Short answer: To learn how to verify a satellite image, find its capture date, not the date the map was updated or the post was shared. Open the same place in Google Earth, read the imagery date, then check a second provider such as Esri Wayback and compare the two capture dates. If they disagree, trust the capture date.
A satellite photo shared online usually arrives with a confident caption: “new images showโฆ”, “just released”, “taken yesterday”. The picture itself rarely carries any of that information. A satellite image is a photograph taken from orbit on one specific day, and everything else, the date a map service published it, the date someone posted it, the date you saw it, is a different date. Confusing those dates is how an old photo gets shared as breaking news. This guide shows you how to separate them, step by step, using free tools, and it finishes with a worked example in which two providers gave two different answers for the same place on the same day.
The key term to hold on to is capture date: the day the satellite’s sensor actually recorded the image. Every other date you will meet is measured against it.
What does “new” actually mean for a satellite photo?
Three different dates travel with almost every satellite image, and only one of them tells you when the photo was taken. Mix them up and even a careful reader gets fooled.
| Date | What it really is | Where you find it | What it proves |
|---|---|---|---|
| Capture date | The day the satellite recorded the scene | Google Earth’s imagery date; Esri’s metadata pop-up; a Sentinel-2 scene’s sensing time | When the photo was taken, the only date that settles “is this new?” |
| Update date | The day a map service published that imagery in its map | A service announcement, a blog post, or a new release in Esri Wayback | Only that the map changed, the photo inside it can be months or years old |
| Share date | The day someone posted the image online | The timestamp on the social media post or article | Nothing about the photo, an old image can be shared for the first time today |
That table is the whole check in miniature: a satellite image is only as new as its capture date. An “update” is a publishing event, and a share is a posting event. Neither one re-takes the photograph. Keep the three dates apart and most false “new imagery” claims fall apart in the first minute.
One more distinction matters before you start. Most map services do not show you a single photograph. They show a mosaic: many images stitched together, each captured on a different day. Google Earth Help explains this plainly, for images with a date range, the “Imagery Date” shows the oldest date in the possible range, and a date is not shown at all at the seam between two images or for 3D imagery. So a single date on screen can still hide a patchwork, and a building near a seam may come from a different day than the field next to it. When the exact day matters, check the spot itself, not just the general area.
Why a map update is not proof a photo is new
The clearest recent example of this confusion comes from Bellingcat, the open-source investigation group. In late September 2026, posts and news reports claimed Google had “just updated” its satellite imagery of Gaza, including Rafah, and the images circulating showed widespread ruins. Bellingcat investigated and found the imagery was not entirely new: a Google spokesperson told Bellingcat the latest Gaza imagery had actually been available since June 2026, and the images themselves had been captured by Airbus in February 2026. The previous Google-visible image of the same area dated from June 2024.
Bellingcat then obtained newer Planet Labs SkySat imagery from September 2026, about seven months after the Airbus captures, which showed further demolition around Rafah municipal stadium, Tal as Sultan, Bani Suheila, Jabaliya and Shujaiya. Its conclusion was careful and worth copying as a habit: the viral images were real, they just did not show September 2026. The claim failed on the date, not on the picture. United Nations data cited by Bellingcat put the scale of the damage in context: about 82% of structures in Gaza damaged, with two-thirds destroyed.
Notice what went wrong. Nobody faked a photograph. A genuine image, captured in February, published by Google in June, was shared in September as if it were September. Each step in that chain was honest on its own; only the caption was wrong. That is the most common way an old satellite image misleads people, and it is also the easiest to catch, if you check the capture date before you believe the caption.
The same trap works in quieter settings. A developer’s “new” aerial view of a building site, a “before and after” pair of a flood, a holiday photo of a coastline, in each case the image may be real and the date may be wrong. The fix is identical every time: find the capture date, write it down with the provider’s name, and compare it with the date in the claim.
