ENVRI (envri.eu) is a community of 27 research infrastructures in environmental sciences across the atmospheric, marine, terrestrial, and biodiversity domains. Since 2011, the European environmental research infrastructures that are part of ENVRI have been sharing a common vision: delivering interdisciplinary and integrated solutions to global challenges.
The ENVRI-Hub serves as a collaborative platform designed by the project ENVRI-Hub NEXT (2024 - 2027) to integrate data and services provided by distributed research infrastructures.
On the ENVRI-Hub, you can discover, retrieve, and analyse data. To do this, you can use three different tools:
The Catalogue of Services for refined web service discovery and data retrieval;
The Knowledge Base, which allows the discovery and contextualisation of datasets;
The Analytical Framework for data analysis and processing.
In this training resource, we will introduce the ENVRI-Hub data harmonisation approach, and the two components of the ENVRI-Hub search architecture: the Catalogue of Services and the Knowledge Base.
1. Essential Variables and Metadata Harmonisation
1. Essential Variables and Metadata Harmonisation
While Earth system observations are rapidly increasing in volume and complexity, data harmonisation becomes vital for robust science. ENVRI-Hub harmonises access to data through the Essential Variables approach.
Essential Variables (EXVs) consist of the minimum set of physical, chemical, or biological variables necessary to monitor the health of our changing planet. The prism of Essential Variables adopted, or being adopted, by ENVRI spans four interconnected domains: Climate (ECVs), Ocean (EOVs), Biodiversity (EBVs), and Geodesy (EGVs). EXVs allow ENVRI to establish exactly what needs to be observed to monitor the Earth system.
Because different research infrastructures store and label data differently, the ENVRI-Hub uses semantic profiles like the EPOS-DCAT-AP, the NERC vocabulary, and also the I-ADOPT framework to standardise how data is described, ultimately enabling automated, cross-domain interoperability.
Key takeways
The Essential Variables (EXVs) define what to observe.
A unified metadata application profile defines how data is described.
Metadata harmonisation enables the integration of data and services across formats, machine-readable workflows, and environmental science domains.
Without harmonisation, EXV-based data remains fragmented.
The ENVRI-Hub does not host or duplicate data and services, but efficiently classifies them in an actionable catalogue. Each service listed in the catalogue contributes to specific EXVs and pertains to at least one environmental domain.
The Catalogue of Services functions as a sophisticated aggregator that automatically harvests service descriptions from distributed Research Infrastructures and provides information on how to retrieve data from the chosen service. Data retrieved can be visualized in the Graphical User Interface (either on a map, a table or a graph depending on the data structure) and accessed through a set of unified and standardised APIs.
Moreover, the Catalogue of Services lists external services and resources, like Jupyter Notebooks, repositories and tools.
2.1 Walkthrough
We focus here mainly on the human access to the Catalogue of Services and therefore on its graphical user interface (GUI).
The interface shows a map and a collection of collapsible lateral panels, including Data, Graph, and Table:
The Data panel contains a free-text search bar, search filters, and the list of available services.
The Graph and Table panels display data for the services that support such a visualisation. Users can adjust how data appears on the map by clicking the "Layer" icon in the top-right corner.
The filters are additive and allow users to restrict the list of services shown in the lower part of the panel.
The filter offers several possibilities for searching services:
Free-text search (among the metadata of the services);
Geographical bounding box;
Time restriction;
Data and service provider;
Services offering specific visualisations (map, graph, table);
The ECVs that the service contributes to.
Each service card has an “i” icon that opens a detailed description, including all the metadata collected from the service owner, and a “star” icon to add the service to the favourites list and allow for the simultaneous visualisation of multiple services. When a service is selected, the corresponding payload is displayed on the GUI when a dedicated Converter is provided. Some services allowthe user to enter additional parameters to refine the payload request.
Some services have a “download” icon to export the payload, either directly or by accessing a dedicated external website.
Sharing data and services
Once users have set up the interface to their liking (including filters, favourites, the zoom level, and other customisations), they can generate a permanent web link (URL) by clicking the "Share" button. This link will recreate the same view anytime it is opened, on any computer. It is also possible to export the active map and legends as an image file.
EXV use cases
The "Use cases" section, available in the upper part of the interface, contains templates for Essential Variables studies. Clicking this button reveals a list of preconfigured study templates. Users can browse these templates, read their descriptions, and see which data services they use. To activate a specific example, the user clicks the button next to it; a pop-up message confirms that this will reset all current filters and favourites. After accepting, the interface reloads to display the selected study.
Access to possibly restricted/embargoed services is managed through the ENVRI ID (powered by GRNET), providing a unified Authentication and Authorisation Infrastructure (AAI) to the ENVRI-Hub.
