EVI vs NDVI: Formulas, Differences and When to Use Each

EVI vs NDVI comes down to one trade-off. NDVI uses only red and near-infrared light, so it is simple and has a NOAA record back to 1981, but it tends to saturate over dense vegetation and reacts to soil color (USGS). EVI adds the blue band and three correction terms, so it stays more sensitive in dense canopy and corrects for some atmospheric and background noise (USGS).

Use NDVI for early-season crops, long-term comparisons and quick field checks. Use EVI when a closed canopy looks flat on an NDVI map. This guide gives both formulas, a side-by-side table, a worked chart of the saturation effect and a calculator for your own pixel values.

EVI vs NDVI at a glance

The table sums up the differences that change a decision on the farm. Each row is explained further down.

What vegetation indices measure

A vegetation index turns light measured by a satellite, drone or aircraft into one number per pixel. Healthy leaves absorb much of the red light they receive and reflect much of the near-infrared. The bigger the gap between those two bands, the greener and denser the canopy tends to be.

An index measures greenness and canopy density. It does not name a cause. A low patch may be water stress, nutrient shortage, disease, pest damage, poor establishment or a planter skip. The map tells you where to look, and a field visit tells you why.

That matters when you compare EVI and NDVI, because neither one diagnoses a problem. They differ in how well they keep showing differences in green cover under certain conditions.

▶ How satellite data moved farming from NDVI to EVI and hyperspectral imaging · 10:49

How NDVI works

NDVI dates to the NASA ERTS research of the early 1970s, including a 1974 paper on monitoring Great Plains vegetation, given at a NASA ERTS-1 symposium. It remains the common starting point, and a 2000 NASA article calls it the most common measurement on global vegetation maps.

The USGS gives the formula as (NIR − R) / (NIR + R), where NIR and R are surface reflectance in the near-infrared and red bands.

Reading NDVI values

The index always falls between −1 and +1. The ranges below come from the USGS and should be treated as a rough guide, not as fixed classes for one crop.

For a longer crop-by-crop reading, see our guide to NDVI ranges.

Strengths of NDVI

• It needs only two bands, so any camera or satellite with red and near-infrared bands can produce it. A standard RGB drone camera cannot.
• It is easy to calculate by hand or in a spreadsheet.
• The USGS notes it can compensate for changing illumination, slope and viewing angle, which helps at continental and global scale.
• It has a long record. NOAA’s daily NDVI climate data record runs from 1981 on a 0.05° global grid.

Limits of NDVI

• It flattens out in thick canopy. Nebraska Extension says it tends to saturate after canopy closure, which hides variation across the field late in the season.
• It responds to soil color when cover is thin.
• The formula has no term for haze, so images need atmospheric correction before you compare dates.

▶ What NDVI shows about crop health and how to use it · 7:48

How EVI works

The MODIS science team built EVI to improve on NDVI. A 2000 NASA article says it does not become saturated as easily over areas with large amounts of chlorophyll and corrects for some distortion from particles in the air and from the ground below the plants.

The general form is EVI = G × (NIR − R) / (NIR + C1 × R − C2 × B + L). Standard values are G = 2.5, C1 = 6, C2 = 7.5 and L = 1, as shown in the USGS Landsat EVI formula.

Bands to use on each satellite

• Landsat 8 and 9: bands 5, 4 and 2 for NIR, red and blue.
• Landsat 4 to 7: bands 4, 3 and 1.
• Sentinel-2: B08, B04 and B02, all at 10 m.
• MODIS: NASA already publishes ready-made EVI and NDVI layers in the MOD13Q1 product, so you do not need to calculate them.

Use surface reflectance as a fraction between 0 and 1. The constant L = 1 assumes that scale, so raw digital numbers give wrong results. Sentinel-2 values, for example, must be converted first.

Strengths of EVI

• It keeps more contrast in high-biomass canopy. The MOD13Q1 description says it has improved sensitivity over high biomass regions.
• It has built-in terms for some atmospheric and background effects.
• It is available in the same MODIS product as NDVI, so you can view both for the same pixels and dates.

