SCIENCE UNDER AUTOMATION
A foundation model for the Earth system
Aurora evaluates forecasts of air quality, ocean waves, cyclone tracks and weather, reporting advantages over operational comparisons.
What this result supports
Training and initial conditions depend on established observational and numerical infrastructure. Forecast skill is separate from collecting samples or testing a new causal explanation.
Evidence record: Peer-reviewed forecasting evaluation. Checked: 2026-10-21.
Where this evidence is used
Earth & environmental science
- Produce a forecast from available state information: Aurora demonstrates several forecasting applications with established data infrastructure.
- Check spatial validity and model outputs: Tool-task scoring and forecast comparisons support checks within their evaluated domains.
- Design and carry out physical sampling: The evaluated systems consume existing observations and API data; field collection remains external to those systems.
- Validate across sites, instruments and conditions: Transfer requires checks beyond the particular forecasting tasks and regional API coverage.