2.2 Backoffice
The Catalogue of Services uses EPOS-DCAT-AP to harmonise metadata describing services, notebooks, and external repositories. EPOS-DCAT-AP extends core entities of DCAT-AP, like dataset and distribution, to also include the description of the web services.
Entities
The dataset contains information about the dataset itself, for instance: title, description, DOI, data provider, and related EXVs.
The distribution entity contains, for example, information about the license under which the dataset is distributed.
Linked to the distribution entity are the web service and the operation.
The web service contains a description of the web service and the web service provider.
The operation contains the URI template and the description of each parameter that is involved in the query of the service.
Filling metadata descriptions
The Catalogue of Services backoffice easily allows updating existing datasets or adding new ones by offering a GUI and test environments to visualise immediately any change made. This graphical application substitutes the manual filling of turtle files (.ttl) making it easier to harmonise services coming from different domains.
If you're a data scientist, most of your time goes into searching, collecting and curating data and other research assets. The ENVRI-Hub Knowledge Base finder expands what is available in the Catalogue of Services.
The Knowledge Base is a semantic search tool that indexes ENVRI-related resources using advanced information retrieval technologies, offering flexible keyword-based search, similarity-based ranking, and AI dialogue support.
Through the Knowledge Base, ENVRI-Hub provides search capabilities to locate environmental content coming from research infrastructures and reliable sources in three ways:
A classical search system powered by a Large Language Model (LLM) that comes with a free-text search field. This search option provides users with an AI-generated summarised response and yields a list of search results, including datasets, web APIs, notebooks, images, and contextual webpages. The search outcomes can also be visualised as a navigable knowledge graph.
A dialogue-based search agent, which complements the classical search with a more natural communication style, allowing users to discuss and analyse content with follow-ups.
A virtual research environment to run the shared notebooks, which we found from the search system, to explore the datasets, metadata, and APIs. Overall, to consolidate all the services, a federated authentication system is implemented that enables RIs to easily and seamlessly index their data while adhering to FAIR principles.
3.2 The Environmental Expert
Powered by LLM, the “Environmental Expert” search agent is trained on specific contextual information coming from the ENVRI-Hub Knowledge Base and the Catalogue of Services.
Standalone Interface: Users can interact with a dialogue agent through a pop-up window or a dedicated interface.
User Guidance: The agent acts as a starting point for research. Users can ask research questions and receive links, guidance, and explanations of search outcomes to help navigate other tools in the ENVRI-Hub.
Specialised Knowledge: While commercial LLMs may perform similarly or better for general public knowledge, this specialised agent is thoroughly trained on the ENVRI ecosystem and provides precise answers for ENVRI-specific topics.
Code Generation: The agent can generate code from ENVRI's analytic framework libraries to compute Essential Variables (EXVs). Users can execute this code either within the ENVRI analytical framework, a Jupyter notebook environment, or a customised Virtual Lab.
Zhiming Zhao | Associate Professor of Informatics at University of Amsterdam / LifeWatch ERIC
References
References
Carval, T., Zhao, Z., Dobler, D., & BODERE, E. (2026). ENVRI-Hub NEXT_D10.1_Semantic Search in ENVRI Catalogue and RI Catalogues (Version V1_Under EC Review). Zenodo. https://doi.org/10.5281/zenodo.18433332
De Nart, D. (2026). ENVRI-Hub NEXT_D7.2_First Report on Integration of Catalogue with the Analytical Framework (Version V1_Approved by the EC). Zenodo. https://doi.org/10.5281/zenodo.18683335
Dema, C., Fiebig, M., Shridhar, J., Vermeulen, A., Turco, A., Gutierrez, M., Thijsse, P., Bumberger, J., D'Amico, G., Ripepi, E., Izzi, F., & La Scaleia, G. (2026). ENVRI-Hub NEXT_D11.2_Metadata and Vocabularies Harmonisation (Version V1_Approved by the EC). Zenodo. https://doi.org/10.5281/zenodo.18668275
Islam, N. T., & Zhao, Z. (2026). ENVRI-Hub Next_D9.1_V1_ D9.1 Knowledge Base Recommender System Design_Public_Approved by the EC (Version V1_Approved by the EC). Zenodo. https://doi.org/10.5281/zenodo.18662776
Turco, A. (2026). ENVRI-Hub NEXT_D7.3_First Release of the Map-based User Interface (Version V1_Approved by the EC). Zenodo. https://doi.org/10.5281/zenodo.18683125
Kety Giuliacci | Metadata curator at INGV
Alessandro Turco | IT Project Lead at EPOS ERIC
Zhiming Zhao | Associate Professor of Informatics at University of Amsterdam / LifeWatch ERIC