The Challenge Cities and infrastructure managers across the globe face mounting climate resilience challenges urban heat islands, wildfire and flood risk, vegetation loss, and rapid land-use change. While Earth Observation (EO) satellites offer a powerful monitoring tool, a critical gap exists: high-resolution commercial imagery is prohibitively expensive, while freely available open data from missions like ESA’s Sentinel-2 often lacks the spatial detail needed for meaningful environmental analytics. This resolution gap has historically limited the reach of satellite-derived insights, leaving many cities, regions, and developing-world geographies underserved.
Our Approach The EASE project directly addresses this challenge by developing state-of-the-art Super-Resolution (SR) deep learning models capable of enhancing medium-resolution Sentinel-2 imagery by up to 10×, without dependence on costly reference data. Crucially, EASE goes beyond image sharpening, it embeds advanced eXplainable AI (XAI) techniques tailored specifically for the SR context. These tools make model outputs transparent, interpretable, and trustworthy, enabling detection of artifacts such as hallucinations and ensuring that the enhanced imagery is fit for high-stakes operational decision-making. The entire pipeline is powered by Luxembourg’s MeluXina High-Performance Computing (HPC) infrastructure, enabling large-scale model training, domain adaptation, and deployment at speed.
What We’re Building EASE is structured across six interconnected work packages spanning data preparation, HPC integration, SR model development, XAI integration, and real-world validation. Use cases include tree detection, sealed surface mapping, heat island classification, and vegetation health monitoring. The final deliverable is a commercially deployable, globally scalable SR-XAI pipeline fully integrated into WEO’s operational analytics platform, reducing costs, expanding geographic reach, and reinforcing the trustworthiness of environmental data products for clients in cities, infrastructure management, insurance, and beyond.
Partners EASE is a collaboration between WEO SAS, a Luxembourg-based environmental analytics company delivering satellite-derived intelligence to cities and infrastructure managers worldwide, and the Luxembourg Institute of Science and Technology (LIST), a leading public research organisation with deep expertise in AI, remote sensing, and HPC-integrated machine learning. HPC resources are provided by LuxProvide via the MeluXina supercomputer. The project is co-funded by the Luxembourg Ministry of Economy (MECO) and the Luxembourg National Research Fund (FNR) under the Joint AI-HPC 2025 call, with a total eligible budget of €1.1 million.