Session Topics

Session 1. Multiscale and Probabilistic Approaches to Flow and Transport Modeling

Organizers: Alberto Bellin, University of Trento, Felipe de Barros, University of Southern California, Aldo Fiori, University of Rome 3

Abstract: Understanding and predicting the movement of fluids and contaminants in water bodies is inherently uncertain due to the environment’s heterogeneity, limited site characterization data, and the complexity of coupled physical, chemical, and biological processes. This session focuses on recent advances in the probabilistic modeling of complex natural systems, including uncertainty quantification (UQ) as applied to surface and subsurface flow and transport phenomena. We invite contributions that address the characterization, modeling, and quantification of uncertainty in surface water and groundwater systems across multiple scales. Topics of interest include, but are not limited to:

• Probabilistic modeling of flow and transport
• Geostatistical methods and spatial variability in hydrogeologic parameters
• Data assimilation, inverse modeling, and Bayesian approaches
• Risk-based decision-making and uncertainty propagation
• Multiscale modeling and computational techniques for UQ
• Data-driven methods for UQ and parameter estimation
• Case studies demonstrating the practical application of stochastic and machine learning methods

This session aims to bring together researchers and practitioners working at the intersection of hydrogeology and hydrodynamics, environmental modeling, and computational methods, fostering discussion on how complexity-aware approaches can enhance the reliability and robustness of predictions in environmental and engineering decision-making.

Session 2. Computational ecohydrology

Organizers: Giacomo Bertoldi, Eurac, Sara Bonetti, EPFL, Xue Feng, University of Minnesota

Abstract: The study of ecohydrology concerns the interactions between the hydrological cycle and ecological systems, involving different biotic and abiotic components of the terrestrial biosphere. Representing such interactions in computational models can be challenging due to variations in their physical controls, human influence, as well as nonlinearities in the system across a multitude of spatial and temporal scales. However, observations at different scales are becoming increasingly available. They should be used to improve our capacity modeling of ecohydrological processes across scales, to advance understanding of ecosystem responses to disturbances or climatic stressors, improve predictions of the global water, energy, and carbon cycles, and help inform better ecosystem management and adaptation policies under climate change. This session invites submissions that advance the modeling of ecohydrological processes from micro to macro scales through (i) novel insights into dominant ecohydrological processes and scaling relationships, (ii) new developments in mathematical representations and model coupling, (iii) model-data integration, (iv) computational methods, (v) data-driven or hybrid modeling leveraging artificial intelligence and deep learning, and (vi) applications of computational models in both natural and managed ecosystems. The topics of interest include, but are not limited to:

• Watershed modeling, water resources, and water quality
• Land-atmosphere interactions
• Ecohydrology of agricultural or urban systems
• Plants’ influence on aquatic systems, hydrodynamics, and sediment transport
• Plant xylem and phloem transport
• Root-rhizosphere interactions
• Belowground water, carbon, and nutrient dynamics
• Population dynamics, including competition
• Ecosystem responses to biotic and abiotic stressors and human influence

Session 3. Computational modeling of subsurface processes for climate mitigation and energy transition

Organizers: Inga Berre, University of Bergen, Sebastian Geiger, TU Delft, Adriana Paluszny, Imperial College London

Abstract: Human-driven climate change presents one of the defining challenges of our time, demanding both substantial reductions in greenhouse gas emissions and the development of sustainable energy systems. The subsurface, long central to fossil fuel extraction, now holds a key role in enabling this transition. Technologies such as CO₂ storage, geothermal energy production, and hydrogen storage rely on a deep understanding of coupled subsurface processes, which includes thermal, hydrological, mechanical, and chemical, and biological processes. Challenges are related to nonlinear coupling, a wide span in spatial and temporal scales, and a structurally complex subsurface environment where the geology is inherently uncertain. This session focuses on advances in computational modeling of subsurface processes. Contributions addressing multiscale and multiphysics modeling, model coupling, data integration, uncertainty quantification, and simulation frameworks for subsurface applications are particularly welcome.

