STAC and Zarr: Climate Data Standards in the Exascale Era

As modern high-performance computing systems march into the Exascale era, kilometer-scale simulations will soon become the workhorses in climate research. Each of these models produces data volumes on the order of several petabytes per simulation. Curating and providing such massive datasets to users worldwide presents a major challenge. To handle these data volumes efficiently, Earth system models (ESMs) such as ICON increasingly rely on cloud-native formats like Zarr for storing model outputs, and on STAC catalogs for FAIR (Findable, Accessible, Interoperable, Reusable) dissemination of datasets.

Represented by Florian Ziemen, Fabian Wachsmann, and Kameswar Rao Modali, the DKRZ participated in the STAC Community Sprint and the Zarr Adopter Summit, held in parallel at ESA ESRIN and in Rome from October 14–17, 2025.

During these meetings, they actively brought in knowledge of the Earth system modeling community and shaped best practice guidance to brainstorming sessions and the development of specifications aimed at fostering synergy between STAC (SpatioTemporal Asset Catalog) and Zarr.

This work marks an important step toward the future adoption of modern data standards such as STAC and Zarr within the geoscience community, paving the way for efficient, interoperable, and future-ready use of Exascale climate data.