Submit a session proposal or short course

Proposed sessions

Click on the titles marked to see a descriptions.

1. AI-driven Numerical Modeling and Data Assimilation in Geodynamics

   Fan Xiao, Dunhui Xiao, Zenghua Li, Xiaohui Li, Zhankun Liu, Xinguang He

   The numerical modeling of geodynamics such as mineralization, diagenesis, and tectonic processes is based on the fundamental principles of mathematics, physics, and chemistry. Integrating geological data allows us to construct robust mathematical-physical models for quantitatively characterizing these geological processes. These models can be solved to simulate the complex geosystems and their spatiotemporal responses under extreme conditions within the inner Earth using numerical analysis methods such as finite element or finite difference and high-performance computing. This approach aims to investigate the mechanisms inherited in complex geological processes, and their evolutionary pathways, and to interpret the conditions of geological events. Nevertheless, geodynamics, including mineralization, diagenesis, and tectonic modeling, possess strong uncertainties due to dependence on unknown or ambiguous boundary conditions, initial conditions, and physical-chemical parameters. This requires the assimilation of observational data into physical-informed numerical models to improve the accuracy of solutions or predictions. In recent years, driven by the analytics of big data, computational science, and non-linear theories and methods in complex physical systems, significant progress has been made in the numerical modeling of geodynamics. Invited submissions for abstracts or/and full papers include (but are not limited to): (1) Numerical simulation of geological processes such as mineralization diagenesis and tectonics; (2) Computational geochemistry; (3) Computational geophysics; (4) Numerical geotectonic and geodynamic modeling; (5) Data-driven numerical modeling approach; (6) High-performance and high-throughput computing for geoscience data; (7) Data assimilation techniques in geosciences; (8) Data-driven modeling and state estimation of Earth's dynamic systems. The full papers can be recommended to be published in the IAMG journal of Computers & Geosciences or Applied Computing and Geosciences.
2. Spatial Association

   Yongze Song

   This session invites researchers to present the latest studies in spatial association methods and their practical applications. Accurate characterization of spatial association is fundamental for exploring spatial factors, improving spatial predictions, and supporting informed geographical decision-making. Models addressing spatial dependence, heterogeneity, geocomplexity, singularity, and similarity are central to describing spatial association. Moreover, recent developments in geospatial artificial intelligence have introduced powerful and precise approaches for analyzing spatial association. We welcome submissions from both theoretical and applied studies that contribute to the advancement of spatial association analysis.
3. Ontologies, Knowledge Graphs, and Large Language Models in Geoscience: Construction and Application

   Xiaogang Ma, Chengbin Wang, Anirudh Prabhu

   As the geoscience community increasingly turns to advanced AI methods, the integration of ontologies, knowledge graphs, and large language models (LLMs) offers transformative potential for data management and knowledge discovery. This session aims to explore innovative approaches to constructing and applying these technologies within geoscience. We invite presentations that demonstrate the development of ontologies and knowledge graphs specific to geoscientific domains, showcasing how these models and frameworks can enhance data interoperability and semantic clarity. In particular, we welcome presentations that demonstrate how the machine-readable semantics are leveraged in reasoning and inference tasks that lead to new knowledge discoveries. Moreover, we encourage discussions on the utilization of LLMs for data science tasks in geoscience, such as automation of literature review, hypothesis generation, and enhanced data querying and analysis. Other presentations such as using knowledge graph to guide and advise LLMs in data-intensive research are also encouraged. By bringing together experts in geoinformatics, artificial intelligence, and mathematical geosciences, this session aims to stimulate interdisciplinary dialogue and identify best practices for leveraging these advanced tools. Join us in exploring and discussion how the synergistic application of ontologies, knowledge graphs, and LLMs can advance a variety of topics in geoscience.
4. AI-driven Mineral Prospectivity Modeling