Session 4. Pore-scale processes of multiphase flow, transport, and geomechanics

Organizers: Sidian Chen, Stanford University, Davide Picchi, University of Brescia, Cyprien Soulaine, CNRS

Abstract: Understanding how fluids, solutes, and the pore matrix interact at the pore scale is fundamental to predicting the behavior of subsurface processes. This session highlights recent advances in experimental, computational, and theoretical methods that reveal the multiphase flow dynamics, transport mechanisms, and geomechanical responses emerging from microscale heterogeneity. Contributions will explore topics such as fluid–fluid and fluid–solid interfacial interactions, capillarity and wettability effects, pore-scale deformation and failure, reactive transport coupling, development of biofilms, and flow of solid suspension. We invite studies leveraging Computational Fluid Dynamics, Pore-Network Modelling, lattice-Boltzmann, high-resolution imaging, microfluidics, and innovative analytical approaches. By improving the understanding of pore-scale coupled physics, this session aims to advance our ability to model and manage subsurface systems across applications including energy storage, carbon sequestration, groundwater remediation, geohazards, and geothermal production.

Session 5. Advances in remote-sensing tools and data driven methods for understanding hydrological processes

Organizers: Manuela Girotto, UC Berkeley, Ben Livneh, UC Boulder, Chrisitan Massari, CNR

Abstract: Remote-sensing tools and data-driven methods are rapidly advancing our ability to observe, quantify, and understand hydrological processes across scales. These advances, including the integration of multi-source remote-sensing data and the use of machine learning and other computational approaches, are enabling new insights into the spatial and temporal variability of water and energy fluxes, including soil moisture, snow, and surface and groundwater dynamics. This session highlights recent progress in the development and application of innovative methods that leverage remote-sensing observations, data assimilation, and artificial intelligence to enhance our understanding of the terrestrial water cycle and freshwater resources. We invite contributions that explore multi-scale, multi-variable, and multi-sensor approaches for characterizing hydrologic states and fluxes, quantifying and reducing uncertainty, and enabling data-driven discovery of complex hydrological processes.

Session 6. Inverse problems in hydrosystem modeling

Organizers: Anneli Guthke, University of Stuttgart, Wolfgang Nowak, University of Stuttgart, Velimir Vesselinov, EnviTrace LLC

Abstract: Inverse problems encompass a wide range of mathematical and computational techniques aimed at estimating model parameters, quantifying uncertainties, and identifying plausible structures of water resources models to improve their predictive capabilities. These problems arise when model outputs, e.g., streamflow, groundwater heads, soil moisture, or solute concentrations, are observed and used to update model states (data assimilation), and to identify model parameters and/or the functional form or plausibility of processes that are difficult or impossible to measure directly (inference). We welcome contributions along the following lines of research, both theoretical developments and practical applications:

• Methods for parameter inference, model selection/averaging, and uncertainty quantification, be they Bayesian, frequentist, optimization-based, or AI/ML-based
• Statistical methods for sensitivity analysis and for optimal experimental design
• Approaches to infer mathematical terms for poorly understood processes and improve system understanding
• Concepts to identify and treat model-structural errors
• Approaches to improve computational efficiency of solving inverse problems, including surrogate and AI/ML-based techniques
• Methods to increase the transparency and robustness of workflows and help communicate inference results
• Variations of all the above specific to low-dimensional, high-dimensional, geostatistical, linear, or non-linear inverse-problem settings.

Session 7. Data-driven and physics-based machine learning methods for forecasting and knowledge discovery of subsurface hydrology

Organizers: Valentina Ciriello, University of Bologna, Ahmed ElSheikh, Heriot-Watt University, Alexandre Tartakovsky, University of Illinois Urbana-Champaign

Abstract: We invite contributions that demonstrate innovative applications of machine learning (ML) techniques for forecasting hydrological and flow behaviors, unraveling complex subsurface interactions, and advancing our understanding of subsurface fluid flow systems. We especially encourage studies that not only leverage ML with big data but also integrate fundamental physical insights and knowledge discovery. This session aims to provide a platform for researchers to share their work at the intersection of cutting-edge ML and key hydrological challenges, inspiring future developments in machine learning for subsurface hydrology and fluid flow. We welcome contributions in the following areas: surrogate modeling and emulation of complex subsurface models; data assimilation and uncertainty quantification; reinforcement learning for reservoir optimization; differentiable modeling to incorporate physical knowledge; explainable and interpretable ML for physical models; benchmarking studies; and the utilization of foundational models in subsurface applications.