Limits of EVI

• It needs a blue band. Some drone cameras have no blue band, so check your sensor before you plan on EVI.
• Its values are not interchangeable with NDVI values. In our example below, the same pixel can read 0.78 on NDVI and 0.59 on EVI.
• Clouds still block the view. NASA is clear that neither index removes every obstacle, including clouds and aerosols.
• Bright surfaces can distort it. With snow-like example inputs of NIR 0.80, red 0.85 and blue 0.90, the denominator falls to 0.15 and the result jumps to −0.83. Mask snow and cloud first.

Key differences between EVI and NDVI

The five differences below decide which index belongs on your map.

1. Sensitivity in dense canopy

NDVI’s weakness is structural. Once a closed canopy has absorbed nearly all the red light, red has little room left to fall. Nebraska Extension explains that red absorption reaches a peak regardless of further biomass. Because near-infrared sits in both the top and bottom of the NDVI ratio, further gains barely move the result.

The chart runs both formulas on the same six example pixels as a canopy thickens. From the fifth step to the sixth, NDVI rises by less than 0.03 and EVI by about 0.10. The inputs are round example values, so compare the shape of the two lines and not the exact numbers.

2. Atmosphere

Haze scatters blue light more than red, which is why EVI uses the blue band. The USGS says the L value, the C coefficients and the blue band together help reduce background noise, atmospheric noise and saturation in most cases. It is a partial correction. It does not rescue cloudy images, and shadows remain a problem for both.

3. Soil and background

Both indices are affected by what shows through a thin canopy. EVI has the L term, but if thin cover and bright soil are your main problem, a purpose-built option exists. SAVI corrects NDVI for soil brightness where vegetative cover is low, using L = 0.5 (USGS). Consider it for young row crops, wide spacing and young orchards.

4. Data and sensor needs

NDVI needs two bands and EVI needs three. Sentinel-2, Landsat and MODIS all carry a blue band, so this mainly limits some drones. Pixel size and revisit depend on the satellite and not on the index, as the charts show. A 250 m MODIS pixel covers about 15 acres, so it suits regional work more than within-field decisions. Sentinel-2’s 10 m bands suit field zones.

5. Length of the record

Choose NDVI when you need regional history from before 2000. The NOAA record starts in 1981, and AVHRR NDVI data have supported long-term drought monitoring work. The MODIS EVI series starts on 18 February 2000. You can also calculate EVI from older Landsat surface reflectance, since the USGS produces it from Landsat 4 onward. That NOAA grid is 0.05°, about 5 km, so it is too coarse for one field. At field scale, the USGS produces both NDVI and EVI from the same Landsat scenes, so neither has a longer record there. For regional climate baselines, NDVI has the edge.

Chart

In this example, NDVI flattens as a canopy thickens while EVI rises

Index value for six example pixels, from thin to very dense canopy (our calculation)

Which index to use, by situation

EVI is not the only late-season choice. Nebraska Extension finds NDRE a better indicator than NDVI for high-biomass crops such as corn after canopy closure, because it replaces red with red edge. Sentinel-2 has red-edge bands, but standard Landsat does not.

Compare NDVI and EVI on your own pixel

Pull surface reflectance for one pixel from your image tool and enter it below. The difference between the two values is not a measure of error. It shows only that the scales are not the same.

Try the numbers for your own farm below.

Using EVI and NDVI in precision agriculture

Both indices can feed the same field jobs. The method matters more than the index: use one index per time series, keep the same sensor where you can, and ground-truth what the map shows.

Crop health monitoring

Track each zone through the season against its own earlier values. A patch that falls behind the rest of the field is a reason to scout. Nebraska Extension shows an NDVI map of a soybean field where lower values matched herbicide injury seen on the ground.

Yield estimates

Index values relate to biomass, so they are often used as an input to yield models. Treat them as one input, calibrated for your crop and region. Saturation is one reason to test EVI or NDRE on high-yield fields.

Irrigation

Falling index values can point to water stress, but the signal may arrive after the crop is already stressed, and it can have other causes. Use a moisture index, soil probes or a field visit before changing a schedule.

Fertilizer zones

Index maps can help draw variable-rate zones. A low value does not prove a nutrient shortage, so confirm with soil or tissue tests. Check the product label and local rules, and get agronomic advice before setting rates.

