# Copernicus NDVI
What NDVI actually is and why it’s used
# NDVI, Copernicus & Satellite Verification
**NDVI (Normalized Difference Vegetation Index)** is a numerical indicator derived from satellite imagery that estimates **vegetation “greenness” and vigor**.
### Formula
[](https://wiki.zelmacorp.io/uploads/images/gallery/2026-01/ndvi.png)
- **Red band**: vegetation absorbs this for photosynthesis
- **Near-Infrared (NIR)**: vegetation reflects this strongly if healthy
### Value range
NDVI value
Interpretation
-1 to 0
Water, snow, clouds
0–0.2
Bare soil, rocks
0.2–0.4
Sparse vegetation, grass
0.4–0.6
Healthy vegetation
0.6–0.9
Dense forest canopy
**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