Earthmover Selected to Power ARIA’s “Forecasting Tipping Points” Simulation Catalogue

Earthmover is proud to announce that we have been selected by the Advanced Research + Invention Agency (ARIA) to provide the Simulation Catalogue for the Forecasting Tipping Points programme.

Greenhouse gas concentrations simulated by NASA. Image credit: NASA.

ARIA is the UK’s new R&D funding agency, designed to take bold bets on transformative science. Their Forecasting Tipping Points programme is one of their most ambitious yet. It brings together 26 independent modelling and observing teams with a single, critical goal: to establish the scientific framework for an early warning system for climate tipping points—giving us the foresight needed to protect our communities and economies from the most severe and abrupt consequences of a changing climate.

To succeed, these teams need more than just raw data and compute power; they need a way to share, discover, and analyze multi-petabyte collections of Earth System Model (ESM) simulations without the logistical “data toil” that traditionally slows down large-scale climate science.

That’s where Earthmover comes in.

The Challenge: From Data Overload to Collaborative Insight

The 26 teams funded through the Forecasting Tipping Points programme–including the UK National Oceanography Center, the UK National Centre for Atmospheric Science, and Cambridge University—will generate massive volumes of scientific data – including climate simulations, satellite observations, and drone measurements. Traditionally, managing this variety of data at the petabyte scale would require a custom development effort that could take months to years.

ARIA needed a solution that was:

• Beyond Tabular Data Lakes: Traditional data lakes designed for tabular data fail when handling the high-dimensional, tensor-based outputs from climate models and Earth observation systems.
• Analysis-Ready Cloud-Optimized (ARCO): Moving beyond simple file storage to a cloud-native, tensor-aware data lake that preserves the performance and insights of tensor data.
• Centralized yet Interoperable: ARIA required a federated solution, seamlessly integrating with both on premises storage (the NERC JASMIN infrastructure and the CEDA archive) as well as commercial cloud.
• Collaborative and Dynamic: The ARIA catalog is not static. Twenty-six teams need to build upon each other’s work in real-time, collaborating and sharing throughout the data life cycle.

Why Earthmover?

Our team at Earthmover is behind the foundational open-source technologies—Pangeo, Xarray, and Zarr—that have become the global standards for scientific data analytics and modeling at the petabyte scale. The Earthmover platform is the commercial culmination of that experience.

The ARIA programme chose the Earthmover platform for these key reasons:

1. Day 1 Value

Unlike custom builds that take months to develop, the Earthmover platform is production-ready today. We are deploying on top of JASMIN’s S3-compatible object storage immediately, allowing ARIA researchers to start doing science on day one, not attending software design meetings. This deployment highlights the platform’s native support for hybrid cloud architectures—bridging the gap between existing on-premise storage like JASMIN and the elastic compute resources of the public cloud. By meeting researcher teams wherever their data lives, we ensure that infrastructure is an accelerator rather than a bottleneck.

2. “GitHub for Data”

One of the core pillars of the Earthmover platform is a catalog and governance model that researchers already understand. By providing a GitHub-like experience for data—complete with organizations, repositories, and robust role-based access control—we have the potential to turn the catalogue into the “connective tissue” of the entire programme. Most importantly, Earthmover provides a self-service environment where scientists can manage their own data iteration cycles without needing to wait on IT departments or infrastructure specialists. In effect, Earthmover provides similar turn-key experience for scientists working with data that GitHub provides for engineers working with code.

3. High-Performance Tensor Storage with Icechunk

The catalogue will leverage Icechunk, our open-source transactional storage engine, to provide a data layer that is AI-ready from the start. By bringing database-style ACID transactions and version control to multidimensional tensors, Icechunk allows teams to safely update datasets and extract the complex spatio-temporal data required for modern AI/ML training up to 100x faster than traditional NetCDF-based cloud workflows.

A Living Research Environment

The partnership between ARIA and Earthmover is about more than just building a repository; it’s about creating a living, connected research environment. Unlike static archives, this catalogue is designed for the modern scientific lifecycle—where data is constantly updated, versioned, and instantly available for cross-team analysis.

By pairing Analysis-Ready, Cloud-Optimized (ARCO) data with scalable, data-proximate computing, we are providing a production-ready foundation for AI/ML training and complex multi-model and model-observation comparisons. This architecture eliminates the need for teams to download massive datasets, so they can move directly to the work that matters: identifying the early warning signals that could protect humanity from abrupt climate changes.

Looking Ahead

We are honored to partner with ARIA and the NERC JASMIN facility on this mission-critical project. This engagement validates Earthmover’s mission: to empower people to use scientific data to solve humanity’s greatest challenges.

We look forward to sharing more updates as the catalogue grows and the first insights from the Forecasting Tipping Points programme begin to emerge.

Interested in how Earthmover can accelerate your team’s scientific workflows? Contact us to learn more or explore our platform documentation.