Partnering for progress

The technology and partnerships behind the Global Ecosystems Atlas
On this page

Building the platform to map the world’s ecosystems to a single standard is too big a job for any one organisation alone. The Global Ecosystems Atlas Technical Consortium brings together IUCN, Ai2, the European Space Agency (ESA), Esri and Google, alongside the Global Ecosystems Atlas core team at the GEO Secretariat and the Global Ecology Lab at James Cook University (JCU).

Each partner plays a distinct role, united by the common goal to deploy cutting-edge technologies to solve the complex data challenges that come with mapping ecosystems at scale. This ensures an approach that is both scientifically rigorous and nationally trusted.

The collaborative work is designed so that the output of one stage feeds into the next, with the contributions of any one partner complementing those of the others. Partners are also aligned on shared aspirations beyond the technical side, including visibility, uptake and impact.

A common language

Combining maps from different sources into a single product only works when ecosystems are described in a consistent way. The Global Ecosystems Atlas uses the IUCN Global Ecosystem Typology (GET) as its common classification scheme, and relies on published cross-referencing guidelines to ensure that the wide array of input data can be harmonised meaningfully into a common geospatial framework.

Where countries or regions already have well-developed classification schemes, our partners help build bridges. ESA, for example, coordinates guidance for translating between European classifications and the IUCN GET, while Esri helps to implement the cross-referencing at a global scale. 

Mapping the unmapped

Many ecosystems have never been mapped. To fill those gaps, the Global Ecosystems Atlas uses geospatial foundation models. These are AI models that are trained on large volumes of satellite data to recognise patterns.

 Ai2 contributes models from its OlmoEarth platform, trained on multi-sensor satellite data to identify patterns and generate high-quality, high-resolution ecosystem maps using known occurrences of ecosystems. Google contributes with AlphaEarth Foundations and its satellite embedding, a set of 64 high-resolution layers that can be applied directly to ecosystem classification globally.

Google Alpha Earth
Diagram showing how AlphaEarth Foundations works, taking non-uniformly sampled frames from a video sequence to index any position in time. This helps the model create a continuous view of the location, while explaining numerous measurements.

In addition, ESA’s extensive Earth observation systems are central to ecosystem mapping efforts around the world. Its programmes in support of national ecosystem mapping, such as the World Ecosystem Extent Dynamics (WEED) project, supplies high-quality inputs for the Global Ecosystems Atlas.

Teaching models what to look for

To map ecosystems at a global scale, AI models need a large set of locations with know ecosystem occurrences. For the Global Ecosystems Atlas, this is where EcoTrain comes in.

Still in progress, EcoTrain will be an open training and validation dataset. Google worked with JCU to design a stratified sampling approach that identified more than 90,000 sample points around the world. These are labelled by experts at JCU using Ai2’s OlmoEarth Studio, a browser-based tool built for fast, high-volume annotation with AI-assisted feedback and an active learning workflow. So far, more than 60,000 of these sampling points have been labelled.

Because EcoTrain will be openly available, all partners contribute to its technical specifications so it is useful beyond the Global Ecosystems Atlas.

OlmoEarth Studio, a browser-based tool built for fast, high-volume annotation with AI-assisted feedback and an active learning workflow.

One map from many

New and existing maps come together through Esri’s scripted cloud-based processing pipeline, which harmonises them into a single synthesis map and accompanying data products.

JCU produces the metadata and technical documentation and leads quality assurance.

How we get data to users

Data only matters if it gets into the hands of people who need to use it. Alongside the Global Ecosystems Atlas platform and a version-controlled spatial data repository, data products will be shared through the Esri ArcGIS Living Atlas, and made publicly available through Google Earth Engine's Data Catalogue. This means analysts and others can work with the data using the same tools they are already familiar with.

Built with countries

At every stage of development, maps are produced with national authorities. The Global Ecosystems Atlas team has engaged with up to 100 national contributors during bilateral meetings, as well as through CBD regional dialogues, partner convenings and other third-party events.

In addition, Esri works directly with national data developers, ESA provides national maps, training and coordination with European countries, and IUCN reaches practitioners through the IUCN Academy and its expert networks via regional offices and the Commission on Ecosystems Management.

For countries that do not yet have their own ecosystem maps, the Global Ecosystems Atlas can facilitate in-country mapping accelerators: structured programmes that pair countries' own scientists and institutions with international technical partners to build maps together.

In the Maldives, international ecosystem scientists and technical partners, including JCU, Ai2 and Google, provided the global technical standards and modelling tools, including OlmoEarth Studio and Google Earth Embeddings. Maldivian ministries and agencies brought first-hand knowledge of the islands, their ecosystems and how they change season by season. Knowledge from the ground was then critical to drive the mapping model, define and describe ecosystem classes, collect example locations and iteratively improve predictions.

What comes next?

The Global Ecosystems Atlas core team and partners will continue to coordinate and improve the shared product, its documentation, and its delivery. Together, the team and many others in the sector will work to address the important challenge of including ecosystem condition and change over time. The team will build on the existing scientific oversight structures, expand and develop partnerships with national and regional institutions, further align with private-sector user base, and ensure that local experts, whose knowledge and validation efforts are essential, remain actively engaged.

Join the early adopters or contribute to the Global Ecosystems Atlas

Organisations or companies working on ecosystem mapping, national reporting or nature-related disclosure can help us co-design pilots and test methods. Contact us: ecosystems.atlas[at]geosec.org