approaches to identify rapid changes in microbial indicators at karst springs. Journal of Contaminant Hydrology, vol. 254, 104129. https://doi.org/10.1016/j.jconhyd.2022.104129
Vucinic, L., O’Connell, D., Coxon, C., Gill, L. (2024). Back to the future: comparing yeast as an outmoded artificial tracer for simulating microbial transport in karst aquifer systems to more modern approaches. Environmental Pollution, vol. 349, 123942. https://doi.org/10.1016/j.envpol.2024.123942
How to cite: Vucinic, L., O'Connell, D., Coxon, C., and Gill, L.: Linking flow cytometry signatures with flow and transport behaviour in karst aquifer systems, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11938, https://doi.org/10.5194/egusphere-egu26-11938, 2026.
Nowadays, the demand for the utilization of geothermal energy is growing worldwide. Several countries have numerous geothermal projects in the preparatory phase, but in many cases, it is difficult to decide which projects should be financed from a limited budget. To aid decision-making, various risk assessments are carried out to rank the projects, defining various technical and geological risk factors. One from risk factors can be the reservoir quality.
In this study, heterogeneity of porous sedimentary reservoirs is examined as a risk factor, since it (1) can significantly affect the amount of extractable water, (2) can have a strong influence on the thermal breakthrough time, (3) and strongly influences the size of the protection zone (PZ) around the geothermal well. Regarding the PZ, the size of the area associated with both the concentration front (SPZ – solute protection zone related to potential pollution) and the thermal front (TPZ – thermal protection zone) was examined separately. Two- and three-dimensional numerical models were built to study the effects of heterogeneity on geothermal well doublet where stochastic permeability distributions generated by sequential Gaussian simulation represent the heterogeneous reservoir. Detailed comparative simulations were used to reveal how (1) the scale of heterogeneity and (2) the permeability anisotropy influence the size of the PZ and the thermal performance of geothermal well doublet at different well distances and yields.
The results reveal how the required PZ increases with the scale of heterogeneity and provide probabilities for the size of PZ at different times for each heterogeneity scale. On a short time scale (<1 yr), there is no significant difference between the sizes of the SPZ and TPZ, and compared to the homogeneous approach, there is an increase of up to 50–75%. This decreases after 50 yr to 20% in the case of SPZ, while the thermal conduction smears the effects of heterogeneity for the TPZ size. The research also provides the probability of heat power achievable for each heterogeneity scale. Therefore, the research facilitates the more accurate risk assessment, i.e., based on the information about reservoir heterogeneity, it yields an estimate of the recoverable heat power and PZ size by calculating probabilities. Finally, the research applies the findings of synthetic simulations to a real geothermal project in Hungary, where the PZ probability map for geothermal well doublet was completed and the risk of heterogeneity was evaluated on the thermal performance.
Project no. KT-2023-900-I1-00000975/0000003 has been implemented with the support provided by the Ministry of Culture and Innovation of Hungary from the National Research, Development and Innovation Fund, financed under the KDP-2023 funding scheme.
How to cite: Molnár, B., Garaguly, I., and Galsa, A.: Numerical investigation of reservoir heterogeneity as a risk factor on the size of the protection zone and on the thermal performance of geothermal well doublet, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1202, https://doi.org/10.5194/egusphere-egu26-1202, 2026.
Subsurface heterogeneity and directional variance in the connectivity of permeable zones influences the movement of fluids which in turn, leads to anisotropy in permeability and flow. Our study investigates this anisotropy in both fractured and porous media by employing a set of synthetic fractal-fracture networks and multifractal models respectively, along with natural datasets. The latter include a high-resolution soil thin-section and a set of outcrop-based fracture maps that are evaluated using a “dynamic” approach. This is achieved by the means of simulating flow using TRACE3D, PFLOTRAN and Processing Modflow. TRACE3D, a streamline simulator, used by reservoir engineers, is employed for generating recovery curves assuming that each flow model is saturated with oil which is “recovered” by injecting water from a series of wells. It essentially serves as an indicator of connectivity, a time-varying “response curve” in our case, a practice not uncommon in the literature where multiple realizations of a pattern are compared using some kind of a “response”. In order to assess flow in x-direction a series of injection and production wells are placed along the boundaries parallel to the y-axis, with no-flow boundaries along the x-axis. The setup is then rotated 90 degrees to evaluate flow in the y-direction. For evaluating anisotropy in terms of equivalent permeability along x and y-directions, PFLOTRAN and Processing Modflow are used. PFLOTRAN is a high-performance, massively parallel simulator designed for modelling fluid flow and reactive transport in geologic porous media. Processing Modflow, on the other hand, implements Darcy’s law and mass conservation equations to simulate groundwater flow in aquifers. The flow setup used in PFLOTRAN and Processing Modflow is similar to the one used in case of TRACE3D, except that, instead of injection and production wells, PFLOTRAN applies pressure heads and Modflow applies hydraulic heads along the boundaries parallel to the y-axis that facilitate flow along the x-direction. This study adopts a “dynamic” approach for delineating subsurface anisotropy in terms of connectivity, permeability and flow in multifractal porous media and fractal-fracture networks.
