What Can You Actually See in 10 m Satellite Imagery? An OSINT Guide by Target

Free Sentinel-2 imagery has a pixel of 10 m, and that single number decides most of what an open-source analyst can and cannot conclude from it. An object smaller than a pixel disappears into its surroundings. An object several pixels across shows up as a shape. Nothing at this resolution shows the type of a vehicle or a marking on a roof. This guide goes target by target through the questions analysts actually ask.

How many pixels does a target need?

A target needs to cover several pixels in each direction before its shape can be read, and a single pixel only when it contrasts sharply with what surrounds it. The Sentinel-2 instrument provides four bands at 10 m, six at 20 m and three at 60 m, so the finest detail available for free optical imagery is a 10 m square.

The arithmetic is direct: divide the size of the target in metres by 10.

Target size on the ground	Pixels across at 10 m	What the image shows
Under 10 m	Less than 1	Nothing, unless it is far brighter or darker than its surroundings
10 to 30 m	1 to 3	A dot or a blob: presence, not shape
30 to 100 m	3 to 10	A rough shape and orientation
Over 100 m	More than 10	Outline, extent, and internal change

Three levels of conclusion follow from this. Detection means something is there. Recognition means the general class is clear, such as a building or a ship. Identification means the specific type is known. Free 10 m imagery reaches detection for mid-sized objects and recognition for large ones. Identification is the work of sub-metre commercial imagery; the wider trade-offs are in the four types of satellite image resolution.

Can you see vehicles, aircraft or people?

No. Cars, trucks, armoured vehicles and people are smaller than one 10 m pixel and cannot be detected individually in Sentinel-2 imagery. Claims that a free Sentinel image "shows tanks" or "shows a convoy" should be treated as unsupported unless the claim is about something else that is visible, such as fresh tracks or a newly cleared area.

What can be seen is the footprint that many vehicles or aircraft leave:

• Disturbed ground. Repeated traffic strips vegetation and exposes soil, which changes colour over an area far larger than a vehicle.
• New hardstanding and revetments. Cleared, levelled or paved areas tens of metres across are visible.
• Large aircraft on an apron. A large aircraft is a few pixels long. It may appear as a bright blob on dark tarmac, enough to say that something large is parked there and not enough to say what.

The honest wording is "activity consistent with", followed by what was actually observed.

Can you see buildings and damage to them?

Large buildings yes, damage only when it is large. A warehouse, hangar or factory hall covers many pixels and its outline is clear. A house is one or two pixels. Damage becomes visible when it changes the roof or the footprint over an area of several pixels: a collapsed hall, a burned-out building, a debris field.

Partial damage, such as a hole in a roof or a destroyed section of wall, is below the resolution. A building that looks unchanged at 10 m may still be damaged. For this reason, a clear image that shows no change supports only a limited statement: no change larger than a few tens of metres is visible on that date.

Radar helps here. Sentinel-1 is sensitive to structure, and a collapsed building returns the radar signal differently from a standing one. The technique that exploits this is described in SAR coherence for change detection.

Can you see ships?

Yes, large ones, and radar is the better sensor. A cargo ship or tanker is many pixels long and stands out against water in both optical and radar imagery. Sentinel-1 is a C-band radar that images day and night regardless of the weather, and metal hulls return a strong signal against the dark, smooth sea surface.

Free imagery supports these conclusions about ships:

• Presence and count. How many large vessels are at an anchorage or alongside a quay on a given date.
• Approximate length. Enough to separate a large ship from a small one.
• Change over time. Whether a berth that is usually occupied is empty, or the reverse.

It does not support identifying a specific ship, its flag or its cargo. Small boats are often below the detection limit, especially in rough water. More on the method is in ship detection with SAR.

Can you see airfields, ports and other large facilities?

Yes. Runways, taxiways, quays, breakwaters, tank farms, rail yards and large industrial plants are far larger than a pixel, and their layout is clear in 10 m imagery. These are the targets where free imagery is strongest, because the questions are about large, slow changes.

Changes that are routinely visible at a large facility:

• A runway or apron being extended, resurfaced or cratered over a wide area
• New buildings, tanks or shelters of warehouse size
• New earthworks, berms and cleared plots
• A quay extended or a basin dredged
• Vegetation cleared along a perimeter

Narrow linear features such as roads and fences can be visible even when they are narrower than a pixel, because a long, straight line of contrasting pixels is easy for the eye to pick out. The feature is detected; its width is not measured.

Can you see fires, burn scars and floods?

Yes, and these are often the clearest signals of all, because they change the colour or the radar response of a large area. A burn scar or a flooded plain is hundreds of pixels, far above the detection limit.

• Active fire. Detected by thermal sensors rather than seen as a shape. NASA notes that each detection marks the center of a pixel flagged as containing one or more fires or other thermal anomalies, at 375 m for VIIRS and about 1 km for MODIS.
• Burn scar. Visible in Sentinel-2 as a dark area, and much clearer in infrared band combinations than in natural colour. See false color composites.
• Flood. Open water is dark and smooth in radar imagery, so Sentinel-1 shows the flooded extent through cloud. Water depth is not visible.

What does this mean for an OSINT conclusion?

Match the wording of the conclusion to the level the imagery reached. Most over-claims in open-source reporting come from stating identification when the imagery supports only detection.

The imagery shows	Supported wording	Not supported
A dark area that was not there before	"A burn scar of about this extent appeared between these dates"	"The depot was destroyed by a strike"
Bright returns on water near a port	"At least this many large vessels were present"	"These are warships"
A changed outline of a large building	"The roof of the hall changed between these dates"	"The building was hit by a missile"
Nothing	"No change larger than a few tens of metres is visible"	"Nothing happened"

When the question needs identification, the free image still does useful work: it shows whether anything changed, where exactly, and between which dates, which is what a commercial image order needs to be specified. The steps for a full check are in how to verify a reported event with free satellite imagery, and the tools are compared in best OSINT satellite imagery tools. Definitions of the terms used here are in the glossary.