   Emmanuel John Carranza, Renguang Zuo

   Mineral prospectivity modeling as a computer-based approach to delineate target areas for exploration of certain mineral deposits in a mineral system has evolved from being knowledge driven to artificial intelligence (AI)-driven. The applications of AI in mineral exploration are ever increasing nowadays to address the complexity of relationships among datasets and with known deposit occurrences. The session welcomes submissions for presentations of: (1) novel AI algorithms and applications for recognition and integration of geo-anomalies to support mineral exploration, in 2D or 3D; and (2) novel AI algorithms and applications for analysis and synthesis of a variety of geoscience datasets to model mineral prospectivity and associated uncertainty, in 2D or 3D.
5. Machine Learning Applications in Geoscience Research

   Enamundram Chandrasekhar, Sang-Mook Lee, Byung-Dal So

   The geoscience data are inherently heterogeneous, encompassing spatial, temporal, and multiscale information. The steady penetration of machine learning (ML) and deep learning (DL) techniques into geoscience research has been emerging as a transformative force, providing a unique ability to detect patterns, make predictions, and enable new insights and methodologies across various applications in geophysics, geology and atmospheric science. This session aims to explore the innovative use of ML/DL techniques in geoscience research from subsurface imaging to environmental monitoring to predicting mineral prospecting zones and natural hazards. We invite contributions in all areas of geosciences that showcase cutting-edge research, case studies and advancements in ML/DL algorithms addressing challenges in geoscience research. Contributions to this session include but not limited to timeseries modelling, geospatial analysis, pattern recognition, data automation and uncertainty quantification.
6. Current Trends in Spatiotemporal Modeling

   Dionysios Christopoulos, Sandra De Iaco, X San Liang, Emmanouil Varouchakis

   The development of space-time methods remains at the forefront of current research in Spatial Statistics and Machine Learning. Flexible and interpretable models that can accurately capture the complex patterns of dynamic processes and reliably estimate uncertainties are needed. The key modeling challenges for the analysis of modern spatiotemporal data include the development of models that scale favorable with large data size, ability to handle heterogeneous data, multivariate dependence, multiple correlation scales, and the development of practical non-Gaussian probability distribution and non-stationarity. This session aims to assemble contributions that will advance spatiotemporal methods in mathematical geosciences by introducing novel concepts and methodologies, computational algorithms, and innovative applications or studies that focus on interesting or challenging spatiotemporal datasets. Topics of interest include the development of novel space-time covariance functions (e.g., non-separable models, models on the sphere and manifolds, multivariate dependence, complex-valued models), covariance-free approaches (e.g., models based on stochastic partial differential equations and explicit precision operators), innovative simulation methods, and computational advances for big space-time datasets. The session also invites contributions that focus on causal inference, uncertainty quantification, applications of deep learning to spatiotemporal datasets, non-Gaussian space-time approaches and multiscale models. Interdisciplinary contributions that combine the spatiotemporal dimension with physics, machine learning, and applied mathematics perspectives are also welcome.
7. Data-Driven Innovations for Mineral Exploration Decision-Making: Addressing Present and Future Challenges

   Behnam Sadeghi, David Zhen Yin, Rian Dutch, Putra Sadikin, Richard Scott

   This session welcomes all data-driven ideas to address the current and future challenges in energy-transition minerals exploration and development, especially the need for rapid and accurate information to make better decisions in addition to sample or survey optimization through data-driven and machine learning methods. Following the landmark Paris Agreement on Climate Change, a multitude of nations committed to a substantial reduction in greenhouse gas emissions. Central to this ambition is the advancement of clean energy technologies. However, the successful deployment of these technologies hinges on the availability and sustainable management of critical minerals like copper, lithium, nickel, cobalt, and rare earth elements. These minerals are fundamental in the manufacture of a wide range of clean energy products, from the batteries that power electric vehicles to the components essential for energy-efficient lighting and advanced electronics. The rising demand for critical minerals, essential for clean energy technologies like electric vehicle batteries and wind turbines, offers economic opportunities but also poses environmental and geopolitical challenges. Addressing this requires global collaboration to implement sustainable mining, enhance recycling, and foster innovation in alternative technologies while investing in infrastructure and workforce development. Recent progress in data science and machine learning has shown great potential to accelerate the discoveries of mineral deposits, improve resource efficiency, and sample/survey optimization. Such data-driven approaches can more effectively ingest geochemical, geological, geophysical, remote sensing, environmental data, and beyond, thereby proving better-informed mineral exploration. Our session will cover the latest data science and machine learning advancements in combining such multi-disciplinary data to enhance sustainable and efficient decision-making in mineral exploration. We encourage studies using new data-driven approaches for geochemical data analysis, geological modeling, geophysical inversion, mineral prospectivlity mapping, decision-making under geological uncertainty, and more.
8. Enhancing Quantitative, Geotechnologies and Programming Skills in Geosciences Education

