Researchers have evaluated five weather Foundation Models (FMs) for their effectiveness in quantifying carbon dioxide removal through enhanced rock weathering (ERW). The study, which utilized 5,192 chronosequences from 264 fields across Brazil and the US, found that these FMs can transfer their representations to ERW tasks. The best-performing model, TerraNova-ERW built on Aurora 1.5, demonstrated the ability to positively order measurements within unseen fields. While zero-shot prediction is not yet feasible due to data scale limitations, few-shot prediction significantly improves efficiency, with TerraNova-ERW recovering over 60% of high-removal fields using just two samples per field and potentially reducing sampling needs by up to 98% in dense fields. AI
IMPACT Demonstrates potential for foundation models to accelerate scientific discovery and environmental solutions.
RANK_REASON Academic paper detailing novel application of foundation models to a scientific problem. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- Aurora 1.5
- Brazil
- CatalyzeX Code Finder for Papers
- CORE Recommender
- DagsHub
- enhanced weathering
- Gotit.pub
- Hugging Face
- IArxiv Recommender
- Influence Flower
- ScienceCast
- TerraNova-ERW
- US
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →