Clouds, revisits and timing: why a new clear image can take days

Revisit and usable image are different things

Satellite imagery is often described by how often a satellite comes back. That number, the revisit, tells you how often a satellite passes over the same place. It does not tell you how often you will get a picture in which your field can actually be seen. A pass becomes a usable image only when cloud, haze and cloud shadow leave enough of the field visible to measure.

The difference matters because a field map built from satellite imagery can look equally confident whether the image behind it is two days old or three weeks old. Knowing why the gap between usable images can stretch, and where to find the image date, makes any satellite-based product easier to read honestly.

How often the satellites come back

The European Space Agency states that the revisit frequency of each single Sentinel-2 satellite is 10 days. Flying two satellites in the same orbit, phased 180 degrees apart, gives the constellation a nominal five-day revisit at the Equator under the same viewing conditions. The two satellites in the nominal constellation are currently Sentinel-2B and Sentinel-2C; Sentinel-2C replaced Sentinel-2A in operation on January 21, 2025. ESA has also kept Sentinel-2A collecting images in an extension campaign, which Copernicus said in October 2026 will continue until December 31, 2027.

ESA gives the five-day figure at the Equator. The number of passes a particular place receives depends on where it is, and this guide does not attempt to count them for any region. Published revisit figures describe how often a satellite is overhead, not how many clear images a given field will receive.

Other missions publish their own figures. The Landsat program describes each Landsat satellite as repeating its coverage every 16 days, with Landsat 8 and Landsat 9 placed half a cycle apart so that together they revisit every 8 days. For Europe's radar mission, the Copernicus programme reported in June 2026 that Sentinel-1 had re-established its nominal six-day repeat cycle with two new satellites, Sentinel-1C and Sentinel-1D.

How cloudy is the planet?

Cloud is the main reason a pass does not become a usable image. NASA Earth Observatory reports that one study, based on nearly a decade of satellite data, estimated that about 67 percent of Earth's surface is typically covered by clouds. The same NASA page notes that cloud is especially dominant over the oceans, where other research shows less than 10 percent of the sky is completely clear of clouds at any one time, and that over land, 30 percent of skies are completely cloud free.

These are global averages. NASA points out that its average map does not illustrate daily or seasonal variations in where clouds are, and cloud over any one region changes with the season and the weather. This guide could not find an authoritative published figure for clear-sky frequency over the Canadian Prairies during the growing season, so it does not give one. The practical lesson is simpler: on any single satellite pass, there is a real chance that some or all of a field is hidden.

Passes versus usable images: a schematic month

The diagram below is a schematic example, not data from any place or satellite. It shows a 30-day window with six passes spaced five days apart. Three of the passes are drawn as cloudy and three as clear, so in this example the field is seen on day 3, day 18 and day 28. Someone opening a field map on day 16 would be looking at the day 3 image, 13 days old, even though the satellite had passed over twice more in between.

The point of the schematic is the shape, not the numbers. A run of cloudy passes can turn a nominal five-day revisit into a much longer wait for a new clear image; in this example the gap between the first two clear images is 15 days. A partly cloudy pass can also leave one part of a field visible while another part is hidden, so two areas of the same field may have different latest dates.

Radar sees through cloud, but sees something different

Optical satellites such as Sentinel-2 and Landsat record sunlight reflected from the surface, so they cannot see through cloud. Radar works differently. Sentinel-1 carries a C-band synthetic aperture radar, or SAR, which sends out its own microwave signal and records what comes back. ESA describes the Sentinel-1 satellites as operating day and night and acquiring imagery regardless of the weather.

That makes radar a natural candidate for filling cloudy gaps, but a radar image is not a cloud-free copy of an optical image. Radar measures how much of its own signal is scattered back to the sensor, and that return is sensitive to surface roughness, moisture content and viewing geometry. A radar image does not record the visible and near-infrared reflectance that optical vegetation indices such as NDVI are calculated from, so values from the two kinds of sensor are not interchangeable, and any colour in a radar picture has been added during processing.

Cloud masks and cloud shadows

Even a mostly clear image can hold cloud over part of a field. Processing systems therefore produce a cloud mask: a layer that marks the pixels judged to be cloud or cloud shadow, so they can be left out before an index or a score is calculated. For Sentinel-2, ESA's Level-2A processing includes the detection of clouds, snow and cloud shadows and produces a scene classification map with classes for clouds, including cirrus, shadows, cloud shadows, vegetation, land without vegetation, water and snow. For Landsat Collection 2, the US Geological Survey provides a pixel quality band, QA_PIXEL, whose cloud, dilated cloud and cloud shadow flags are set by an automated algorithm called CFMask.

Cloud shadow deserves particular attention. A shadow falls on the ground some distance from the cloud that casts it, and a shaded part of a field can look darker or weaker in an optical index even though nothing on the ground has changed. Masks are automatic classifications, not checks on the ground: thin haze, small clouds and their shadows can be missed or misclassified. A careful product tells you how much of the field was usable instead of hiding that information.

Why a field map may show an older image

When the newest pass over a field is cloudy, a careful product has a few honest choices: show the most recent usable image with its date, show the clear part of a partly cloudy image with the hidden part marked, or show nothing new. What it should not do is present an old observation as if it were current. This is also why a map processed yesterday can still rest on an image observed two weeks ago: the processing date and the observation date are different things.

Gaps can also arrive at inconvenient times. A cloudy spell is a spell of unsettled weather, which can be exactly when people most want a fresh look at their fields. A delay before a new clear image is a normal property of optical satellite data, not necessarily a fault in the product that displays it.

Reading any satellite-based field product

Whatever tool you use, a few questions help separate what an image shows from what it cannot show.

• Find the observation date: the day the satellite recorded the image, not the day the map was made.
• Look for a cloud, cloud shadow or quality flag, and for how much of the field was usable.
• Check whether part of the field is masked, and whether a pattern sits next to a masked area or a shadow.
• Ask which sensor the image came from: an optical value and a radar value are not interchangeable.
• Compare with the previous usable image instead of assuming a change happened since the latest pass.
• Use a pattern as a reason to look on the ground, not as a conclusion.