Digital agriculture: satellites, NDVI and decision tools
How satellites see crops, what NDVI measures and how to calculate it, what digital tools can do for farmers, and the questions of data and access.
On this page
Every few days, satellites photograph almost every farm field on Earth, in wavelengths the eye cannot see. Combined with weather data, soil maps, sensors in fields and phones in farmers' pockets, these images feed digital agriculture: tools that estimate crop health, water use and yield, warn of pests and droughts, and advise on what to do.
Remote sensing and digital tools are now part of agronomy, agricultural engineering and extension syllabi, and of policy debates on data and access. This article covers how satellites see crops, the most-used index (NDVI), and what the tools are good for.
How a satellite sees a crop
Sunlight reaching a leaf is absorbed, transmitted or reflected, and the pattern depends on wavelength:
- Visible light is mostly absorbed by chlorophyll, especially blue and red, with a little more green reflected, which is why leaves look green.
- In the near infrared (NIR), from about 700 to 1,300 nm, healthy leaves reflect strongly, because light scatters inside the spongy mesophyll.
- The sharp rise between red and NIR, around 700 to 750 nm, is the red edge; it shifts as chlorophyll changes and is used to detect stress early.
- In the shortwave infrared, water in leaves absorbs strongly, so these bands track plant water content.
Bare soil reflects gradually more with wavelength; water reflects little and almost nothing in the NIR.
Vegetation indices: NDVI and others
Because healthy vegetation is dark in red and bright in NIR, combining the two bands gives a sensitive measure of green, living cover. The most widely used is the normalised difference vegetation index:
NDVI ranges from −1 to +1. Dense, healthy vegetation typically gives about 0.6 to 0.9; sparse or stressed crops 0.2 to 0.5; bare soil around 0.1 to 0.2; water and snow near zero or negative.
Calculating NDVI from reflectances
Using reflectance values like those in the figure:
- Healthy crop: red 0.04, NIR 0.48. NDVI = (0.48 − 0.04) ⁄ (0.48 + 0.04) = 0.44 ⁄ 0.52 = 0.85.
- Stressed crop: red 0.10, NIR 0.30. NDVI = 0.20 ⁄ 0.40 = 0.50.
- Bare soil: red 0.20, NIR 0.28. NDVI = 0.08 ⁄ 0.48 = 0.17.
On the Sentinel-2 satellites, red is band 4 (about 665 nm) and NIR band 8 (about 842 nm), both at 10 m resolution; on Landsat 8 and 9, red is band 4 and NIR band 5. With images of one field taken through the season, NDVI rises with canopy growth, peaks near flowering and falls at senescence, so its curve tracks the crop's development.
NDVI has limits. It saturates in dense canopies (an LAI above about 3 to 4 gives similar values), is affected by soil background in sparse crops, and shows that something is wrong without saying what. Other indices address these: the EVI (enhanced vegetation index) reduces saturation and atmospheric effects; NDRE uses the red edge for chlorophyll and nitrogen status; NDWI and related indices track water.
Sources of data
| Source | Typical resolution | Revisit | Notes |
|---|---|---|---|
| Landsat (since 1972) | 30 m | about 8–16 days | free; the longest record, used for change over decades |
| Sentinel-2 (Copernicus) | 10–20 m | about 5 days | free; red-edge bands; the workhorse of field monitoring |
| MODIS and similar | 250–1,000 m | daily | free; regional crop and drought monitoring |
| Commercial constellations | about 0.5–5 m | daily or more | paid; small fields, detailed monitoring |
| Radar (SAR), e.g. Sentinel-1 | 5–20 m | about 6–12 days | sees through cloud; crop type, flooding, rice mapping |
| Drones | centimetres | on demand | very detailed, small areas |
Cloud cover is the great limitation of optical satellites in the humid tropics and monsoon seasons, which is why radar, which sees through cloud, matters for rice and rainy-season crops.
What digital agriculture is used for
- Crop monitoring and scouting: maps of in-field variation that direct a farmer or adviser to problem areas.
- Crop type and area mapping for statistics, food security early warning and subsidy or insurance administration.
- Yield forecasting combining remote sensing with weather and crop models.
- Water: estimating actual evapotranspiration from thermal and optical data (for example FAO's open WaPOR database), irrigation scheduling and water accounting.