Session 8. Computational methods for human–water system modeling and decision making

Organizers: Hoori Ajami, UC Riverside, Matteo Camporese, University of Padova, Adam Szymkiewicz, Gdańsk University of Technology

Abstract: Sustainable management of water resources increasingly depends on our ability to represent and predict the complex interactions between human and natural systems. Advances in computational modeling have enabled the integration of hydrological, ecological, social, and economic processes within unified frameworks that capture feedback between human decisions and terrestrial hydrologic cycle. This session invites contributions that leverage physically-based and data-driven methods to improve understanding, prediction, and management of coupled human–water systems, particularly those affected by climate and land use change. We welcome studies that employ numerical modeling, systems analysis, optimization, data assimilation, or machine learning approaches to address decision-making under uncertainty, resilience of water systems, and adaptive management strategies. Topics of interest include, but are not limited to, agent-based and network models of human behavior, coupled socio-hydrological modeling frameworks, optimization of water allocation and infrastructure, and participatory or multi-objective decision-support systems. By fostering cross-disciplinary discussion, this session aims to advance computational methods that link hydrological processes with human actions, supporting more robust, equitable, and sustainable water management.

Session 9. Computational Techniques for Hydroclimatic Simulation, Forecasting, and Analysis

Organizers: Giuseppe Mascaro, Arizona State University, Simon Michael Papalexiou, Hamburg University of Technology, Elena Volpi, University of Rome 3

Abstract: This session focuses on recent advancements in simulating and forecasting key hydroclimatic variables, including precipitation, temperature, wind, humidity, radiation, soil moisture and discharge. We emphasize emerging computational methods for modeling time series, spatially distributed fields, and extreme events. We also focus on techniques to simulate and/or detect the effect of climate variability and long-term climate change. Topics include:

• High-resolution numerical models of weather–climate–land interactions
• Stochastic and probabilistic simulations, including extremes and compound events
• Machine learning (ML) and hybrid physics–ML methods for hydroclimatic forecasting
• Computational techniques for detecting changes in the water cycle (e.g., trends, nonstationarity, regime shifts)
• Data assimilation, ensemble prediction, and uncertainty quantification
• Integration of remote sensing, reanalyses, and real-time monitoring networks

We welcome case studies from diverse regions that demonstrate the capabilities and limitations of these methods, as well as their relevance to water resource planning, early warning applications, and climate risk assessments.

Session 10. Advances in reduced-order modeling, data assimilation, and digital twins

Organizers: Mario Putti, University of Padova, Gianluigi Rozza, SISSA, Daniel M. Tartakovsky, Stanford University

Abstract: The rapid expansion of monitoring networks delivering high-resolution real-time measurements has generated widespread interest in data assimilation and, more generally, the interplay between data and mathematical models, which is transforming water resources research dealing with the environment, energy, climate, public health, etc. These and other fields of research rely on science-based predictions that combine integrated multidisciplinary models and data of different types and spatiotemporal resolutions. This session explores emerging computational approaches that bridge observation and prediction to enhance decision-making. In this context, Reduced Order Modeling enables rapid, accurate simulations with reduced computational cost, forming the foundation for Digital Twins—dynamic virtual systems that assimilate real-world data for continuous forecasting and control. Despite recent progress, significant challenges remain in scaling these approaches to large, multi-physics systems and achieving seamless coupling between model reduction, data assimilation, and real-time control. This session invites contributions showcasing theoretical advances and applications in model reduction, data assimilation, and digital twins that push the boundaries of predictive modeling and real-time decision-making for water systems.

