Skip to content

Vegetation indices (NDVI, GNDVI, NDRE)

Vegetation indices are calculated values that reflect the condition of the crop (healthy, stressed, weed, diseased). DroneField supports three classical indices: NDVI, GNDVI and NDRE.

All three require a multispectral drone (P4M or M3M). RGB-only drones (P4R, M3E) cannot produce them.

NDVI — Normalized Difference Vegetation Index

Section titled “NDVI — Normalized Difference Vegetation Index”

The classic “how good is the stand” measure. From the NIR (Near Infrared) and Red bands:

NDVI = (NIR - R) / (NIR + R)

Range: −1 … +1. Positive values indicate photosynthesis; around 0 indicates bare soil, water or shadow; negative values are rare on ground imagery.

NDVI works best on dense stands. A saturated canopy (>0.85) is no longer differentiated — everything “very green” gets pushed to ~1.0.

Uses Green instead of Red:

GNDVI = (NIR - G) / (NIR + G)

More sensitive to chlorophyll content and nitrogen stress. Better at separating healthy and very-mature plants than NDVI.

Uses the Red Edge band instead of Red:

NDRE = (NIR - RE) / (NIR + RE)

The most accurate index for mature, dense stands — it does not saturate where NDVI does. The go-to index for precision N fertilisation. Use this on mature corn, wheat or rapeseed.

Video: NDVI or NDRE? Pick the wrong index and dense crops all look “healthy” while problems hide under the canopy:

Indices are part of the processing flow:

  1. Enable Process multispectral images in the Process panel.
  2. At the end of processing all three indices appear in the Layers/ folder.
  3. The map shows a new NDVI layer (NDVI is active by default; you switch between indices in the side panel).

When the NDVI layer is on, a floating panel appears at the bottom-right of the map:

  • AI analysis — AI-assisted zoning and a draft spray plan from the Luna assistant (see AI vegetation analysis)
  • Index selectorNDVI, GNDVI, NDRE (switches instantly)
  • Range — min/max for the visualisation
  • Preset ranges — one-click settings
  • Colour ramp — legend underneath
  • Export — PNG / JPEG / GeoTIFF (QGIS)
  • Process — for on-demand index processing (see below)

The full index range (−1 … +1) is usually too wide for visual interpretation. DroneField offers four one-click presets:

Highlights the small NDVI differences between bare soil and weeds. Range: roughly 0.05 … 0.30.

Use it for a spring field, to see where weed populations are springing up between rows.

For detecting stressed canopy (water, nutrient, disease). Range: roughly 0.30 … 0.65.

Use it for summer rapeseed, corn — where the canopy is already dense but yellowing in places.

Fine-grained view of the mature, healthy canopy. Range: roughly 0.65 … 0.90.

Use it for dense stands — e.g. pea before flowering — to see where development is faster / slower.

Focuses on near-zero and negative values. Range: roughly −0.20 … 0.10.

Use it for standing water detection, bare-soil patches.

Every preset uses a six-step colour ramp:

Index valueColourMeaning
LowestDark redWater, soil, bare
LowRedStress
Mid-lowOrangeWeak
Mid-highYellowAverage
HighYellow-greenHealthy
HighestDark greenStrong, dense

The ramp maps linearly onto the chosen range. With a range 0.30 … 0.65, 0.475 is yellow.

The panel’s Range slider accepts any min/max. Useful for:

  • Centring on a decision threshold (e.g. NDRE 0.40–0.65 = treatment zone for N fertilisation)
  • Matching your field’s typical range

Changes take effect immediately — you see the result live on the map.

If you ran processing without the multispectral switch (e.g. quick orthomosaic only), you can request the indices later:

  1. Click Process in the vegetation index panel.
  2. The list of missing bands appears (e.g. “NDRE — missing bands: RE, NIR”).
  3. Click Continue.
  4. DroneField processes only the missing indices without redoing existing work.

Faster than rerunning the whole processing.

The panel’s Export button offers three formats:

The visualisation is saved as an image with the chosen colour ramp. No georeference — a simple picture, fine for email or presentations.

The most valuable export. A .tif (GeoTIFF) file:

  • Float32 depth — every pixel is the actual index value (e.g. 0.6234)
  • Georeferenced in WGS84, EPSG:4326
  • Loads directly in QGIS, ArcGIS, ENVI and other GIS programs

QGIS auto-detects the projection on import. You can re-render the values with any colour ramp.

See Exporting to QGIS for details.

Export/
├─ NDVI/
│ ├─ ndvi_visualisation.png
│ ├─ ndvi_visualisation.jpg
│ └─ ndvi_geotiff.tif
├─ GNDVI/
│ └─ ...
└─ NDRE/
└─ ...
  • Don’t base decisions on NDVI alone — no index tells the whole story. Soil sampling and ground inspection always complement it.
  • Don’t compare two different drones — P4M and M3M filter spectra differently; values are not 1:1 comparable.
  • Don’t mix morning and afternoon flights within one project — changing solar angle shifts index values.