How to verify a satellite image in Google Earth
Google Earth is the first place most people look, and it is the first place to check, but you have to look in the right mode. In ordinary browsing mode, Google Earth web shows no imagery date in the footer at all, only the attribution line. The date appears when you open historical imagery. That single detail catches out a surprising number of people, who assume the map they are looking at must be current because nothing on screen says otherwise.
Follow these steps in Google Earth on the web, which is the route Google now recommends. Google has announced that the Earth Pro desktop app will no longer be available to download from 25 June 2027, although existing installs will keep working, so learning the web route now saves re-learning it later.
- Open Google Earth on the web and go to the place in the claim. Use the search box, or paste the coordinates from the claim if it gives any.
- Zoom in until the feature you care about, the building, the road, the field, fills a good part of the screen. A date for the whole city is not a date for one street.
- Click the clock icon in the top toolbar to open historical imagery. A timeline appears, with the imagery date shown at the top and in the footer at the bottom left.
- Read the newest date offered. That is the capture date of the newest imagery Google holds for that spot, or, where the view is a mosaic, the oldest date in the capture range, as Google Earth Help explains.
- Step backwards through the timeline. Each date is a different capture. Note the date of the image that matches the claim, and the date of the image just before it, so you can say when a feature first appeared or disappeared.
- Remember that historical-slider dates mean the image was captured on or before the date shown, according to Google Earth Help, and that Google also states plainly that its images are not in real time.
If you still use Earth Pro on a desktop, the same check lives in the status bar at the bottom of the window, and the clock icon opens the same time slider. The dates mean the same thing in both versions. Whichever version you use, record the date exactly as shown, with the provider name “Google Earth”, before you move on, memory is a poor substitute for a written note when you are comparing providers.
A useful cross-check inside Google Earth itself is the reverse image search step. If the satellite image you were sent is a screenshot or a cropped still, run it through a reverse image search first. An earlier appearance of the same frame, dated before the claimed date, ends the check immediately, the image cannot be newer than its first appearance. Our guide to when a photo was taken covers the same dating logic for ordinary photographs, and much of it transfers directly to satellite stills.
How do you cross-check the date with a second provider?
One provider is a claim; two providers are a check. Different services buy imagery from different satellite operators on different schedules, so their newest images for the same place are rarely from the same day. Comparing them tells you how fresh “new” really is, and it protects you from a trap the competitors in this field warn about: two providers can quietly share the same underlying imagery, so a second source only counts if it is genuinely independent.
Esri Wayback is the most useful second provider for most readers, because it keeps an archive of past versions of Esri’s World Imagery map, each with its own release date. But its dates need the same care as Google’s. Esri’s own blog explains that the date on a Wayback layer is the map’s publication date, and that it does not indicate the actual vintage of the individual images inside the map. The real capture date sits in a metadata pop-up that appears when you click a location on the map. A release dated October can therefore contain a photograph taken the previous August, which is exactly what happened in the worked example below.
- Open Esri Wayback and go to the same place you checked in Google Earth.
- Pick the newest release in the list. Note its release date, and label it “release”, not “capture”.
- Click directly on the feature you care about. Read the metadata pop-up: provider, satellite, capture date, resolution. That capture date is the number to compare with Google’s.
- Step back through older releases the same way. Each release’s pop-up tells you when its imagery was actually captured, so you can see whether the map got newer photographs or just a newer publication date.
- Compare the two capture dates. If they are months apart, say so plainly. The gap itself is a finding, not a failure.
Two more free providers fill gaps the first two leave. The Copernicus Browser shows Sentinel-2 imagery, which revisits the same place every five days at the equator, according to the official Copernicus Sentinel-2 page, with its sharpest bands at 10 metres. That resolution cannot show a single house clearly, a house is only a few pixels at 10 metres, but it can settle whether a field was flooded, a forest was burnt, or a large building existed on a given date, because you can pick the exact sensing date and time from the scene’s metadata. NASA Worldview is the fastest of the three for recent events: NASA’s own Worldview booklet says daily global imagery is generally available within three to four hours of satellite observation, with an archive back to the year 2000 and more than 1,200 imagery layers. Its date box sits in the lower-left corner, and it can compare two dates side by side.