Session 11. Data-driven and physics-based machine learning methods for forecasting and knowledge discovery of surface hydrology

Organizers: Dipankar Dwivedi, Lawrence Berkeley National Laboratory, Uwe Ehret, Karlsruhe Institute of Technology, Basil Kraft, ETH

Abstract: We invite abstracts for a session on data-driven and physics-based machine learning (ML) methods for forecasting and knowledge discovery in surface hydrology. The session focuses on innovative approaches that integrate empirical ML techniques (e.g., LSTMs, CNNs, PINNs) with physical principles (e.g., mass and energy conservation) to enhance predictions of streamflow, flooding, runoff, and lake water quality. We particularly welcome contributions on hybrid modeling frameworks, scalable forecasting tools, surface–subsurface emulators, and interpretable ML approaches that advance process understanding. Topics of interest include physics-guided ML for rainfall–runoff modeling, large-sample hydrology (e.g., CAMELS), theory-guided data science, and applications in flood and drought prediction and water quality assessment. Submissions should emphasize novel methodologies, case studies, or open challenges in ML-driven hydrology.

Session 12. Transport and mixing in heterogeneous porous and fractured media across scales

Organizers: Branko Bijeljic, Imperial College, Marco Dentz, IDAEA-CSIC, Delphine Roubinet, CNRS, Geoscience University of Montpellier

Abstract: Physical and chemical heterogeneity are intrinsic to geological media and control flow, dispersion, mixing and reactive transformations across a broad range of spatial scales in heterogeneous porous, fractured and karst media. Spatial variability ranges from the scales of pores, fractures and conduits to the regional scales of porous formations, fracture and karst networks. The quantification of these processes is key for a broad range of applications including groundwater flow and transport, underground hydrogen and carbon dioxide storage, geothermal energy production, geological storage of radioactive waste. The aim of this session is to discuss new experimental, numerical, and theoretical approaches to characterize and quantify medium structure, transport, mixing and reaction at micro, meso and macro scales, and the interaction between these processes across scales.

Session 13. Computational modeling of hydrodynamics and morphodynamics processes in coastal and river systems and mass flows

Organizers: Vittorio Di Federico, University of Bologna, Matthew Farthing, ERDC, Hui Tang, German Research Center for Geosciences

Abstract: Geophysical flows occur over a wide range of temporal and spatial scales across coastal and riverine environments as well as the land surface. Wind-induced waves, tidal forcing, and runoff process drive circulation patterns and sediment transport, while density differences and turbulence create complex interactions between fluid flow and evolving topography. Predicting these interactions is critical for mitigating hazards, quantifying landscape change, and understanding the impacts of extreme weather and sea level rise on water resources and built infrastructure. Numerical models to represent these processes must be accurate, efficient, robust, and capable of handling multi-scale phenomena from individual events to decadal and longer timescales. This session encourages submissions on novel computational algorithms, integration of data and physics-based modeling through machine learning and artificial intelligence, multi-scale modeling approaches, model coupling strategies, and large-scale parallel implementations for coastal, riverine, and geophysical flow applications. Applications of interest include studies of flood mitigation systems, sediment transport and morphodynamics, contaminant dispersion, wetland and delta evolution, freshwater/saltwater interaction, and mass flow hazards over complex topography.

Session 14. Advances in computational modeling of vadose zone processes

Organizers: Tissa H. Illangasekare, Colorado School of Mines, Bo Guo, University of Arizona, Jan Vanderborght, Juelich

Abstract: The vadose zone, located at the heart of the critical zone, extending from the water table to the land surface, has a strong control on various processes such as aquifer recharge, evapotranspiration, and chemical transport. Decades of research to address problems in soil physics and subsurface hydrology have advanced the modeling of vadose zone processes. However, the challenges associated with highly nonlinear dynamics, heterogeneity of soils at all scales, and a multitude of process interactions still remain, particularly in applications associated with the transport and retention of chemicals (especially the recent emergent contaminants) and land-atmosphere interaction that require higher spatially and temporally resolved modeling. Innovative computational methods are needed to develop models at multiple spatial scales, specifically in combination with limited measurement capabilities in the subsurface. We invite abstracts that address these challenges and advance our understanding and computational representation of vadose zone processes. Topics include, but are not limited to, the transfer of small-scale understanding to larger scales, representation of heterogeneity across scales, effective representation through constitutive relationships, vadose zone modeling at regional scales, representation of vadose zone processes in climate and earth system models, experiments for improved computation, as well as improved numerical solutions including the use of machine learning.

