NDVI, NDMI and NDRE: how to read satellite maps of your fields

How a satellite "sees" a plant

A satellite does more than take a conventional colour image. It measures reflected sunlight across a number of narrow wavelength bands. Four are particularly useful in agriculture: red, red-edge, near-infrared (NIR) and shortwave infrared (SWIR).

A healthy leaf behaves in a very distinctive way. Chlorophyll absorbs red light, because the plant uses it for photosynthesis. At the same time, the internal structure of the leaf strongly reflects near-infrared. The more green, well-functioning leaves there are, the bigger the difference between these two bands. Bare soil reflects both ranges in a similar way, so the difference is small.

Indices such as NDVI, NDRE and NDMI are simply a way of turning that difference into a single number for each pixel. Thanks to normalisation (dividing by the sum of the bands), the result falls between −1 and 1 and is less sensitive to whether the day was brighter or duller.

In Poland, the most commonly used data come from Sentinel-2, part of the European Copernicus programme. The mission's two satellites pass over the same area every few days. The red and NIR bands have a resolution of 10 m, so one pixel covers 10 × 10 m, while the red-edge and SWIR bands have 20 m. The same data feed satellite field monitoring in Nirby, which calculates NDVI, NDMI and NDRE for every field.

NDVI: what it is and how to read the colours

NDVI (Normalized Difference Vegetation Index) is calculated with the formula NDVI = (NIR − Red) / (NIR + Red). If a plant strongly reflects infrared and strongly absorbs red, the numerator is large and the result approaches 1.

On the map, values are converted into colours. The most common scale runs from red through yellow to dark green. Red means low NDVI, i.e. little green biomass or bare soil. Dark green means a dense, active crop canopy. The exact palette depends on the software, so always look at the legend, not just the colour.

Absolute values only tell part of the story. An NDVI of 0.5 may be perfectly normal for winter wheat in April but a warning sign in June. What matters most is the pattern within the field: which areas differ from the rest on that particular date.

When NDVI works best

NDVI shows differences across a field best from emergence to roughly canopy closure. In this period you can see overwintering, uneven emergence, damage by game, waterlogged patches and differences in soil. Later, when the crop is dense across the whole field, values approach the upper limit and the map turns uniformly green. This is known as NDVI saturation.

NDRE: when NDVI stops showing differences

NDRE (Normalized Difference Red Edge) is calculated in a similar way, but uses the red-edge band instead of red: NDRE = (NIR − RedEdge) / (NIR + RedEdge). Red-edge light penetrates deeper into the canopy and is absorbed less than red, so the index saturates later.

In practice, this means NDRE still shows differences in a dense crop when NDVI is already high everywhere. It is also more sensitive to chlorophyll content, which is why it is often used to assess crop condition in the second half of the season, for example before the next nitrogen application.

• Cereals from stem extension onwards: NDVI is often already high across the whole field, while NDRE still separates weaker and stronger areas.
• Oilseed rape and maize at full development: dense, tall biomass at which NDVI loses sensitivity.
• Assessment before top-dressing: paler patches on NDRE may point to areas where plants are less well supplied, but you need to confirm the cause in the field.

NDRE values are naturally lower than NDVI for the same crop, so do not compare the two numerically. The red-edge band on Sentinel-2 has a 20 m resolution, so the image is somewhat less detailed than an NDVI map.

NDMI: water in plants and water stress

NDMI (Normalized Difference Moisture Index) compares near-infrared with shortwave infrared: NDMI = (NIR − SWIR) / (NIR + SWIR). Water in plant tissue strongly absorbs SWIR radiation. The more water in the leaves, the less SWIR returns to the satellite and the higher the NDMI.

NDMI does not measure soil moisture. It shows how well hydrated the plants are, which does depend on soil moisture, but with a delay. On soil without plants, the result says more about the soil surface than about the water available in the profile.

It is most useful during drought. When clearly lower values appear on the NDMI map on lighter soils, knolls or sandy strips, these are usually the places where plants are first to suffer from lack of water. They often show up there earlier than you would see on NDVI, because a plant loses water before it changes colour. Like NDRE, NDMI from Sentinel-2 relies on a 20 m band.

NDVI, NDRE and NDMI compared

In practice, NDVI is usually the starting point, with NDRE and NDMI adding context. If the same area repeatedly shows lower values across all three indices and across several seasons, that points to a persistent factor such as soil conditions, drainage or topography.

Common mistakes when reading satellite maps

Clouds, shadows and haze

Sentinel-2 is an optical satellite, so it cannot see through clouds. Fully cloudy images are usually rejected, but thin cloud, streaks and their shadows can get through. A cloud shadow looks on the map like a distinct, sharp-edged patch of low values, often with a shape that has nothing to do with the field. In cloudy weeks there may be few clear images.

Headlands, field edges and roads

A 10 × 10 m pixel on the field boundary partly covers a road, a ditch, trees or a neighbouring crop. That is why the strip along the edge often shows values that are too low or too high. On top of that, headlands are more compacted and receive uneven rates. Do not judge a field by its margins.

Saturation and comparing different dates

A uniformly green NDVI map in June does not mean the field is even. It may simply mean the index is saturated, so check NDRE. And when comparing images from different days, remember that lighting, atmospheric conditions, crop growth stage and moisture all change. Compare the pattern of patches in the field rather than values to two decimal places.

Confusing the symptom with the cause

Low NDVI can mean nitrogen deficiency, drought, waterlogging, weeds, disease, damage by game, poor emergence, compacted soil or simply a different variety on part of the field. The map cannot tell these apart. Weeds can even raise NDVI, because they are green biomass too.

From the map to a decision in the field

A satellite map is most useful when it leads to a specific action. Below is a typical sequence, from the simplest use to the most advanced.

1. Targeted scouting. Instead of walking the entire field, visit two or three areas that stand out on the map and one representative area for comparison. Check the crop, soil conditions, and any signs of disease or pests.
2. Compare with history. If the same weak patch appears over several seasons and across different crops, the cause is likely persistent — for example soil properties, drainage or topography.
3. Management zones. Use historical imagery to identify stable areas with similar yield potential. Those zones can then support sampling, fertiliser planning and VRA.
4. Soil sampling by zone. Instead of one sample from the whole field, you take samples separately in each zone. We explain how to plan this in our guide to soil sampling by management zone.
5. Variable rate application. Once you know the soil nutrient status in each zone, you can prepare a fertiliser plan and a variable rate application (VRA) map for the machine terminal.

Example: on an April NDVI map, a winter wheat field has a clearly weaker strip on a knoll. On the May NDMI map the same strip dries out first, and images from the three previous seasons show it in the same place. That is a good candidate for a separate zone, where it is worth checking soil nutrient status and pH before you decide on rates.

In Nirby, this workflow is connected end to end: zones are created from historical satellite imagery, soil test results are assigned to those zones, and the fertiliser plan can then be turned into a variable rate application map in ISOXML or Shapefile format.

Checklist: how to read a satellite field map

• Check the image date and the true-colour image: are there clouds, shadows or haze?
• Look at the legend, not just the colours. Every program may use a different scale.
• Match the index to the growth stage: NDVI early in the season, NDRE in a dense crop, NDMI in drought.
• Read the map in relative terms: look for areas that stand out from the rest of the same field.
• Ignore the strip along edges and headlands, unless the problem extends well into the field.
• Compare with earlier images and seasons: is the patch new or does it keep coming back?
• Compare the map with what you already know about the field: application and field-work history, varieties, drainage and topography.
• Go out into the field and check the cause before you change a rate or carry out an operation.
• Turn consistent differences into zones, soil samples and variable rate fertilisation.