Microsoft and Terradot described TerraNova on November 12, 2026, an AI model that uses weather-model representations and field data to help estimate enhanced rock weathering. The reported sampling gains come from preliminary simulations; they do not establish independently verified carbon removals. Terradot also lists the report with that date.
From weather forecasting to soil chemistry
According to Microsoft’s announcement, TerraNova adapts Aurora to predict chemical changes in soil using data from fields where Terradot works with farmers. Its approach combines predictions with physical soil samples rather than replacing measurements. The company says further development and extensive validation are needed.
Aurora’s research publication, published in Nature in May 2025, describes a foundation model trained on diverse geophysical data and adaptable to environmental forecasting tasks. TerraNova extends that research direction into geochemistry.
What the sampling results actually describe
The authors’ Zenodo working-paper abstract reports training and evaluation on 5,192 enhanced-rock-weathering measurements from 264 fields in the United States and Brazil, with grouped cross-validation across 64 sites. These are the authors’ reported results, rather than an independent replication.
Across the seven largest sites, the abstract reports a 75% sampling reduction at a target error of 0.75 tonnes of CO₂ equivalent per hectare per year. At the tighter target error of 0.5 tonnes, the reported reduction is 63%. The headline percentage therefore depends on the error target; it is not a uniform reduction established for every field.
Microsoft describes the findings as simulations and the research as a forthcoming preprint ahead of peer review. The Zenodo record is dated November 12 and exposes an abstract, while listing the paper’s files as embargoed until December 12, 2026.
Predicted weathering and verified net removal
Carbon accounting adds another layer beyond predicting rock reactions. The Isometric Enhanced Weathering in Agriculture protocol, version 1.3, accounts for gross removal alongside lifecycle greenhouse-gas emissions to calculate net removal. Its requirements include evidence verified by independent third parties.
Section 6.1 calls for a project design document describing relevant geochemical models and a comprehensive sampling plan. Section 6.2 requires independent project validation and verification of the net-removal statement. Those requirements explain why a reduction in samples needed for an estimate does not, by itself, demonstrate certified net removals. They are an accounting benchmark, not evidence that TerraNova has received registry approval.
Why the report matters for research institutions
For university researchers, the development raises a concrete question about external validity: how reliably can a model carry information between fields with different conditions? For research offices communicating the work, the useful distinction is between a predictive result, its supporting field measurements and independently verified net carbon removal. TerraNova’s reported gains make that distinction more consequential as AI enters environmental measurement research.
Sources
- Microsoft Source: TerraNova report, November 12, 2026. Checked 2026-11-14.
- Terradot: News and Press. Checked 2026-11-14.
- Munukutla and colleagues: Foundation models for predicting enhanced rock weathering at operational scale, v1. Checked 2026-11-14.
- Microsoft Research: A Foundation Model for the Earth System. Checked 2026-11-14.
- Isometric: Enhanced Weathering in Agriculture, v1.3. Checked 2026-11-14.