- Drought and flood monitoring, and index-based insurance that pays out on satellite-measured triggers.
- Precision agriculture: prescription maps for variable-rate seeding, fertiliser and irrigation, using satellite, drone, yield-monitor and soil-sensor data.
- Advisory services: weather, pest and price advice delivered by text message, voice, apps and video, increasingly with AI-based chat and image diagnosis.
- Traceability and compliance: proving that produce did not come from recently deforested land, now required by some import rules.
Making it useful
Digital data creates value only when it leads to a better decision at the right time. The questions that decide whether a tool helps:
- Ground truth: has the satellite signal been checked against field observations for this crop and region?
- Actionable advice: does the tool say what to do, and is that action available and affordable?
- Timeliness: does the information arrive before the decision has to be made?
- Fit: does it work on small, mixed and irregular fields, in local languages, with intermittent connectivity?
- Cost and business model: who pays, and is it sustainable once a project ends?
Data, access and fairness
Digital agriculture raises questions beyond agronomy:
- Data ownership and use: who owns data collected on a farm by machines, apps and sensors, and who profits from it? Voluntary codes of practice exist in several regions, but legal protections vary widely.
- The digital divide: smallholders, women and remote communities often have less access to phones, connectivity, digital skills and credit. Tools designed without them can widen gaps.
- Concentration: platforms that combine inputs, data, advice and marketing can gain strong market power.
- Public goods: free satellite data (Landsat, Copernicus) and open tools have greatly widened access; many governments now invest in digital public infrastructure for agriculture such as farmer registries and land records.
Key points
- Healthy leaves absorb red and reflect near-infrared light strongly; the red edge lies around 700–750 nm.
- NDVI = (NIR − Red) ⁄ (NIR + Red); about 0.6–0.9 for dense healthy crops, 0.1–0.2 for bare soil.
- NDVI saturates in dense canopies and shows stress without its cause; EVI, NDRE and water indices complement it.
- Free data: Landsat (30 m), Sentinel-2 (10 m, about 5 days); radar sees through cloud.
- Uses: monitoring, crop mapping, yield forecasting, water and drought monitoring, insurance, precision agriculture, advisory and traceability.
- Value depends on ground truth, actionable and timely advice, fit to farmers, and fair data and access.
What the exams ask
- Agricultural engineering, agronomy and remote sensing courses: spectral signatures, vegetation indices, sensors and resolutions, and NDVI calculations.
- Research-fellowship and postgraduate entrances: red-edge and thermal sensing, saturation, radar for crop mapping, yield forecasting and ET estimation.
- Civil-service and policy examinations: digital agriculture for smallholders, data governance, digital public infrastructure and the digital divide.
Described in general terms: the kind of examination, not a past paper. Laisha never reproduces exam questions.
Questions people ask
Can satellites tell me what is wrong with my crop?
They can show where the crop is weaker and when the problem started, which is valuable for scouting. Identifying the cause, whether disease, pests, nutrients, water or waterlogging, still needs a visit, a sample or other data.
Are satellite images useful for very small farms?
Free 10 m imagery covers a 0.1 ha plot with about ten pixels, enough for rough monitoring. For detailed work on small or mixed plots, higher-resolution commercial images or drones help, often accessed through service providers.
Why does my NDVI map look patchy even in a good field?
Variation in soil, water, sowing density and drainage is normal. The useful question is whether the pattern is consistent across seasons and whether it corresponds to something you can manage.
Sources
- Jones, H. G. and Vaughan, R. A. Remote Sensing of Vegetation: Principles, Techniques and Applications. Oxford University Press, 2010.
- Tucker, C. J. "Red and photographic infrared linear combinations for monitoring vegetation." Remote Sensing of Environment 8, 127–150 (1979).
- European Space Agency. Sentinel-2 User Handbook. ESA, 2015.
- FAO. The State of Food and Agriculture 2022: Leveraging Automation in Agriculture for Transforming Agrifood Systems. FAO, 2022.
- FAO. WaPOR: Water Productivity through Open-access of Remotely sensed derived data. Online portal.
- World Bank. What's Cooking: Digital Transformation of the Agrifood System. World Bank, 2021.
More in latest in agriculture
Discussion
Questions, corrections and notes from readers on this article. Every comment is checked before it appears: keep it on the topic, and do not share phone numbers, links or contact details.