Sentinel-2 and Landsat: two ways of seeing a Prairie field

Who runs the two programmes

Sentinel-2 is part of Copernicus, the European Union's Earth observation programme. The European Commission leads the programme in partnership with the European Space Agency. ESA develops, builds, flies and operates the Sentinel satellites.

Landsat is a joint programme of NASA and the U.S. Geological Survey. For Landsat 9, NASA was responsible for the instruments, the spacecraft, the launch and the checkout in orbit. Responsibility then passed to the USGS, which runs day-to-day operations and distributes the data.

Both programmes publish their imagery at no cost. The USGS has kept Landsat data free and open since 2008. Under European Union law, users have free, full and open access to Copernicus Sentinel data. A simple registration is needed to download products, and anyone who shares the data must credit the source.

Which satellites are in orbit in October 2026

Sentinel-2A was launched in 2015, Sentinel-2B in 2017 and Sentinel-2C in September 2024. On January 21, 2025 Sentinel-2C took over Sentinel-2A's place in the standard constellation, which now pairs Sentinel-2B with Sentinel-2C.

Sentinel-2A did not retire. It was moved to a new orbit and began an extension campaign in March 2025. The most recent notice, published in October 2026, keeps Sentinel-2A operating until December 31, 2027. Until then, Sentinel-2A, Sentinel-2B and Sentinel-2C are all collecting images.

Landsat 8 and Landsat 9 are the two Landsat satellites in operation. Landsat 9 was launched on September 27, 2021. Landsat 7 ended its official mission in 2024 and was decommissioned on June 4, 2025.

What each satellite measures

Each Sentinel-2 satellite carries one instrument, the MultiSpectral Instrument. It records 13 spectral bands between 443 and 2190 nanometres. Four bands have a 10 m pixel, six have a 20 m pixel and three have a 60 m pixel. The 60 m bands are used mainly to correct for the atmosphere and are rarely used to look at the crop.

Sentinel-2 includes three red-edge bands, centred near 705, 740 and 783 nanometres, all at 20 m. The red edge is the narrow part of the spectrum where plant reflectance rises steeply from red to near-infrared. Few public satellites record it in this much detail.

Landsat 8 and Landsat 9 each carry two instruments: an Operational Land Imager and a Thermal Infrared Sensor. Together they record 11 bands. The multispectral bands have a 30 m pixel, the panchromatic band has a 15 m pixel and the two thermal bands have a 100 m pixel. The panchromatic band records brightness across one wide range of wavelengths rather than in separate colours.

Swath width is the width of ground imaged in one pass. A Sentinel-2 satellite images a strip 290 km wide. A Landsat satellite images a strip 185 km wide.

How often each one passes over

Revisit is the time between two looks at the same place. ESA describes Sentinel-2's revisit as 10 days at the equator with one satellite and 5 days with two satellites. The figure is given for the equator because orbits sit closer together at higher latitudes, where neighbouring passes overlap.

Each Landsat satellite repeats its path every 16 days. Landsat 8 and Landsat 9 are offset by 8 days, so together they can image any location every eight days.

A pass is not the same as a usable image. Clouds, haze and cloud shadow can hide a field on the day the satellite flies over. That is why the number of passes matters: more passes mean more chances of a clear view.

How many pixels cover a quarter section

A quarter section on the Prairies is 160 acres, or about 64.75 hectares. One hectare is 10,000 square metres, so a quarter section covers about 647,500 square metres.

A 10 m pixel covers 100 square metres, so about 6,475 pixels fit on a quarter section. A 20 m pixel covers 400 square metres, giving about 1,619 pixels. A 30 m pixel covers 900 square metres, giving about 719 pixels.

Seen another way, a quarter section is about 805 metres on a side. That is roughly 80 pixels across at 10 m, 40 at 20 m and 27 at 30 m. These counts assume a square field with straight edges aligned with the pixel grid. Real fields have pixels along their edges that are partly field and partly road, slough or neighbouring crop, so the number of clean pixels inside the field is smaller.

What pixel size means for seeing variation in a field

A pixel reports one average value for everything inside it. Anything smaller than a pixel blends with its surroundings. A wet patch, an eroded knoll or a headland that differs from the rest of the field shows up only when it covers several pixels.

At 10 m, a patch about 30 m across covers roughly nine pixels and can stand out as a small cluster. At 30 m, the same patch covers about one pixel and may disappear into its neighbours. Finer pixels give a sharper map of patterns within a field. Coarser pixels still show broad differences between parts of a field and between fields.

• Check which band a map comes from: the Sentinel-2 red-edge bands are 20 m, not 10 m.
• Look for edge effects where the field meets roads, sloughs or shelterbelts.
• Treat a single pixel as an average of its whole area, not a point measurement.
• Compare dates only when both images are clear over the field.

Pixel size is only part of the picture. The band, the date, the cloud cover and the processing all affect what an image shows. An unusual area on a satellite map tells you where to look on the ground, not what you will find there.

Why combining both programmes helps

Because the two programmes fly on different orbits, combining them gives more passes over a field than either one alone. The difficulty is that their bands, pixel sizes and processing differ, so their values cannot simply be mixed.

NASA's Harmonized Landsat and Sentinel-2 project, known as HLS, deals with this. It processes Landsat 8 and Landsat 9 data and Sentinel-2 data into one consistent surface reflectance dataset with 30 m pixels. NASA states that the combined record gives an average of one observation every 1.4 days worldwide. The gain comes from more satellites flying, not from any one satellite passing more often.

The trade-off is detail. To match Landsat, HLS resamples Sentinel-2 to 30 m, so the 10 m detail is set aside in exchange for more frequent looks. Which matters more depends on the question: the pattern inside one field, or how a field changes from week to week.