How to cite: Biswas, R., Sahoo, B., Roy, A., and Tripathy, S.: Anisotropy in Fractured and Porous Media: A Simulation Approach, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2985, https://doi.org/10.5194/egusphere-egu26-2985, 2026.
Karst aquifers are formed by intricate conduit networks where variations in geometry significantly influence groundwater flow and transport. At the conduit scale, pronounced wall roughness (k/D∼10 -1), variable cross-sections, and tortuous centrelines challenge classical hydraulic descriptions based on smooth or idealised pipes.
In this work, we investigate flow dynamics in real karst conduits using numerical simulations over a wide range of Reynolds numbers (Re=1-104). The objectives are to quantify key geometrical and hydraulic descriptors, including effective cross-sectional areas, conduit centrelines, velocity distributions, and friction factors. Roughness characterisation is performed through spectral analysis of the conduit walls using Fourier-based techniques.
The conduit geometries are directly obtained from high-resolution scans of natural karst formations and retain their full geometric complexity. Flow simulations are carried out without geometric simplification, allowing wall-induced disturbances and roughness effects to be fully resolved.
The results highlight strong deviations from classical smooth-conduit behaviour, with geometry-driven heterogeneity significantly affecting velocity fields and friction losses across all flow regimes. These findings show that the dominant contribution to flow dispersion arises from large-scale roughness, allowing simplified conduit representations that preserve these features to yield hydraulic predictions comparable to those obtained in fully resolved geometries.
How to cite: El Mellas, I., Hidalgo González, J. J., and Dentz, M.: Roughness effects on flow in karst conduits, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6992, https://doi.org/10.5194/egusphere-egu26-6992, 2026.
Tracer tests in alpine karst aquifers often produce characteristic breakthrough curves, but turning them into reliable transport information remains difficult. Multiple peaks, strong asymmetry, and persistent tailing challenge many routinely applied models. We present a long-term tracer dataset from a karst spring in the Eisenerz Alps (Austria), comprising 15 tracer experiments conducted with the same setup over more than a decade. This exceptional consistency allows us to evaluate tracer transport across a wide range of hydrologic conditions and to systematically test how different modeling concepts respond to changing flow states. We examine the performance of commonly used approaches, from classical moment-based analyses and advection–dispersion models to mobile–immobile and multi-dispersion formulations. While more complex models offer greater flexibility, they also introduce issues of non-uniqueness, overfitting, and limited physical interpretability—especially when attempting to reproduce both peak structure and long-term tailing. Instead of favoring model simplicity or complexity, we focus on aligning model structure with the information content of the tracer data. Tracer transport is interpreted within a physically informed conceptual framework that integrates modeling results with independent evidence from local geology, hydrogeology, and speleological observations. This combination allows transport models to be constrained by system-specific knowledge, providing a more defensible basis for interpreting breakthrough curves across contrasting hydrologic conditions. By combining long-term tracer observations with physically constrained modeling and explicit uncertainty considerations, this contribution outlines a robust framework for interpreting alpine karst tracer tests—an essential step toward conservative assessments of contaminant persistence and risk at alpine karst springs.
How to cite: Seelig, S., Seelig, M., Goeppert, N., and Winkler, G.: Rethinking Tracer Interpretation in Alpine Karst: Comparing Physically Plausible Models for Comprehensive Transport Assessment, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9085, https://doi.org/10.5194/egusphere-egu26-9085, 2026.
