NDVI, NDRE, EVI: which vegetation index actually matters?

Every drone demo we have ever attended ends with an NDVI map. Green where the canopy is healthy, red where it isn't, a satisfied nod from the sales rep, a quoted price per acre. NDVI is the famous index, and for that reason it is the one most growers end up paying for. It is also, for most modern precision-ag use cases, the wrong primary choice.

This is a working practitioner's guide to which vegetation index to actually use, and when.

What an index is, briefly

A vegetation index is a mathematical combination of the reflectance values from different bands of light (visible, near-infrared, red-edge, sometimes shortwave infrared) that correlates with some property of the plant canopy — vigor, chlorophyll content, leaf area, water stress, and so on.

Different indices use different combinations of bands. The choice of index is largely a choice of which bands you weight, which controls what the index is sensitive to and where it fails.

NDVI — the famous one

NDVI = (NIR − Red) / (NIR + Red)

What it measures: broadly, canopy vigor and biomass. Healthy plants reflect a lot of near-infrared (NIR) and absorb a lot of red light (because chlorophyll absorbs red for photosynthesis); the contrast between the two bands is a proxy for "this is healthy green vegetation."

Where it works: early to mid season, when canopy is sparse to moderate. Pasture, grain crops before canopy closure, scouting for emergence issues.

Where it fails: NDVI saturates at high biomass. Once a canopy is dense — closed corn at V10, dense vineyard canopy mid-season, full vegetable canopy near harvest — NDVI flattens out and stops being sensitive to further variation. Two fields with very different chlorophyll content can show the same NDVI value because both are above the saturation point.

This is the single most important fact about NDVI, and it is the one most often glossed over in vendor demos. If your decision needs to discriminate among dense canopies, NDVI cannot give you the answer.

NDRE — the one most growers should use more

NDRE = (NIR − Red-edge) / (NIR + Red-edge)

What it measures: chlorophyll content, with much better performance in dense canopy. The red-edge band sits between red and NIR and responds to chlorophyll more linearly across the full range of canopy density.

Where it works: mid to late season in any crop where canopy is dense. In-season nitrogen status, late-season disease pressure, crop-stress detection where NDVI has already saturated.

Where it fails: requires a sensor with a dedicated red-edge band. Many consumer-grade drone multispectral cameras lack red-edge or have it as a noisy add-on. Verify the sensor specification before contracting for NDRE work.

For most modern row-crop, vineyard, orchard, and high-value vegetable operations, NDRE is a better primary index than NDVI. If you are buying a multispectral sensor in 2026, red-edge support is the spec to insist on.

EVI — the satellite-grade workhorse

EVI = 2.5 × (NIR − Red) / (NIR + 6 × Red − 7.5 × Blue + 1)

What it measures: canopy vigor, with corrections for both soil-background influence and atmospheric scattering. The blue-band term in the denominator suppresses atmospheric noise; the constants suppress soil influence at low canopy density.

Where it works: satellite imagery (Sentinel-2, Landsat, commercial constellations) where atmospheric noise and soil background are both significant. Multi-temporal analysis across a season or across years.

Where it fails: less commonly available from drone sensors (requires a blue band, which not all drone multispectral cameras carry). Overkill for low-altitude drone imagery where atmospheric correction is unnecessary.

If your work involves satellite imagery — multi-year baselines, regional-scale analysis, anything where atmospheric variation matters — EVI is the standard. If your work is exclusively low-altitude drone, NDRE or NDVI are usually sufficient and EVI buys you little.

SAVI — the underrated one for sparse canopy

SAVI = ((NIR − Red) / (NIR + Red + L)) × (1 + L)

L is a soil-adjustment factor, typically 0.5 for intermediate canopy density.

What it measures: same thing as NDVI, but with explicit correction for soil reflectance bleeding into the signal.

Where it works: early-season scouting (where bare soil dominates the pixel), arid and semi-arid systems with permanently sparse canopy, range and pasture assessment. Anywhere the soil is a large fraction of what the sensor sees.

Where it fails: dense canopy (where the soil correction adds noise without helping) and applications where you genuinely want soil influence in the signal.

If your operation includes any work in sparse-canopy systems, SAVI in the early season and a transition to NDRE or NDVI as the canopy closes is a reasonable workflow.

Crop-by-crop recommendations

Row crops (corn, soy, wheat): NDVI for emergence and early season. NDRE from V6/V8 forward, especially for nitrogen-status work. EVI for satellite baselines across the season.

Vineyards: NDRE for in-canopy chlorophyll variation, especially in the second half of the season when canopy is dense. NDVI is largely saturated and not useful.

Orchards: NDRE for tree-by-tree chlorophyll and stress. Thermal (not an index but worth mentioning) is often more useful than any of the visible-light indices for irrigation decisions.

Leafy greens and high-density vegetables: NDRE almost exclusively. NDVI saturates almost immediately.

Pasture and range: SAVI in arid systems, NDVI in temperate. EVI for satellite-scale work.

Sensor implications

Whatever index you choose, the sensor has to capture the bands the index requires. Verify before committing.

NDVI requires NIR and Red. Almost all multispectral sensors carry both.

NDRE requires NIR and Red-edge. Many lower-cost multispectral sensors do not carry red-edge, or carry it at significantly lower spatial resolution than the other bands.

EVI requires NIR, Red, and Blue. Most drone multispectral cameras carry RGB plus NIR plus red-edge but not a separate blue calibrated for vegetation index work.

If you are scoping a drone scouting program in 2026, the question to ask the integrator is not "does this drone do NDVI." Every drone does NDVI. The question is "what is the red-edge band specification and how is it calibrated."

Common mistakes

The first is defaulting to NDVI because it is the famous one. We have walked into too many operations that were paying for NDVI mapping when the underlying questions — late-season nitrogen status in dense corn, water stress in a closed-canopy orchard — could not be answered by NDVI in principle.

The second is applying satellite-era indices to drone imagery without recalibration. Index thresholds calibrated for Sentinel-2 do not transfer to drone multispectral with different band centers and bandwidths. The math is the same; the meaningful thresholds are not.

The third is using a single index in isolation. Most serious operations layer two or three indices (commonly NDRE + thermal, or NDVI + NDRE for transition periods) and look at the combination. Single-index analysis is a starting point, not a finished workflow.

The bottom line

The right vegetation index depends on your crop, your sensor, your growth stage, and what decision you are trying to make. NDVI is the famous one because it was the first one, not because it is the best one for most modern decisions. NDRE is the index most growers should be using more often, and the spec to insist on when buying a new sensor.

When a vendor's pitch deck shows you an NDVI map and stops there, ask the next question.