The AETHER project has published its Requirements and Activity Baseline Report, marking a key milestone following the successful completion of the Requirements Review (RR) phase.

This report establishes the scientific, technical, and operational foundation of the project and provides a shared baseline for all subsequent development activities. It consolidates the project’s scope, methodology, system architecture, data requirements, ethical considerations, and use case definitions, ensuring alignment across the consortium and with the European Space Agency (ESA).

About the Report

The document presents a comprehensive overview of the AETHER self-explainable AI (S-xAI) framework, designed to improve transparency, interpretability, and trust in Earth Observation (EO) – based artificial intelligence systems. It details how AETHER integrates concept-based model explainability with retrieval-augmented generation (RAG) to produce human-understandable, scientifically grounded explanations of AI-driven EO outputs.

The report also defines the technical requirements, including software, hardware, data, and governance aspects, and describes a unified, modular architectureapplicable across all project use cases.

Key Content Includes

• State-of-the-art review and gap analysis in explainable AI for Earth Observation
• Description of the proposed Self-Explainable AI (S-xAI) methodology
• Technical and operational requirements for the proof-of-concept system
• Ethical, privacy, and governance considerations aligned with EU and ESA policies
• Detailed descriptions of the three AETHER use cases:
  • Biodiversity monitoring
  • Crop yield prediction and rapid damage assessment
  • Urban heat island detection and analysis
• Identification of end users and expected scientific and societal impact

Role in the Project Lifecycle

The Requirements and Activity Baseline Report serves as a reference document for the entire project duration, guiding system design, implementation, validation, and review activities. It will be iteratively refined where necessary, while maintaining continuity and traceability across project deliverables.

The report has been reviewed in close collaboration with ESA and the Φ-lab, ensuring alignment with ESA’s strategic objectives for trustworthy, transparent, and operational AI for Earth Observation.

Access the Report

The full report is available for download here.