NDVI — Vegetation Health Index
NDVI (Normalised Difference Vegetation Index) is the primary metric Canopex uses to detect vegetation change. It measures how much healthy green vegetation is present on the ground by comparing how plants reflect red and near-infrared light.
Reading the scale
NDVI values range from -1.0 to +1.0:
| NDVI Range | What It Means | Typical Land Cover |
|---|---|---|
| < 0.0 | Water or artificial surfaces | Rivers, lakes, roads, buildings |
| 0.0 – 0.1 | Bare ground, rock, sand | Recently cleared land, desert, mining |
| 0.1 – 0.2 | Sparse or stressed vegetation | Arid grassland, degraded pasture, post-fire recovery |
| 0.2 – 0.4 | Light vegetation | Crops (early growth), savanna, managed pasture |
| 0.4 – 0.6 | Moderate vegetation | Mature crops, scrubland, open woodland |
| 0.6 – 0.8 | Dense, healthy vegetation | Tropical forest, temperate woodland |
| > 0.8 | Very dense forest canopy | Primary rainforest, dense plantation |
Reading the NDVI sparkline
In the evidence panel, the NDVI sparkline chart shows the mean NDVI value for your parcel at each captured time frame (typically seasonal imagery from 2021 onwards, with historical baseline from Landsat going back to 2013).
- Flat line at 0.5–0.8 — stable forest cover. Good news for EUDR compliance.
- Gradual decline — could be drought, disease, or slow degradation. Check the weather panel.
- Sharp drop — sudden clearing event. If it happens after Dec 2020, this is the key EUDR concern.
- Seasonal waves — normal for agriculture and deciduous forest. Not a concern unless the baseline shifts down.
Climate Context — Weather Correlation
Not every NDVI decline is deforestation. Drought, flooding, and seasonal temperature changes all affect vegetation health. Canopex fetches historical weather data for your parcel's location and displays it alongside the NDVI trend.
What to look for
- NDVI drops during a dry period — likely drought stress, not clearing. The weather panel will show low precipitation in the same period.
- NDVI drops with normal rainfall — more likely human-caused clearing. Vegetation should be maintaining or growing with adequate water.
- NDVI drops after a fire hotspot — the fire/flood panel may show FIRMS fire detections in the area, confirming the cause.
Why this matters for EUDR
EUDR due diligence requires you to show that deforestation — human-caused forest loss — did not occur. Being able to attribute an NDVI decline to drought rather than clearing strengthens your compliance case. The weather data in Canopex gives you this evidence automatically.
Corroborating Data Sources
Canopex doesn't rely on a single dataset. Each assessment draws from multiple independent sources. When they agree, your determination is stronger. When they disagree, flags are raised for closer review.
| Source | What It Shows | Why It Matters |
|---|---|---|
| Sentinel-2 L2A | 10m optical imagery, NDVI time series (2017 onwards) | Primary vegetation change detection — the core evidence layer |
| Landsat C2 L2 | 30m NDVI baseline (2013 onwards) | Extends the record before Sentinel-2 existed, establishing what the land looked like years before the EUDR cutoff |
| ESA WorldCover | 10m land classification (tree cover, grassland, cropland, etc.) | Independent baseline of what land cover class the parcel was — confirms whether it was forest |
| IO Annual LULC | Year-over-year land-use change (2017–2023) | Detects transitions like forest → cropland independently from NDVI |
| ALOS PALSAR FNF | 25m radar-based forest/non-forest classification | Works through cloud cover (radar), providing an independent forest confirmation |
| FIRMS / MODIS | Fire hotspot detections | Identifies whether fires occurred in or near your parcel — relevant for burn-and-clear deforestation patterns |
| WDPA | Protected area boundaries | Flags parcels that overlap with internationally recognised protected areas — higher regulatory scrutiny |
| Open-Meteo | Historical temperature and precipitation | Distinguishes weather-driven vegetation changes from human-caused clearing |
For a full technical reference, see the EUDR Data Sources page.
Deforestation-Free Determination
After analysing all evidence layers, Canopex produces an algorithmic determination for each parcel. This is the headline result you'll use in your EUDR due diligence records.
✓ Deforestation-free
No significant vegetation loss detected after 31 December 2020. No change-detection flags were raised. This parcel can be included in your due diligence statement as assessed compliant.
