Interoperable Research Infrastructures for Earth System Science: Lessons from ICOS

Advancing knowledge discovery in Earth System Science requires research infrastructures to provide not only high-quality and reusable data, but also interoperable, machine-actionable services that can be combined across domains and scaled for data-intensive and AI-driven research. The Integrated Carbon Observation System (ICOS) delivers harmonised, long-term observations of greenhouse gas concentrations and fluxes across atmosphere, land ecosystems and oceans, underpinned by common standards, persistent identifiers, rich metadata and open access services aligned with FAIR principles. These characteristics position ICOS as a mature contributor to the European Open Science Cloud (EOSC) and as a practical example of how domain-oriented research infrastructures can support cross-disciplinary and AI-ready science.

In this contribution, we reflect on ICOS experience in implementing EOSC-relevant recommendations and discuss how these can foster cross–research infrastructure interoperability and support large-scale AI applications. We highlight how EOSC thematic nodes can act as coordination and integration layers that align standards, workflows and services across multiple research infrastructures, lowering barriers for cross-domain discovery and reuse. Building on concrete Earth system use cases, we outline how ICOS could contribute to such a thematic node by providing interoperable data services, domain expertise and reference implementations for AI-ready workflows. We further identify remaining gaps in semantic alignment, service orchestration and scalable access, and formulate recommendations for strengthening EOSC as a federated ecosystem capable of supporting next-generation, data- and AI-driven Earth System Science.