NDVI, Copernicus & Satellite Verification

NDVI (Normalized Difference Vegetation Index) is a numerical indicator derived from satellite imagery that estimates vegetation “greenness” and vigor.

Formula

$$NDVI=(NIR-Red)(NIR+Red)NDVI\; =\; \backslash frac\{(NIR\; -\; Red)\}\{(NIR\; +\; Red)\}$$

• Red band: vegetation absorbs this for photosynthesis

• Near-Infrared (NIR): vegetation reflects this strongly if healthy

Value range

Key point: NDVI does not detect “trees” directly — it detects photosynthetically active biomass.

 

Copernicus & Sentinel-2: what data you actually get

 

Sentinel-2 key specs

• Spatial resolution:

  • 10 m (Red, NIR) → NDVI is 10×10 m per pixel

• Temporal resolution:

  • Every 5 days globally

• Cost: Free, open data (huge win for DAOs)

• Bands: 13 spectral bands (visible + infrared)

This means:

• One pixel ≈ 100 m²

• You cannot see individual saplings

• You can see land cover trends over time

What NDVI can verify 

Strong, defensible claims

These are blockchain-safe claims you can attach to smart contracts:

1. Vegetation increase over time

“NDVI increased from 0.21 → 0.47 over 18 months in polygon X”

This is robust, auditable, and repeatable.

2. No-regression guarantees

“NDVI did not fall below baseline for 12 months”

Useful for:

• Anti-deforestation clauses

• Delayed DAO payouts

3. Seasonal survival trends

By comparing same season year-over-year, you can show:

• Whether vegetation persists

• Whether the area is stable or degrading

4. Area-based verification

You can verify:

• Hectares showing sustained vegetation

• Continuity of canopy cover, not just spot planting

Canopy cover analysis (beyond basic NDVI)

NDVI alone is not enough for forest-quality claims. Typically you combine:

Techniques used in practice

1. NDVI thresholding

   • e.g. NDVI > 0.5 = likely canopy

2. Time-series smoothing

   • filters out noise & weather artifacts

3. Change detection

   • ΔNDVI between years

4. Masking

   • exclude water, agriculture, urban areas

What canopy analysis can say

• “This area transitioned from low biomass to sustained canopy”

• “Vegetation density increased and stabilized”

What it cannot say

• Tree species

• Forest age

• Biodiversity quality

• Native vs invasive plants

Limitations you MUST be explicit about (to avoid greenwashing)

1. Cloud cover 

• Optical satellites cannot see through clouds

• Tropical regions can lose 30–60% of observations

• Mitigation:

  • Cloud masking

  • Multi-month aggregation

  • DAO rules that accept “data gaps”

On-chain implication:
Smart contracts must tolerate missing data windows.

---

2. Resolution constraints (10 m)

• Young trees <2–3 years may be invisible

• Agroforestry can look like forest

• Shrubs can inflate NDVI

You cannot claim:
“10,000 trees planted”

---

3. Species identification is impossible

NDVI:

• Does not know what is growing

• Does not know why it’s growing

A monoculture plantation can score higher NDVI than a biodiverse forest.

Mitigation strategies

• Combine with:

  • Field audits

  • Community attestations

  • NGO validation NFTs

• Use NDVI as one oracle, not the only one

---

4. Cause vs correlation

NDVI increase ≠ successful reforestation
It could be:

• Natural regrowth

• Agricultural cycles

• Irrigation changes

Honest framing:

“This project contributed to vegetation recovery as verified by satellite indices”

Not:

“This project restored a forest ecosystem”

---

6. What claims are DAO-safe vs DAO-dangerous

Safe, verifiable, professional claims

• “Vegetation cover increased”

• “NDVI stabilized above baseline”

• “No major deforestation detected”

• “Land cover change confirmed by Copernicus”

Dangerous claims (avoid on-chain)

• Exact tree counts

• Species restoration

• Biodiversity recovery

• Carbon sequestration without models + field data