In real carbonate geothermal systems, the groundwater flow can be influenced by both forced convection driven by the water table topography, as well as free thermal and haline convection induced by buoyancy forces (topothermohaline convection). The interaction of different driving forces has an impact on reservoir parameters such as water temperature and salt content, the most accurate knowledge of which is extremely important for the planning, implementation, and operation of successful geothermal projects. In this study, (1) we performed a series of 2D synthetic simulations to quantify the role of each driving force in the topothermohaline convection system, and (2) we applied the model to the Buda Thermal Karst (BTK) system as a demonstration area (Galsa et al. 2025).
During the synthetic model study, the effect of the water table gradient, the bottom heat flux, and the bottom salt concentration were examined on the dynamics of the coupled system using control parameters. We found in the base model that the recharge area is controlled by cold, fresh, and young infiltrated waters driven by water table topography, while a thermohaline dome characterized by warm, saline, and old waters forms beneath the discharge area. By increasing the water table gradient, topography-driven forced convection becomes dominant, so that intense regional groundwater flow sweeps warm, saline and old water out of the model, leading to a purely advective, stationary solution. By increasing the bottom heat flux, thermal convection becomes prevailing, and thus, paradoxically, intense thermal convection effectively cools the model. By enhancing the salt concentration, the density of the water increases, and a multi-layered thermohaline dome forms beneath the discharge zone with extremely high temperature, salt content, and water age.
This methodology was applied along a 2D section crossing the Buda Thermal Karst system to examine the evolution of the geothermal reservoir. After 10 kyr, precipitation saturated the western, unconfined karst by cold, young and fresh water, while in the eastern, confined part of the deep reservoir, thermohaline convection dominated the groundwater flow. The boundary separating the two regions shifted slowly but steadily to the east, so after 1 Myr, the zone with high-temperature (>200 °C), old (>100 kyr) brackish water retreated to the eastern, deeper (>2–3 km) part of the BTK. Overall, the thermally and chemically mixed water moves westward directly beneath the clayey Oligocene aquitard, can promote karstification and reach the surface in the vicinity of the Danube River, the main discharge zone of the area, producing both cold and lukewarm springs.
By taking into account the interaction of various forces driving groundwater flow, we can obtain a more accurate picture of the temperature, salinity, and water age distribution of the reservoir, which facilitates the precise design and efficient and sustainable operation of deep geothermal systems, minimizing the vulnerability of water resources.
References
Galsa, A., M. Szijártó, Á. Tóth, J. Mádl-Szőnyi, Topothermohaline convection – from synthetic simulations to reveal processes in a thick geothermal system, Hydrology and Earth System Sciences, 29/17, 4281–4305, 2025.
How to cite: Galsa, A., Szijártó, M., Tóth, Á., and Mádl-Szőnyi, J.: Topothermohaline convection – from synthetic simulations to reveal processes in the thick geothermal system of the Buda Thermal Karst, Hungary, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10893, https://doi.org/10.5194/egusphere-egu26-10893, 2026.
In this work we employ openKARST, which has been developed to simulate free-surface and pressurized dynamics in large conduit networks. The solver is designed to model transient flood waves as well as steady flow states under mixed flow conditions. Turbulence is represented via Darcy–Weisbach friction, with support for classical correlation function as Churchill and Colebrook-White. A particle-tracking and advection-diffusion module allows to model tracer transport and travel-time analyses. Verification and validation against analytical benchmarks and applications to large cave networks, such as the Ox Bel Ha system, demonstrates the numerical robustness. In this work we introduce a steady-state analysis workflow that condenses complex network solutions into simpler hydraulic metrics. Using conduit-specific Reynolds numbers and friction factors based on high fidelity simulations, we quantify how effective resistance evolves from partial filling to pressurized conditions with discharge, viscosity and diameter along karst conduits. To recover the observed trends without running full simulations, we propose a simple steady-state approximation in which the longitudinal water surface is represented by a generalized water depth profile, where cross-sectional hydraulics follow circular partial-filling geometry. This reduced model enables to compute conduit-averaged Reynolds numbers and friction factors and further provides conduit profiles that can be compared directly to simulated distributions, including regimes influenced by pressurization. Hence, this work offers a practical bridge between complex network hydraulics and compact predictive relationships to enable systematic exploration of parameter space and upscaling.