| Provider | Date it shows | Best for | Watch out for |
|---|---|---|---|
| Google Earth (web) | Imagery (capture) date in historical imagery mode | High-resolution views and a long archive at one place | No date in normal browsing mode; mosaic dates are ranges |
| Esri Wayback | Release date in the list; capture date in the click pop-up | Comparing dated releases of the same place | Release date is not the capture date, always click for the pop-up |
| Copernicus Browser (Sentinel-2) | Sensing (capture) date and time per scene | Exact-date checks and change over days, not years | 10 m resolution, too coarse for single buildings |
| NASA Worldview | Date you select, with layers hours old | Very recent events such as fires and floods | Daily global layers are coarser than commercial imagery |
Use the table as a routine, not a menu: Google Earth first, Esri Wayback second, then Copernicus or Worldview when the claim needs an exact day or a very recent image. If all the free archives stop months before the claimed date, that is also an answer, say the claim cannot be confirmed from free imagery, rather than stretching an old image to fit it.
Worked example: one place, two providers, two different dates
On 10 November 2026 I ran the full cross-check on one place: Battersea Power Station in London (51.4822, -0.1445), a riverside landmark whose restoration is complete, with building work still going on to its east towards Nine Elms. If the claim you are checking gives you no coordinates at all, first work out where a photo was taken, then bring the coordinates back here. The point of the exercise was not the power station. It was to see what two providers would say about the same spot on the same day, using only their free tools and without signing in anywhere.
In Google Earth on the web, the ordinary view showed no imagery date at all, only the attribution line. Opening historical imagery with the clock icon changed that: the top bar read “Historical Imagery, Jul 20, 2026”, and the footer read “older, 7/20/2026”. I checked it twice, at two nearby points, and both gave the same date. The timeline’s earliest offering for the site was 31 December 1945, a black-and-white aerial view showing dense terraced housing where the power station now stands, a reminder that Google’s archive at a famous London site reaches back eighty years.
Esri Wayback told a different story. Its newest release for the same spot was dated 29 October 2026, and with the local-changes filter off it listed 198 releases back to 20 February 2014. But the release date was not the photograph’s date. Clicking on the power station building itself opened the metadata pop-up, which read: a Vantor (WV03) image captured on 6 August 2025, as shown in the 2026-10-29 version of the World Imagery map, at 0.30 metres resolution. A second click just north of the building gave the same provider and the same capture date.
| Provider | Newest date shown | What that date is | Capture date of the image |
|---|---|---|---|
| Google Earth (web) | 20 July 2026 | Imagery date | 20 July 2026 |
| Esri Wayback | 29 October 2026 | Release (publication) date | 6 August 2025 |
The two providers disagreed by 348 days, about eleven and a half months, on how new the imagery was. By release date, Esri looked newer: 29 October 2026 beats 20 July 2026. By capture date, Google was newer by almost a year. Only the capture dates answer the question “how new is this photo?”, and on that measure Google’s image was the fresher one. Anyone who had quoted Esri’s release date as the photo’s date would have overstated the image’s freshness by more than a year, in the wrong direction.
There was no dramatic visible change to report between the two newest images, and that is worth saying honestly. In both, the power station stands complete and restored, white chimneys, dark roof with two glass courtyards, finished curved residential buildings around it, and landscaped public space to its north. In Google’s July 2026 view, the ground immediately east towards Nine Elms still appeared as extensive bare, cleared ground with industrial sheds and construction material, but I did not have a matching clear Esri view of that exact ground, so I cannot claim a change there between the two providers. A check that finds no change is still a successful check: it tells you the scene was stable between the two captures, which is itself useful when someone claims a sudden transformation.
Run the same routine on any claim and you will get one of three verdicts. The capture dates support the claimed date, the capture dates contradict it, or the free archives have nothing recent enough to decide. All three are publishable answers. Only the fourth option, guessing, is not.
