Rice paddies emit roughly a tenth of all human methane. Draining the field cuts that by half, and farmers save water doing it. Funding stalled for a decade on one missing piece: proof. We built it.
Methane is the fastest lever we have. Rice is the biggest one, and it has been stuck on measurement while the agronomy sat solved.
In August 2024 Verra invalidated 37 rice methane projects covering 4.5 million credits, which was 99.9% of every rice credit ever issued, on additionality and verification grounds.
Proof rested on farmer surveys and occasional spot audits across tens of millions of plots smaller than one hectare. Scrutiny broke that evidence base, and registries have moved on from it.
The methodology has since been rebuilt. Rice methane avoidance is now approved by the Integrity Council under Gold Standard, and represents 0.002% of the voluntary carbon market. Demand is committed. Supply sits near zero. Measurement is the one thing standing between them.
Alternate Wetting and Drying means letting the paddy dry periodically instead of holding it flooded all season. Saturated soil starves of oxygen and breeds methanogens. Introduce air and the process halts. A perforated tube in the mud is the entire apparatus.
One of the cheapest large scale climate interventions anywhere. Agronomists have understood it since the 1990s. The farmer was always ready.
Every existing agricultural MRV stack runs on optical imagery, which needs a clear sky. Rice grows under monsoon. C‑band radar transmits its own signal and reads what comes back, so it sees straight through cloud, smoke and darkness. Water bounces the signal away and reads black. A drained field scatters and appears bright. Flooding state is legible in the physics.
Simulated cloud transit over a paddy scene. The left sensor loses the observation. The right sensor holds it.
Sentinel 1 C‑band backscatter detects standing water through cloud at ten metre resolution, revisiting every six to twelve days. Free, public, and continuous since 2014, which means a defensible historical baseline already exists for every paddy in Asia.
Low cost water level sensors on a statistically designed subset of plots, placed to maximise information gain across soil, slope and irrigation class. A few hundred sensors speak for a few hundred thousand hectares. This is what collapses cost per hectare by two orders of magnitude.
Inferred flooding regime converts to methane emissions per plot, calibrated against field measurement and aligned to the approved Gold Standard protocol so the output arrives registry‑ready.
Error propagated from sparse sensors across millions of heterogeneous plots, producing plot level estimates that carry honest confidence bounds. This layer is still open at smallholder scale, and it decides whether a credit issues.
A registry buys one thing. The error bar.
Regrow Ag, Yard Stick and Perennial serve row crop agriculture under clear skies, alongside incumbent manual survey and audit. Their architecture assumes large contiguous fields and optical imagery. Ours starts from one hectare under cloud. That is a different technical problem, which is why moving down market would mean rebuilding from the sensor up.
India. Roughly 44 million hectares of rice, majority smallholder operated, average holding near one hectare, grown under heavy monsoon cover. The hardest version of the problem, and the right place to start.
Vietnam, Thailand, Indonesia, Bangladesh and the Philippines. With India, roughly 40 million serviceable hectares and about 160 million dollars.
Carbon project developers, farmer producer organisations, agri corporates with rice supply chains, and registries. Bayer, Shell and GenZero have jointly committed to rice methane programmes built on remote sensing verification. Every one of them already holds the farmer relationships.
Roughly four dollars per hectare each season, scaling down with volume, about six percent of credit value. Predictable, low customer risk, and the model we lead with while reference deployments and unit economics establish themselves.
Where we originate directly, a share of issued credits targeting ten to twenty percent. Longer cycle, higher ceiling, greater exposure to credit price. Introduced selectively once the service layer proves out.
TRL 3. Pre seed. Our Sentinel 1 processing pipeline classifies field level flooding state across a validation district over one full growing season. Ground calibration starts in Q1 2027, our first provisional patent follows in Q2, and partnership conversations are underway. We would sooner tell you exactly where we stand.
If you develop rice carbon projects, aggregate smallholders, or buy verified methane reductions, we would like to measure a district with you.