Pest and disease scouting

A sudden drop can flag a problem, but a satellite cannot name the pest or pathogen. Use it to direct scouting, and identify the cause in the field. Our overview of common crop diseases and pests can help with that step.

Crop stage tracking

The USGS notes that NDVI averaged over time can set a “normal” for a region at a given time of year. Comparing this season with that normal shows whether a crop is ahead or behind. Crop phenology work is possible with either index if you stay consistent.

▶ Tutorial on interpreting satellite data for agriculture in our apps · 4:01

Where to get EVI and NDVI data

• Sentinel-2 (ESA/Copernicus): 10 m visible and near-infrared bands, with a 5-day revisit at the Equator.
• Landsat 8 and 9 (USGS/NASA): 30 m bands and a 16-day repeat cycle each, offset by 8 days.
• Harmonized Landsat Sentinel-2 (NASA): a 30 m combined product with a revisit of 1.4 days with five satellites in 2025. The same NASA page gives a global median repeat of 2 days.
• MODIS MOD13Q1 (NASA): 250 m, 16-day composites with both indices ready to use.

Chart

Pixel size of common EVI and NDVI data

Metres per pixel for the visible and near-infrared bands; smaller is finer

Frequent passes do not guarantee clear images. A cloudy fortnight can leave a gap on any optical satellite. MOD13Q1 handles this by choosing the best pixel from each 16-day period, based on low cloud, low view angle and high index value. Its NDVI layer also keeps continuity with the older AVHRR series.

Chart

How often each data source updates

Days between looks; Sentinel-2 at the Equator, MODIS as composite period, before cloud losses

What to compare satellite indices with

Satellites, drones and in-field sensors answer different questions. Satellites repeat over whole farms with no flight to organize, but cloud can interrupt them and pixels are larger. Drones give much finer detail on demand, but need an operator and suitable flying weather, and some cannot measure blue. Soil and weather sensors report continuously, but each covers one point.

The options often work well together: satellites to find the problem area, then a drone or a walk to inspect it, and sensors for the moisture detail. Comparing costs needs real local quotes, because they vary widely by farm size and provider.

Common mistakes when comparing the two

• Using one threshold for both. An NDVI threshold of 0.6 does not carry over to EVI.
• Mixing sensors in one trend. Band widths differ between satellites, so a change may come from the sensor. HLS was built to reduce this problem for Landsat and Sentinel-2.
• Comparing dates without masks. Thin cloud and shadow can look like crop decline.
• Using raw digital numbers in EVI. The fixed L term needs reflectance from 0 to 1.
• Reading a color map as a diagnosis. Color ramps differ by tool. Check the legend and scout before acting.

Trying both indices on your fields

Farmonaut’s Satellite + AI Farm Monitoring System shows NDVI and EVI layers for mapped fields, so a grower can check them through the season. Setup details are in the satellite data API pages and the developer docs, or you can open the web app from our home page.

Related pages on our site: agriculture monitoring systems; regenerative agriculture; sustainable farming practices; agricultural cooperation (Russian); drill and blast monitoring.

FAQ

What is the main difference between EVI and NDVI?

NDVI uses red and near-infrared light. EVI adds blue light and correction terms. As a result, EVI is more sensitive in dense vegetation and corrects for some atmospheric and background effects (USGS).

Is EVI better than NDVI for crops?

Not at every stage. NDVI works well in early and mid season and is simpler. EVI helps most when a thick canopy makes NDVI look flat. Test both on one field before you switch.

Can EVI be higher than NDVI?

Yes, though on most example pixels here it is lower. On the densest example in our chart, EVI reads 0.97 and NDVI reads 0.93. The two are on different scales, so compare each only with itself over time.

Can I calculate EVI from a drone image?

Only if the camera records blue, red and near-infrared and you calibrate to reflectance. With no blue band, use NDVI or, if the camera has red edge, NDRE.

What does EVI mean in remote sensing?

EVI stands for Enhanced Vegetation Index. It is a satellite measure of greenness designed for the MODIS sensor, and it now also comes from Landsat and can be calculated from Sentinel-2 bands.

How often can I get new values?

That depends on the satellite, not the index. Sentinel-2 revisits every 5 days at the Equator, combined Landsat 8 and 9 every 8 days, and MODIS composites arrive every 16 days. Cloud can lengthen the gap.