Satellite Crop Monitoring and Remote Sensing

How remote sensing turns light into crop data

Every surface reflects sunlight in its own pattern across the spectrum. Healthy leaves absorb red light for photosynthesis and reflect strongly in the near-infrared; bare soil reflects more evenly; water absorbs almost all near-infrared. An optical satellite records that reflected light in a set of bands, and a vegetation or water index is simply arithmetic on two or three of those bands that makes one property stand out.

Three kinds of resolution decide what a sensor is good for. Spatial resolution is the size of each pixel: Sentinel-2’s visible and near-infrared bands are 10 m, Landsat’s are 30 m. Temporal resolution is how often the same spot is imaged: about every five days for the two Sentinel-2 satellites together, and every eight days for Landsat 8 and 9 combined. Spectral resolution is how many bands there are and how narrow they are, which is where Sentinel-2’s red-edge bands and hyperspectral sensors with hundreds of bands come in. Radar satellites such as Sentinel-1 work differently: they send their own microwave signal, so they see through cloud and at night. Our guide to remote sensing in agriculture covers the satellites in more depth.

Choosing an index for the question you have

Most confusion about indices comes from picking one before deciding what you want to know. Start from the question:

The NDWI naming trap catches a lot of people. One NDWI, built from green and near-infrared bands, maps open water. Another, built from near-infrared and shortwave infrared, measures water held in leaves and is often called NDMI to avoid the mix-up. Our posts on NDVI ranges, the NDRE formula and MNDWI vs NDWI give the formulas and the bands for each.

Reading a whole season of images

A single index map shows how a field compares with itself on one day. The useful information is in the change: a zone that was average in week four and is now lagging, or a whole field that dropped after a heat wave. That’s why remote crop monitoring works best as a time series, with each new image compared against the last few passes and against the same date in earlier seasons.

Two practical problems come with that. Clouds and their shadows corrupt optical pixels, so every image needs a cloud mask, and in wet seasons radar fills the gaps. And an index shows a symptom without naming the cause: low NDVI can mean drought, nitrogen shortage, disease, waterlogging or a planter that skipped. Someone still has to walk to the spot the map points at and find out which. Our article on comparing satellite images of the same field shows how to do this with NDVI maps.

Land classification and deforestation checks

Remote sensing is also used to sort land into classes, such as cropland, forest, grassland, water and built-up areas. Land cover describes what’s physically on the ground; land use describes what people use it for, which imagery alone can’t always tell you. Classification is done either supervised, where an analyst trains a model on known examples, or unsupervised, where an algorithm groups similar pixels and a person labels the groups afterward.

The same methods now support deforestation checks under the EU Deforestation Regulation, which requires companies placing cattle, cocoa, coffee, oil palm, rubber, soya or wood on the EU market to geolocate each plot and show it wasn’t deforested after 31 December 2020. In practice that means comparing each plot’s boundary against forest maps and imagery from before and after the cutoff. Our piece on remote sensing for EUDR compliance goes through the workflow, and our traceability software keeps the plot records alongside the supply chain data.

Putting satellite maps to work on your fields

The articles in this hub explain the science. If you want to see it applied to your own fields, our satellite crop monitoring page describes how we combine optical and radar data into 16 views per field at 10-meter detail and turn them into a field report with crop health, irrigation, soil, yield, pest risk and fertilizer advice. For a step-by-step method, our guide on how to monitor crop health using satellites goes through it in six steps, from getting the field boundary right to scouting the zones the map flags and keeping a record.