Session 15. Advances in integrating surface and subsurface hydrological modeling

Organizers: Xingyuan Chen, PNNL, Julie Mai, University of Toronto, Alessandra Marzadri, University of Trento

Abstract: The stress of water resources globally has and will increase due to continuously growing demands and changing driving forcing. Central to addressing this challenge and proposing sustainable management strategies is using integrated models that represent the tight connection between surface and subsurface environments at scales ranging from individual channel reaches up to the entire global river network. With the increase in computational resources and the advent of artificial intelligence (AI) and machine learning (ML) methods, we are at the cusp of major advances in physical, chemical, and biological representations of these models, in the integration with observations and in using these models for prediction, management, and decision making. This session welcomes contributions on advances in integrated hydrological modeling, including better representations and coupling of physical and biogeochemical processes (e.g., recharge processes, snow melting, transport and transformation of nutrients and carbon), the integration of AI and ML, data assimilation, as well as case studies covering the wide range of scales where surface and subsurface integration plays a key role.

Session 16. Computational methods for sea-ice modelling and predictions: data assimilation, machine learning and physics

Organizers: Alberto Carrassi, University of Bologna, Tobias Finn, ENPC, Rym Msadek, CERFACS

Abstract: Understanding and predicting sea-ice variability remain central challenges in climate science, with profound implications for polar ecosystems, global weather patterns, and human activities in high-latitude regions. This session focuses on recent advances in computational approaches for modelling sea ice, ranging from traditional physics-based systems to cutting-edge artificial intelligence and hybrid methodologies. We invite contributions that explore innovations in numerical sea-ice models, including developments in dynamical and thermodynamical formulations, improved discretization schemes, scalable solvers, and high-resolution or multi-scale modelling frameworks. Submissions addressing hybrid modelling strategies—integrating physical principles with data-driven algorithms—are particularly encouraged, as are studies evaluating model skill, uncertainty, and robustness across temporal and spatial scales. A second theme of the session centers on data assimilation techniques tailored to the unique challenges of sea ice. We welcome work on assimilation of satellite and in-situ observations; advances in filtering, variational, and ensemble-based methods; and studies investigating how observational uncertainty and sparsity influence predictive capability. We also welcome contributions on hybrid data assimilation and AI methods. Finally, the session seeks abstracts on sea-ice prediction across weather, seasonal, decadal, and climate timescales. Topics may include forecast system development, machine-learning-based prediction architectures, coupled modelling approaches, and evaluation frameworks for assessing predictive performance. Overall, this session aims to bring together researchers developing innovative computational tools to improve our understanding of sea-ice processes and enhance predictive skill. By fostering dialogue across modelling, data assimilation, and machine learning communities, we seek to identify emerging opportunities and common challenges in the next generation of sea-ice research.

Session 17. Model coupling, domain decomposition, and solvers for multiphysics problems

Organizers: Jakub Both, University of Bergen, Nicola Castelletto, Lawrence Livermore National Laboratory, Andrea Franceschini, University of Padova

Abstract: This session aims to collect talks on the analysis and efficient numerical methods for coupled problems arising in water resources. Coupled problems may involve interactions between thermal, mechanical, flow, and chemical processes, and can include both locally coupled systems and boundary-coupled phenomena. Of interest are computational approaches that improve efficiency and accuracy, including adaptive methods, error estimation, and advanced linear and non-linear solvers. Contributions on domain decomposition, model coupling techniques, and other algorithmic developments that enable robust and scalable multiphysics simulations are especially welcome.

Sessions CMWR 2026

Bologna

June 28 – July 2