Satellite imagery can show that a building's roof changed between two acquisitions. It usually cannot, by itself, establish who caused the change, which weapon was used, how many people were inside or whether a military unit stopped functioning. Those are additional claims requiring additional evidence.
The strength of an image depends on its provenance, collection time, location, sensor and viewing conditions. A striking before-and-after graphic is the end of a comparison process, not a substitute for explaining that process. Readers should be able to distinguish what is visible from the interpretation attached to it.
Date the collection, not just the post
An acquisition date records when a sensor collected the image. A publication date records when someone released it. An image posted today can show a site as it looked days, months or years earlier. A caption that provides only the posting date leaves a crucial question unanswered.
Analysts start with the provider or original product, acquisition time and coordinates. They compare stable landmarks, such as road junctions and building footprints, to check that the panels show the same location. Screenshots and social-media crops may remove exactly the metadata needed to assess them.
A valid before-and-after pair gives a time interval for visible change. If a feature appears in the first image and is absent in the second, the images bracket a change between those collections. They do not locate it at a precise minute or necessarily connect it to the attack named in a caption.
Cloud cover, seasonal vegetation, construction and different viewing angles can also alter appearance. A useful comparison chooses a baseline that allows those alternatives to be tested. The broader rules for checking battlefield imagery apply, but overhead images add sensor and collection questions of their own.
Resolution is more than a pixel size
Ground sample distance describes the ground spacing represented by image samples. It is often summarized as a pixel size. That does not guarantee that an object of the same size can be identified, let alone classified reliably as a particular military system.
The British Antarctic Survey's spatial-resolution guide explains why resolution terminology and processing matter. Contrast, shadows and image quality influence what an analyst can distinguish. Sharpening a display can make it easier to view without creating new observations that the sensor never captured.
Off-nadir imagery is collected at an angle rather than straight down. That changes perspective: a tall structure can hide nearby ground, and the same object can look different in another pass. Shadows can help reveal shape when collection time and sun geometry are known, but can also hide features or resemble damage.
Commercial imagery products expose relevant fields such as acquisition time, cloud cover and viewing angle. The provider's streaming-imagery documentation is useful for understanding those fields. A provider's advertised capability should still be checked against the metadata of the particular image being discussed.
Optical and radar images answer different questions
Optical imagery records reflected light and often resembles a photograph. Cloud, haze, darkness and shadow can limit what it shows. A satellite passing overhead is not the same as obtaining a usable optical view of the object of interest.
Synthetic-aperture radar, or SAR, transmits radar energy and measures the return. ESA's Sentinel-1 instrument documentation describes imaging through cloud and rain, by day or night. Its common land mode has a 250-kilometer swath and ground resolution of 5 by 20 meters. Those specifications describe that mode, not every radar product or every satellite.
Radar brightness and geometry require specialist interpretation. They do not correspond directly to natural colors in an optical image. A radar change may warrant investigation without identifying a weapon, building condition or activity on its own.
Revisit frequency also needs care. It describes collection opportunities, not a guarantee that the requested site will be imaged at that interval with suitable conditions and processing. Tasking is a request for collection; it is not proof that a usable image already exists.
Visible damage has an evidence boundary
A documented historical example is UNOSAT's assessment of the 2014 Gaza conflict. The analysis used high-resolution imagery, including 50-centimeter Pleiades imagery, and compared acquisitions to classify damage visible from above. Its methods and limitations are more informative than an isolated damage graphic.
UNOSAT explains that satellite and field assessments complement each other. Overhead imagery can cover places that ground teams cannot reach, but may miss damage such as holes in walls or broken windows. A building that looks intact from above is not necessarily undamaged or usable.
Likewise, visible roof collapse supports a description of that visible condition, within the image's limits. It does not count casualties or prove the cause. Even an apparent disappearance of equipment does not establish whether it was destroyed, moved or obscured.
The strongest public assessments identify the imagery, explain the comparison and separate observation from attribution. They then test possible causes against other records, including authenticated video, witness accounts and official findings. Air-defense reports require the same discipline: a detected object, a claimed outcome and a proven effect are different pieces of evidence. This article explains those methods; it does not claim an independent authentication of a current strike image.