NDVI satellite verification for carbon credits: a primer for buyers

The problem with static ratings

Most carbon credit quality scores are static. A panel of analysts assesses a project at a point in time, issues a rating, and updates it once or twice a year. The rating cannot tell you whether the forest is still standing this week. For a buyer making a multi-million pound commitment, that gap is uncomfortable.

“

Sylvera tells you the rating. ClimSen shows you the evidence — refreshed every week.

What NDVI is

The Normalised Difference Vegetation Index (NDVI) is a value between -1 and +1 calculated from satellite imagery. It measures the difference between near-infrared light (which healthy vegetation reflects strongly) and red light (which vegetation absorbs for photosynthesis). Bare soil and water return NDVI values below 0.2. Sparse vegetation, grassland or cropland is around 0.3 to 0.5. Dense, healthy tropical forest is 0.7+.

The European Space Agency's Sentinel-2 satellites photograph every point on Earth every 5 days at 10-metre resolution. The data is free and public. Anyone can calculate NDVI for any plot of land in the world. That is the foundation of modern remote-sensing carbon verification.

How ClimSen uses it

For every land-use project on the marketplace (afforestation, reforestation, REDD+, mangrove restoration, agroforestry), ClimSen runs a weekly Sentinel-2 NDVI scan over the project area. The first scan establishes a baseline. Every subsequent scan compares the current vegetation density against that baseline.

If NDVI drops by more than 8% versus baseline, our system creates an anomaly trigger. A human reviewer then assesses whether the drop reflects a genuine integrity issue (illegal logging, fire, drought) or an expected seasonal pattern. If the trigger is confirmed, the project's Premium Score is reduced and may fall below the 70-point threshold required for marketplace listing.

1. 1

   Weekly cron triggers a Sentinel-2 scan

   Every Monday at 05:00 UTC, our scheduler kicks off a scan across all monitored projects. Only nature-based project types (ARR, REDD+, mangrove, agroforestry, IFM) are queued — industrial sites are excluded.

2. 2

   ClimSen fetches NDVI stats over a 4 km² area

   We pull the past 30 days of Sentinel-2 L2A imagery, mask out clouds, and compute mean NDVI per pixel inside the project's centroid box.

3. 3

   Result is compared against baseline

   The first scan establishes a baseline NDVI for the project. Every subsequent scan is compared to that baseline. A drop of more than 8% creates an anomaly trigger.

4. 4

   Human review on every score change

   Triggers never auto-modify the Premium Score. They appear in /admin/twin for review. A ClimSen analyst confirms or rejects the change.

5. 5

   Buyer-facing page reflects the verdict

   If confirmed, the project's Premium Score drops. If the new score falls below 70, the project is delisted from the marketplace and existing buyers are notified.

What NDVI cannot do

• It cannot measure carbon stock directly — only vegetation density as a proxy
• It cannot distinguish a productive plantation from native forest of the same density
• It cannot see through dense cloud cover, which limits monitoring in tropical wet seasons
• It does not work for industrial carbon removal projects (DAC, biochar, mineralisation) which have no vegetation to monitor

Why this still matters

NDVI is not a complete answer to carbon credit verification. But it is the only continuously updated, independent, and publicly verifiable signal that exists today for land-use projects. For a CSRD-disclosed credit, having weekly satellite verification on the project page is the difference between 'we trust the registry' and 'we have ongoing independent evidence the carbon is still where the project says it is'.