⚠ Potential deforestation detected
One or more flags were raised. This does not necessarily mean deforestation occurred — it means the satellite evidence warrants closer investigation. See the flags list for specific concerns.
Confidence levels
| Confidence | Meaning | What to Do |
|---|---|---|
| High | Multiple data sources agree. Sufficient time frames were analysed. | Result is reliable. Include in your due diligence records. |
| Medium | Some evidence supports the determination, but one flag or limited data creates uncertainty. | Review the specific flag. Consider supplementing with ground-level verification. |
| Low | Insufficient satellite data to make a reliable assessment (e.g., persistent cloud cover, limited imagery availability). | Do not rely on this determination alone. Seek alternative evidence sources or ground-level verification. |
Common flags and what they mean
"Vegetation loss X% (Y ha) in [period]"
The change detection algorithm found a significant drop in NDVI between two seasonal comparisons. Check the weather panel — if there was a drought in the same period, the loss may be natural. If rainfall was normal, this could indicate clearing. The percentage and hectarage help you judge the scale.
"Overall NDVI trajectory is declining"
Across all captured time frames, the trend is downward. This suggests ongoing degradation rather than a single clearing event. Consider whether the parcel is in a region experiencing long-term drought or whether it reflects progressive land-use change.
"IO LULC detected significant land-cover change"
The independent IO Annual LULC dataset — which classifies land use globally every year — found a transition on this parcel (e.g., from "forest" to "cropland"). This corroborates the NDVI evidence and increases concern.
"IO LULC tree cover trend is declining"
The year-over-year tree cover percentage from the IO LULC dataset shows a downward trend, suggesting progressive forest loss.
"Overlaps a WDPA protected area"
Your parcel falls within an internationally recognised protected area. This doesn't affect the deforestation determination, but it increases regulatory scrutiny — production in protected areas may face additional legal requirements beyond EUDR.
"Mean NDVI delta below threshold"
The average change in NDVI between compared time frames is more negative than the detection threshold. This is a quantitative signal that vegetation health across the parcel is worsening.
What to Do With Flagged Parcels
A flagged parcel is not automatically non-compliant. It means the satellite evidence needs human review. Here's how to work through a flagged result:
- Read the specific flags. Each flag tells you what was detected and when. A "vegetation loss 8% in Q2 2023" is different from "overall trajectory declining."
- Check the weather context. Was there a drought or extreme weather event in the flagged period? If so, the vegetation change may be natural.
- Review the imagery timeline. Use the RGB layer and timeline slider to visually inspect the parcel before and after the flagged period. Look for obvious signs of clearing (geometric patterns, exposed soil) versus natural change (browning from drought, seasonal leaf loss).
- Cross-reference with other sources. Did the IO LULC and ALOS radar layers also flag changes? If multiple independent sources agree, the signal is stronger. If only one source flagged it, it may be noise or a sensor artefact.
- Consider local knowledge. If you or your supplier have ground-level information (planned replanting, controlled burns, infrastructure development), factor that in. EUDR concerns deforestation specifically — not all vegetation change is deforestation.
- Document your reasoning. Whether you conclude the parcel is compliant (natural change, not deforestation) or non-compliant (actual clearing after Dec 2020), record your reasoning. The EUDR audit PDF with scene IDs provides the satellite evidence; your recorded interpretation completes the due diligence file.
Working With Your Export Files
EUDR Audit PDF
The PDF report contains everything an auditor needs: parcel locations, NDVI time-series data, per-parcel deforestation-free determinations with confidence levels, flag details, satellite scene IDs for reproducibility, and a methodology summary. Keep this document in your due diligence records alongside your supplier documentation.
GeoJSON
A machine-readable file containing your parcel geometries with all computed metrics as properties. You can open this in QGIS, ArcGIS, or any GIS tool for further analysis. You can also load it into compliance databases that accept geospatial formats.
CSV
A tabular export with one row per parcel and columns for all key metrics (NDVI mean, area, determination, confidence, flags). Open in Excel, Google Sheets, or any spreadsheet tool. Useful for creating summary tables or importing into your own compliance tracking systems.
eudr-pdf, eudr-geojson, eudr-csv) include additional EUDR-specific fields like the cutoff date reference and regulation citation.Need more help?
Check the methodology or start your next assessment.