   Francisco Tognoli

   A quantitative approach, supported by a survey of teachers and students, has provided valuable insights and inspired this session proposal. Today, quantification, geotechnologies and programming are increasingly essential skills for geoscientists. However, do geoscience programs worldwide provide students with a solid quantitative and spatial analysis foundation and foster essential programming skills? Or are we, instead, training "button-pushing" geoscientists? Discussions on the current state of geoscience curricula—and assessments of how well they prepare students for the demands of modern geoscience—are crucial. Given the field's rapid evolution, are classical teaching methods and traditional strategies in geology still relevant in the 2020s? Are there programs that have successfully equipped students to tackle global challenges and adapt to the profession's ongoing changes? An IAMG session can drive this critical discussion to enhance geoscience education. Submissions on curriculum design, teaching strategies, case studies and quantitative analyses are highly encouraged.
9. The 3D/4D geological modeling and targeting for mineral exploration

   Gongwen Wang, John CARRANZA, Deng Hao, Zhiqiang Zhang

   All the authors are International Association for Mathematical Geology (IAMG) Life Member. They have some 3D targeting projects communication and cooperations each others. Professor John Carranza is Editor-in-Chief of NRR IAMG, and he is key memeber of EIS (2024).
10. Recent developments in constructing geological structures: Beyond conventional methods

    Weisheng Hou, Mathieu Gravey, Baoyi Zhang, Nan Li, Yanshu Yin, Jiateng Guo, Qiyu Chen, Xiaohui Li

    The artificial intelligence (AI), multiple-point statistics (MPS), and other methods have significantly enhanced 3D geological modeling, overcoming the limitations caused by sparse data and complex shapes. This session bring together researchers in AI, MPS, stochastic simulation, and conventional geological modeling who have a common research question: detection, characterization, and reconstruction of patterns and structures of geological and geophysical data. We invite contributions showcasing novel methods with applications in mineral perspective, gas and reservoir, engineering geology, digital twins, regional geological investigation, and beyond, for which can be valuable additions to the methodological toolbox for reconstructing geological structures.
11. Mining geostatistics, optimization and geometallurgy

    Raimon Tolosana Delgado, Jörg Benndorf, Julian Ortiz, K. Gerald van den Boogaart

    The sessions aims to bring together all aspects of mining-relevant mathematical methods, along the whole mining cycle from exploration targeting to mine closure and on all scales from microstructure characterization to the long term mine planning and multi mine site operation planning. Important areas are: quantification and modelling of rock microstructures, geostatistics of geometallurgical variables incl high-order methods, modelling of beneficiation processes, structural modelling with uncertainty, stochastic mine planning, real time mining updating, and predictive process optimization. Contributions from all fields of application or development of geomathematical methods for mining are welcome.
12. Intelligent Reservoir Characterization and Modeling
13. Intelligent Reservoir Characterization and Modeling

    Qiyu Chen, Guillaume Pirot, Zhesi Cui, Shaoqun Dong, Gang Liu, Shu Jiang

    Reservoir characterization and modeling are crucial for informed decision-making in resource exploration, development, and management. These processes are widely applied in the exploration and production of subsurface reservoirs for hydrocarbons, geothermal energy, groundwater, and other resources. Recent advancements in computational techniques, machine learning, and integrated geological modeling have significantly enhanced the ability to characterize the complex spatial patterns and processes of subsurface reservoirs. This session will focus on the latest innovations in intelligent reservoir characterization and modeling, combining traditional geological expertise with emerging AI-based technologies. Presentations will cover a variety of topics, from data-driven modeling, AI-based pattern reconstruction, and digital twins, to the integration of multiple-disciplinary datasets such as seismic, well logs, and production data. The session will showcase how these intelligent approaches help improve reservoir characterization, simulation, and management, enabling more efficient and sustainable resource extraction.
14. Compositional and density data analysis in geosciences