How to cite: Kordilla, J., Dentz, M., and Hidalgo, J.: openKARST: From detailed karst conduit dynamics to simplified hydraulic flow dynamics, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11205, https://doi.org/10.5194/egusphere-egu26-11205, 2026.
Groundwater flow in karstic aquifers is highly dependent on the geometry of the conduit network. Not integrating the geometry explicitly leads to inaccurate predictions of the aquifer's behavior under both high and low flow conditions.
At the same time, discharge time series at the spring contain information about the internal network organization. Hence, we study how hydrological properties can be related to the statistical metrics developed to characterize network geometry and topology.
One possible way to answer this question is to compare the results of flow simulations in an ensemble of karst networks. We propose using pyKasso, a pseudo-genetic karst network simulator, to obtain a large number of karst networks constrained by the same geological and hydrological settings. We then run flow simulations with openKARST, a flow simulator that handles turbulent and laminar flow in complex karst networks.
Here we present the first results of this comparison exercise on a catchment in the Chartreuse Mountains (France). The Aup du Seuil catchment (~10 km2) was chosen due to its 17-year record of discharge measures at the outlet, and the 25 kilometers of conduits explored and mapped by speleologists. In addition, its geology is relatively simple.
How to cite: Dufour, D., Renard, P., Kordilla, J., and Straubhaar, J.: Karst network geometry and groundwater flow in the Aup du Seuil catchment (Chartreuse, France), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12570, https://doi.org/10.5194/egusphere-egu26-12570, 2026.
Groundwater modelling serves as an important tool for the study and management of karst water resources. In such context, and especially when we try to obtain karst system understanding through groundwater models, it is not only important to get the right answers (e.g., a well-fitted spring discharge simulation), but we have to get those answers for the right reasons (i.e., because our model realistically reflects the underlying system). This problem has been addressed before, e.g., via model evaluation upon multiple system signatures such as spring discharge and solute concentrations. However, because the conduit network and the geological structure of karst systems is usually largely unknown, lumped models are often employed, which simplify the karst system and simulate the processes in a spatially aggregated manner. Due to those simplifications, assessing model (structural and process related) realism beyond a model’s ability to simulate system output signals is often impossible. Yet, using models to better understand karst system functioning is vital for sustainable karst water resources management. This discrepancy necessitates the study of karst systems and corresponding (lumped or spatially aggregated) models on the basis of systems that are fully known. For this purpose, we develop a virtual laboratory for karst groundwater modelling. This virtual laboratory facilitates the generation and simulation of synthetic karst systems and system signatures such as spring discharge using state-of-the-art spatially distributed numerical modelling. The virtual laboratory makes the generation of synthetic systems more accessible to a wider hydro(geo)logical community, taking a step towards more realistic process representation in (lumped) karst models in the future. Building on our previous work, this combination allows for the stochastic generation of conduit networks using pyKasso [1] and their subsequent embedding and simulation with the spatially distributed discrete-continuum model MODFLOW CFPv2 [2] via the CFPy package for the Python programming language [3]. We demonstrate the capabilities of the virtual laboratory for a number of cases, show upcoming development goals, and discuss opportunities for future applications.
[1] Miville, F., Renard, P., Fandel, C., & Filipponi, M. (2025). pyKasso: An open-source three-dimensional discrete karst network generator. Environmental Modelling & Software, 186, 106362.
[2] Reimann, T., Giese, M., Geyer, T., Liedl, R., Maréchal, J. C., & Shoemaker, W. B. (2014). Representation of water abstraction from a karst conduit with numerical discrete-continuum models. Hydrology and Earth System Sciences, 18(1), 227-241.
[3] Reimann, T., Rudolph, M. G., Grabow, L., & Noffz, T. (2023). CFPy—A python package for pre‐and postprocessing of the conduit flow process of MODFLOW. Groundwater, 61(6), 887-894.
How to cite: Rudolph, M. G., Reimann, T., Sinha, N., Renard, P., and Hartmann, A.: Towards a Virtual Laboratory for Karst Groundwater Modelling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16875, https://doi.org/10.5194/egusphere-egu26-16875, 2026.