What if the image is not old, but fake?
So far this guide has assumed the photograph is genuine and only its date is in doubt. Sometimes the image itself is the problem. Researchers have shown that AI tools can generate satellite-style images convincing enough to pass as real: as the SecJuice guide documents, researchers at the University of Washington demonstrated in 2021 that AI-generated satellite images of Tacoma could look authentic. A fully synthetic image has no true capture date, so no date check can catch it, which is why provenance matters as much as dates.
Three quick checks help. First, ask where the image came from: a frame downloaded from Google Earth, Esri, Copernicus or Worldview has a provider and a date you can re-check yourself, while a lone screenshot forwarded through three group chats has neither. Second, look for the things a real satellite image cannot hide: shadows that point the wrong way for the claimed time of day, a season that does not match the vegetation, or cars, boats and aircraft that appear in identical positions in “before” and “after” frames taken months apart. Third, remember resolution: at Sentinel-2’s 10 metres, a single damaged house is a handful of pixels, so any image that claims to show fine detail of one building from a free low-resolution source deserves extra suspicion.
If the image is a still from a news report rather than a map screenshot, the same habits apply alongside the checks in our verify a news photo guide, and if the picture has been recycled from an earlier event, our out-of-context photo check walks through tracing its first appearance. A photo can be old, mislabelled, or fake, and the three failures need three different checks, so do not stop at the first one that passes.
How should you record a satellite image as evidence?
Checking a date is only half the job. The other half is writing it down so that someone else can repeat your check and reach the same answer. Fact-checkers keep a short citation habit for exactly this reason, and it costs less than a minute per image.
Record these seven details every time you rely on a satellite image: the provider (Google Earth, Esri, Copernicus, NASA Worldview), the platform or satellite (for example Vantor WV03 or Sentinel-2), the capture date exactly as shown, the latitude and longitude of the spot you checked, the resolution if the tool gives it, any edits you made such as cropping or annotation, and the date you accessed it. That list is your evidence. Without it, your verdict is an opinion; with it, anyone can open the same tools, enter the same coordinates, and see the same dates.
Two habits make the record stronger. First, save the original file or take a screenshot that includes the date line and the provider’s credit, before you crop or annotate anything, a cropped image without its credit is much harder for the next person to verify. Second, when the claim concerns a place where people live, keep the check to the public claim itself: the question is whether the image is new and where it shows, not who lives there or what they are doing now. This guide is about verifying a claimed image, and the checks above are designed for that job.
Finally, state your verdict in one plain sentence and show your working beneath it. “The image was captured on 6 August 2025, so it cannot show an event from October 2026” is a complete answer. “This is new imagery” without a capture date is not an answer at all, it is a caption. The difference between the two is the whole skill this guide teaches, and it is a skill that transfers to every photograph you will ever be asked to believe: find the capture date, check it twice, write it down.
FAQs
How can I tell when a satellite image was taken?
Find its capture date in the tool that serves it. In Google Earth on the web, open historical imagery with the clock icon and read the imagery date. In Esri Wayback, click the spot itself and read the capture date in the metadata pop-up, not the release date. Then confirm the date with a second provider before you trust it.
Is Google Earth imagery live?
No. Google Earth Help states plainly that its images are not in real time. The imagery date is the capture date of the photograph, or the oldest date in the range for a mosaic, so always read the date rather than assuming the map is current.
Why do Google Earth and Esri show different dates for the same place?
They buy imagery from different satellite operators on different schedules, so their newest captures rarely match. Esri also labels its versions by release date, which is a publication date, while the photograph inside may be months older. Compare capture dates with capture dates, never a release date with an imagery date, and the disagreement usually makes sense.
Can I trust a satellite image shared on social media?
Trust it after you have checked it. Run a reverse image search for earlier appearances, find the capture date in Google Earth or Esri Wayback, and cross-check it with a second provider. A genuine image can still carry a false date, as Bellingcat showed in October 2026, so a real picture with a wrong caption is the most common trap.