    Karel Hron, Alessandra Menafoglio, Jennifer McKinley, Caterina Gozzi

    Compositional and density data analysis in geosciences This session will focus on the statistical/machine and deep learning treatment, modelling and interpretation of compositional and density data in geochemical applications, particularly for geochemical exploration and mapping. It will address the challenges of geochemical mapping with relative data, either with compositional data (typically geochemical data) and density data (typically particle size distributions). Geological survey data typically have both relative and spatial elements, both of which must be considered for meaningful analysis. The performance of empirical geochemical approaches also deteriorates when the geochemical datasets are too large, typically consisting of tens of thousands of observations and tens of variables. Conversely, such large datasets are advantageous for the distributional (functional) data approach that will be discussed in this session. The session will explore popular unsupervised multivariate/functional data analysis methods, such as dimension reduction (PCA) and clustering, to uncover inherent relationships and patterns in the relative data. It will also discuss process discovery and validation using techniques such as discriminant analysis, machine learning and deep learning methods for pattern identification and classification. All contributions on the application of (1) multivariate geoscience data processing within a compositional framework in a geoscience context (2) functional data analysis processing with distributional data by considering their relative character are welcome.
15. Marginal Seas - Modeling of interfaces between continents and oceans for sustainable development

    Jan Harff, Junjie Deng, Joanna Dudzinska-Nowak, Jinpeng Zhang

    Large parts of the world's population live near coasts and estuaries, so that the interactions between land and ocean play a fundamental role in the living conditions of these people. Disturbances of natural balances in the land-ocean mass and energy transfer caused by human activities and climate change require new holistic management concepts for the protection and sustainable use of the resources in marginal seas. These concepts base on the transdisciplinary interpretation of scientific data and the simulation of the development of marginal sea processes on integrated time scales. Numerical methods and models addressing climate and environmental change in the geological past but also for future projections are available, but more recently there are promising methods of big data analysis, Artificial Intelligence (AI) and Machine Learning (ML) debated and applied. The qualified use of these methods requires transdisciplinary co-operation chains between the natural sciences, socioeconomics and the humanities. In the proposed session, we invite representatives of these disciplines, together with data scientists and modelers to contribute through lectures and discussions to the development of advanced methods supporting sustainable planning and management of marginal seas.
16. Towards Large AI Models for Geosciences

    Suihong Song, Xiaocai Shan, Keyu Liu, Xin Li, Mingliang Liu

    As geosciences face increasingly complex challenges, large AI models—whether versatile fundamental models for multiple tasks or task-specific models with fewer parameters—may offer transformative opportunities. By leveraging cutting-edge algorithms, established geoscientific knowledge, and vast quantities of observable data, these AI models could reshape the landscape of geoscientific research and applications. This session focuses on (1) AI methodologies possibly used for large AI models, (2) construction of large-scale and high-quality training datasets, and (3) practices of already trained large AI models. Submissions are invited on AI techniques applicable to geosciences, including but not limited to Generative Adversarial Networks (GANs) and diffusion models conditioned on observable geoscientific data, discriminative and segmentation methods tailored for domain-specific applications, as well as large language model (LLM) related technologies. A major challenge in geosciences is the development of robust training datasets. We welcome contributions exploring workflows, methodologies, software solutions, best practices, and semi-technical narratives related to data collection and labeling. Some training data, such as conceptual geomodels, are especially scarce and could be constructed using simulation-based approaches, such as process-mimicking methods and efficient physics-based simulations. Long-term satellite imagery, spanning decades, also offers immense potential for dataset creation. Abstracts detailing methods, tools, or case studies for constructing such datasets and introducing new data libraries are particularly encouraged. Some large AI models may have already been trained for specific tasks. We encourage submissions that delve into the datasets, methodologies, performance benchmarks, and insights derived from these models. Lastly, the session seeks visionary contributions that explore the challenges and future prospects of large (or multi-task fundamental) AI models in geosciences. These grand perspectives can inspire novel approaches and set the stage for future advancements.