The hydrogeological regime and the persistent moisture intrusion phenomena at the Holy Monastery of Kleiston are fundamentally dictated by a complex tectonic framework within the Sub-Pelagonian Unit of Mount Parnitha, NW of Athens, Greece. This study identifies the monastery’s location as a site of intense structural deformation, where the stratigraphic sequence is governed by a series of successive tectonic nappes. The structural architecture is defined by two primary thrust faults: an upper thrust (No 1) that positions carbonate rocks over volcanosedimentary formations, and a lower, sub-parallel thrust (No 2) that repositions the volcanosedimentary sequence atop an underlying lower carbonate series. Central to the water intrusion mechanism is the identification of two low-angle fault surfaces, designated as No 3 and No 4, which act as primary hydraulic discontinuities. Crucially, both of these tectonic surfaces are situated within the mass of the lower carbonate rocks and are oriented sub-parallel to the lower thrust, effectively mirroring its geometry. These intra-lithic discontinuities serve as the primary structural controls for groundwater movement, with the upper surface (No 3) coinciding with the main foundation level and the roof of the Catholicon, while the lower surface (No 4) dictates the base of the local epikarst hydrogeological system.
The interaction between this tectonic fabric and the carbonate lithology has facilitated the development of an extensive epikarst zone, exceeding 10 meters in thickness, characterized by high secondary porosity. This zone is defined by two principal discontinuity systems -striking NNW-SSE and NE-SW- alongside secondary fractures that have undergone significant karstification. The mechanical widening of these joints is further enhanced by the deep penetration of root systems, which extend up to seven meters into the rock mass, creating vertical conduits. Hydrogeologically, the epikarst functions as a perched aquifer, recharged through a combination of direct autogenic precipitation and lateral allogenic contribution from upstream debris and weathered volcanosedimentary mantles.
The manifestation of water and humidity within the monastery's functional spaces is the direct result of epikarst spring fronts emerging at the intersections of these specific tectonic surfaces with the building infrastructure. The upper fault surface (No 3) directs groundwater discharge into the Catholicon and the adjacent storage caverns, a process exacerbated by the thin carbonate cover which offers minimal lag time between precipitation events and intrusion. Simultaneously, the lower fault surface (No 4) facilitates discharge into minor caves and areas beneath the monastery’s retaining walls and communal spaces. This structural control explains the persistence of moisture even during arid periods, as the complex network of tectonic voids and karstified joints within the epikarst serves as a shallow reservoir. Consequently, the study concludes that the water intrusion at the Holy Monastery of Kleiston is a structurally driven phenomenon, where sub-parallel tectonic discontinuities within the lower carbonates serve as the primary conduits for the localized hydrogeodynamic discharge of the epikarst aquifer.
How to cite: Filis, C., Skourtsos, E., Vassilakis, E., Kotsi, E., Konsolaki, A., and Lekkas, E.: Hydrogeological coupling of epikarst dynamics and tectonic discontinuities: Impacts on water intrusion at the Holy Monastery of Kleiston, Greece, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16916, https://doi.org/10.5194/egusphere-egu26-16916, 2026.
The Chakrata region in the Lesser Himalayas of Uttarakhand, India, is characterized by complex lithology, steep topography, and a strong dependence of local communities on natural springs for domestic and agricultural water supply. This study investigates the hydrogeochemical characteristics, controlling processes, and drinking-water suitability of representative springs across the region. The results suggest that the spring water is fresh, transparent, and odorless in nature. The results suggest that surface water is slightly acidic to highly alkaline in nature (pH: 6.7 to 9.02) and moderately mineralized (TDS: 8 to 252 mg L-1and EC:16 to 503 µS/cm). The temperature of springs water varied from 6.5˚C to 26˚C, which depends on the ambient temperature of the infiltrating water, thermal conductivity of reservoir rocks and groundwater movement along with its residence period.The major ionic composition followed the order Ca²⁺ > Mg²⁺ > Na⁺ > K⁺ and HCO₃⁻ > SO₄²⁻ > Cl⁻ > F⁻, with Ca–HCO₃ and Ca–Mg–HCO₃ as dominant water facies. Various trace elements such as Al, As, Ba, B, Cr, Cu, Li, and Sr found in the spring water samples that contribute to its mineral profile and potential health benefits. Gibbs plot reveals that rock–water interaction, particularly the dissolution of carbonate and silicate minerals, primarily controls the chemistry of spring water. The LULC raster was clipped using WQI zone boundaries, and class-wise area statistics were computed. Zone III (WQI 18–47), is dominated by forest cover (~68.1%) and rangeland (~30.7%), while built-up areas (~0.2%), water bodies (~0.7%), bare land (~0.09%), and agriculture (~0.002%). Similar LULC patterns were observed in Zones I and II. Most samples fall within the permissible limits of BIS (2012) and WHO (2011) standards, indicating good to excellent water quality, although localized chloride enrichment suggest minor anthropogenic influence. The study provides critical baseline data for sustainable spring management and highlights the importance of lithology and recharge dynamics in governing spring water chemistry in the Lesser Himalayan region.
