Artificial intelligence is changing the way geomorphologists observe and interpret landscapes. Its role is no longer limited to speeding up the identification of landforms. Data-driven methods are now being used to reconstruct incomplete surfaces, follow rapid environmental changes, estimate process dynamics, and explore how landscapes may evolve under future conditions.
This session focuses on the use of artificial intelligence to investigate both geomorphic forms and the processes that create and modify them. We seek contributions dealing with terrestrial, coastal, submarine and planetary environments, including rivers, slopes, badlands, karst areas, deserts, glaciers and periglacial regions. Studies may address landform recognition, surface and sediment characterization, topographic or bathymetric reconstruction, environmental monitoring, palaeolandscape analysis, susceptibility assessment and the prediction of geomorphic change.
We are particularly interested in research that combines information from different sources, such as satellite imagery, UAV surveys, digital elevation models, LiDAR, photogrammetry, geophysical data, field measurements and video observations. Contributions based on machine learning, deep learning, computer vision, generative approaches, foundation models and physics-guided methods are welcome, especially when they provide new ways of linking observations with geomorphic processes.
The session will also provide space for critical discussion. High predictive accuracy alone does not necessarily guarantee that a model is reliable or scientifically meaningful. We therefore encourage studies examining uncertainty, spatial dependence, model interpretation, transferability between regions, data limitations and unsuccessful applications. Comparisons between data-driven methods, traditional statistical approaches and process-based models are equally relevant.
By bringing together geomorphologists, remote-sensing researchers, modellers and data scientists, the session aims to discuss how artificial intelligence can help us move from the automatic description of landscapes towards a deeper understanding of their behaviour and evolution.