An Erasmus+ cooperation partnership integrating AI into water emergency management and environmental science education across three leading European universities.
Ai-LEARN is a pioneering model for integrating advanced AI into water emergency management and environmental science education — deployed across three leading universities in the Netherlands, Spain, and Greece.
The project creates high-quality, internationally accessible training modules on flood forecasting, drought analysis, and water management, combining state-of-the-art machine learning with real-world EU satellite data and field case studies.
At its core, Ai-LEARN rethinks how education is delivered: AI-driven systems — including Large Language Models (LLMs) — guide learners through complex environmental models, making advanced methodologies accessible and pedagogically sound. The consortium also develops AURA-Learn, an agentic framework in which student, teacher, and evaluator AI agents collaborate to personalise the learning process and continuously improve course quality.
Four interconnected goals driving educational innovation in water and environmental science.
Embed state-of-the-art AI tools — Large Language Models and ML algorithms — into educational modules, creating adaptive, interactive learning environments that foster digital literacy.
Create modules using real-time remote sensing data and advanced environmental modelling. Provide hands-on experience simulating flood and drought scenarios for climate adaptation.
Build a scalable educational framework through active partnerships among TUC, UPV/EHU, and IHE Delft — merging digital innovation and environmental expertise across borders.
Conduct rigorous evaluation studies monitoring improvements in digital competencies and practical environmental problem-solving skills, with a target of 30% improvement in learning scores.
Three regional case studies grounding the educational content in real environmental challenges.
Groundwater data infilling and drought assessment using ANN models and geostatistical methods applied to aquifer data from Cretan islands.
Hydrological analysis integrating EU satellite data (MTG) with remote sensing techniques and ML models, focused on the Basque Country region.
Urban and rural flood forecasting using LSTM networks applied to data from the Amazon and Magdalena River Basins, with collaboration from the Magdalena River Basin Authority (Colombia).
Five complementary work packages spanning the full project lifecycle from management to dissemination.
Overall coordination, financial oversight, risk management, and communication across all partners. Bi-annual General Meetings (one online + one in-person per year) and bimonthly WP operational meetings ensure continuous alignment. A dynamic risk register and standardised progress reporting keep all partners on track.
Development of the technical AI infrastructure: specialised TensorFlow/Keras models for flood forecasting (LSTM/ANN), groundwater data infilling, and drought assessment. An Educational LLM Integration Framework with prompting techniques and instructor guides enables AI-assisted learning. Three case studies — Greek islands, Spain hydrology, and Latin American watersheds — are fully implemented as structured, project-based learning modules.
Design and development of the WaterTech Learning Portal — an internationally accessible virtual learning environment building on IHE Delft's eCampus Moodle infrastructure. Integrates AI tools from WP2, supporting MSc-level education, short courses, and professional development. Designed with WCAG 2.1 AA accessibility standards and multilingual support (English, Spanish, Greek).
Comprehensive communication, dissemination, and exploitation strategy. Includes this public website, Open Educational Resources, public outreach events (minimum 15) in each partner country, policy briefings for regional environmental authorities, and engagement with environmental professionals and policymakers. MOOC development for global reach beyond the project lifecycle.
Systematic monitoring of educational impact, interdisciplinary collaboration, and research innovation. Pre/post knowledge assessments measure 30% learning improvement targets. User testing cycles at M8, M16, and M24 refine platform usability. Structured reports document outcomes at key milestones.