How to cite: Painuly, A., Rai, S. K., and Singh, V. B.: Understanding Spatio-Temporal Variability in Spring Water Hydrogeochemistry of the Chakrata Region, Lesser Himalaya, India, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18331, https://doi.org/10.5194/egusphere-egu26-18331, 2026.
How to cite: Renard, P., Straubhaar, J., Lauzon, D., and Trunz, C.: Karst network simulation with statistical learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18554, https://doi.org/10.5194/egusphere-egu26-18554, 2026.
The interaction between the low-permeability matrix and high-permeability conduits is a governing mechanism in karst hydrogeology, yet it remains difficult to quantify, especially under the disturbance of underground engineering. This study investigates the spatiotemporal evolution of groundwater exchange and the impact of tunnel drainage in a typical karst system in Southwest China. We developed a regional dual-permeability model by integrating GemPy for 3D geological structural modeling and MODFLOW-CFPv2 for hydraulic simulation. The model achieved a high reliability with a Pearson correlation coefficient of 0.98 between simulated and observed heads, outperforming traditional MODFLOW-Drain/River methods in capturing non-Darcy conduit flows.
Simulation results under natural conditions reveal significant spatial heterogeneity: while the matrix-to-conduit recharge dominates the spatial extent (37% of the area), the conduit-to-matrix leakage dominates the total exchange volume (58%). Sensitivity analysis identifies matrix hydraulic conductivity and conduit wall permeability as the most critical factors controlling exchange intensity, whereas rainfall intensity primarily influences local flow directions rather than the overall exchange pattern. Furthermore, the study quantifies the disturbance of tunnel drainage. Tunnel construction caused a maximum groundwater drawdown of 5.44 m and a residual drawdown of 2.00 m during the stable phase. Crucially, drainage activities induced flow reversals (transforming matrix recharge zones into leakage zones) and reduced the regional exchange flux by an average of 2,987 m³/d. These findings clarify the non-linear controls on karst flow exchange and provide a robust scientific basis for groundwater management and geo-hazard mitigation in engineering-disturbed karst aquifers.
How to cite: yang, D. and huang, T.: Quantifying Groundwater Exchange Mechanisms and Tunnel Drainage Impacts in a Regional Karst Aquifer: A Coupled GemPy and MODFLOW-CFPv2 Approach, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18855, https://doi.org/10.5194/egusphere-egu26-18855, 2026.
Karst landscapes are characterised by dissolution landforms, including cave conduit networks along which meteoric waters flow. Experimental setups and field monitoring reveal that two main geometric parameters, namely hydraulic diameter and relative roughness, control head losses and flow in a karst conduit. In empirical flow models, the first can be roughly estimated from the topometric data of traditional cave surveys, while the second is usually sampled from a range of plausible values to fit the observed hydraulic signals. A higher resolution representation of cave geometry, in the form of a dense 3D point cloud allows 1) for more complex geometric descriptors and their spatial correlations to be computed, and 2) their impact on hydraulic response to flow conditions to be tested.
We collected field data by mapping the cave passages of interest with a high-resolution dynamic laser scanner in various hydrologically active caves of the European Alps. We modelled the cave walls as triangulated meshes using Poisson reconstruction. We subsequently developed and compared various strategies to automatically extract a cave centreline based on these triangulated meshes. First, we approximated the curve skeleton of the cave conduit by contraction of mesh vertices. Second computed a path supported by a voxel set of the cave interior, maximising clearance to the cave walls. Third we implemented an iterative virtual stepper, finding at each step the next optimal direction of travel based on local cave geometry. We finally confronted these purely geometric approaches to build curvilinear objects with two hydraulic paths from numerical experiments.
Travelling along this 1D object, we implemented an algorithm to find optimal section orientations and compute sequential mesh-plane intersections. We computed a suite of 2D shape descriptors on this family of 2D polylines and analysed the spatial correlation of key descriptors designed to summarise passage size, shape and roughness. Using this newly assembled pipeline for dense point cloud geometric description, we highlight the value of generating new morphometric indicators adapted to the complexity of cave datasets to calculate equivalent hydraulic radii, or along passage roughness, both of which can be used to inform cave-network scale models of flow and transport.
How to cite: Racine, T., Noetinger, B., Cheong, O., Cabello, S., El Mellas, I., Straubhaar, J., and Renard, P.: Systematic characterisation of cave conduits geometry from dense 3D point clouds, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20101, https://doi.org/10.5194/egusphere-egu26-20101, 2026.
Identifying Structural Controls on Nonlinear Flow and Transport in Network Representations of Heterogeneous and Karst-Like Media Using Interpretable Machine Learning
Understanding how structural heterogeneity and connectivity control flow and transport remains a central challenge in fractured-porous media and karst systems, where strong velocity contrasts, preferential pathways, and non-Fickian transport are commonly observed across scales. Conceptual network representations provide a physically grounded framework to investigate these processes, but isolating the combined effects of geometry, topology, and finite connectivity on flow and transport behavior remains difficult.
In this work, we use pore network models as generic network representations of heterogeneous and karst-like media and combine them with interpretable machine learning to systematically identify the structural characteristics that govern flow and transport responses. Large ensembles of synthetic networks are generated with controlled variations in coordination number, throat size distributions, throat lengths, and connectivity. For each network, single-phase flow and advective-diffusive transport are simulated, and metrics characterizing flow heterogeneity and transport nonlinearity, such as velocity and flow-rate distributions, dispersion coefficients, spatial moments, and breakthrough curve scaling, are extracted.
Interpretable machine learning is used as a diagnostic tool, rather than a surrogate model, to quantify the influence of geometric and topological descriptors on flow localization, preferential channeling, and anomalous transport behavior. Feature importance and sensitivity analyses identify dominant structural controls and interactions, highlighting how connectivity, heterogeneity, and finite network structure shape nonlinear flow and transport. The results provide insight into the mechanisms controlling transport in strongly heterogeneous systems and illustrate how data-driven analysis can support physics-based modeling of fractured and karst environments.
How to cite: Puyguiraud, A., Gouze, P., Hyman, J., and Dentz, M.: Identifying Structural Controls on Nonlinear Flow and Transport in Network Representations of Heterogeneous and Karst-Like Media Using Interpretable Machine Learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20128, https://doi.org/10.5194/egusphere-egu26-20128, 2026.
Studying fluid flow in porous media is essential for various applications which include hydrocarbon extraction, groundwater management, sequestration or oil and gas extraction. Drainage efficiency influences recovery, storage capacity, and the long-term stability of reservoirs. During primary drainage, invasion of a non-wetting fluid through pore bodies and throats leaves behind trapped and disconnected clusters of wetting fluid leading to higher residual saturation. Recent studies have shown that capillary bridges and corner films can connect such trapped clusters and enable their subsequent drainage.
This work develops a pore-network model that incorporates film flow and adjustable surface tension on a 2D porous matrix generated by Random Sequential Adsorption and analyzed via Delaunay–Voronoi geometry. Pore pressure–volume relations are derived from a Young–Laplace description and combined with Poiseuille-type fluxes. In addition, pressure diffusion is studied on heterogeneous pore networks with fractal-like geometry, to quantify how this type of structure controls the spreading of pressure and dissolved species at the surface.
The objective of this study is to investigate how such film flows enhance connectivity and reduce residual saturation during drainage. Additionally, the transport of dissolved species that modify wetting properties and surface tension will be modeled, exploring their impact on flow efficiency. By combining numerical simulations with theoretical analysis, our work aims to quantify the impact of surface-chemical interactions on macroscopic flow behavior, providing new insights into optimizing fluid displacement and improving efficiency in subsurface processes.
How to cite: Rustamova, S., Toussaint, R., Reis, P., Moura, M., and Maloy, K. J.: Modelling the impact of film flow and adjustable surface tension in reservoir flow management, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21153, https://doi.org/10.5194/egusphere-egu26-21153, 2026.
