7.1 Where and when
Most false claims about images are not fakes. They are real pictures with the wrong place or the wrong date attached: last year's flood shared as today's, or a street in one town presented as another. Geolocation and chronolocation test those two claims.
- Geolocation works out where an image or video was captured, by matching what it shows against reference material such as maps, satellite imagery and street-level photos.
- Chronolocation works out when it was captured, from shadows, weather, the state of the scene and the first time it appeared online.
The result is never a bare pin or a bare timestamp. It is a place with an honest radius, a time window, and a statement of how confident you are and why.
7.2 Reading an image for location clues
Before you search, write down what the image shows. An inventory stops you from fixing on the first clue and forgetting the rest. Sort each clue by how long it is likely to stay the same:
| Clue type | Examples | Value |
|---|---|---|
| Fixed and distinctive | coastline, hills, a bridge, a tower, a road junction, a railway | Strong: it shows on maps and satellite imagery |
| Fixed but common | building styles, street lamps, kerbs, road markings | Useful to narrow a region, weak on its own |
| Text | shop signs, street names, notices, vehicle livery | Strong if legible, but check the language and script against the claim |
| Changeable | parked cars, market stalls, posters, scaffolding | Weak for place, often useful for time |
| Natural | vegetation, soil colour, snow, the angle of the sun | Helps with region and season |
Location metadata rarely helps. Most platforms strip it, and SITREP removes all metadata, including GPS location, from images before anyone else sees them. Where a file you hold lawfully still carries a location, treat it as a claim to test: metadata can be edited, and phones record wrong positions indoors and in cities.
7.3 Narrowing the search area
Start with the claim. If a post says the video shows Kestrel Harbour, test Kestrel Harbour first: it is quicker to confirm or rule out a stated place than to search the world. If the claim fails, widen the search step by step, using the strongest clues first.
- Language and script on signs narrow the country or region.
- Road clues such as the side of the road traffic uses, the colour of lane markings and the shape of signs narrow it further.
- Terrain and vegetation tell you whether you are looking at a coast, a plain or hills, and at what season.
- Distinctive structures such as a three-arched bridge or a harbour wall let you search maps and imagery directly.
Beware of look-alikes. Many towns share the same chain shops, road signs and building styles. A match on common features is a lead, not an answer.
7.4 Matching against reference imagery
Reference imagery includes maps, satellite and aerial imagery, street-level photos and earlier media that has already been verified. A geolocation is only as good as its matches. Aim for:
- At least three independent fixed features that appear in both the image and the reference.
- The same relative positions. The features must sit in the same order, spacing and alignment from the camera's point of view, not just exist nearby.
- A line-of-sight check. Ask whether the matched features could really be seen together, in that order, from somewhere in the scene. This is a private check of your match, and it stays in your log. What you state, propose or publish is where the event or scene is, never where the camera stood. If the check points to a home, or to someone who may still be there, note only that the line of sight works and go no further.
- Dated references. Satellite and street-level imagery can be months or years old. A building that is missing from the reference may be newer than it, not proof that the place is wrong.
Record every match side by side, ideally as an annotated comparison image with the same features marked in both. That comparison is your evidence, and on SITREP it is what reviewers will check.
7.5 An honest radius
Precision and accuracy are different. A point placed to the metre is precise; it is accurate only if the event really happened there. False precision misleads readers and, in a conflict, can put people in danger.
SITREP proposals use a fixed set of radii so that nobody claims more than they know:
| Radius | Choose it when |
|---|---|
| 25 m | The event or scene is fixed to one spot, such as a single building or a junction, with several matching features |
| 100 m | You have matched the street or block, but the exact spot along it is uncertain |
| 500 m | You have matched the neighbourhood, a landmark in view, or a stretch of waterfront |
| 1 km | You have the district or the approach to a town |
| 5 km | You have the town or a stretch of road between two towns |
| 25 km | You have only the area or region |
The radius always describes where the event or scene is, never where the camera stood. When in doubt, choose the wider radius. A wider radius that turns out right is worth more than a narrow one that turns out wrong.
7.6 Chronolocation: bracketing the time
You will rarely prove the exact minute an image was taken. Instead, bracket it: find the earliest time it could have been captured ("not before") and the latest ("not after"), and narrow the gap.
| Evidence | What it gives you |
|---|---|
| The first appearance online | A "not after" limit. Find the earliest copy (module 6); the image cannot be newer than its first upload |
| Shadows | Time of day and season. Shadow direction gives the sun's direction; shadow length against the object's height gives the sun's height. A sun-position calculator turns both into time windows for the place |
| Weather | Rain, snow, cloud and wind can be checked against historical weather records for the place |
| Season | Leaves, crops, snow and daylight narrow the time of year |
| State of the scene | Construction, damage and repairs compared with dated satellite or street-level imagery give "not before" and "not after" limits |
| Clocks, screens and notices | Direct evidence, if legible, but clocks can be wrong and notices can be old |
| Context | Scheduled events, such as a market day or a published road closure |
Always convert to UTC in your log (module 1), and watch for daylight saving changes. An upload time shown in your own time zone is a common source of error.
7.7 SITREP community geolocations: proposing
On SITREP, members can propose where something is, and other members confirm or dispute it. Community geolocations are the community's conclusion, not SITREP reporting, and they are labelled "Community geolocations. Not SITREP reporting."
What you can geolocate: a post in an alert thread, a public channel or the community feed, an image in such a post, or an alert. The alert itself is never changed. Posts in private channels and incident reports cannot be geolocated this way (incident reports have their own rules, section 7.9).
Who can propose: members from the Member trust level upward, within a daily limit that grows with trust. New accounts cannot propose. You can have one open or held proposal of your own per target.
What a proposal contains:
- a point, set on the map, and a radius from the fixed set in section 7.5
- a category: event site, landmark, infrastructure, military position or other
- your reasoning, in plain text of up to 1,000 characters: what matched, and how
- evidence: up to three links to the original sources and up to four images, such as your annotated comparison
Your reasoning is checked by the same safety rules as a post. Set the location with the map point only. Do not type coordinates, grid references or map links into your reasoning: text like that is held for review. Link to original sources, never to mirror sites or link shorteners.
Never allowed: locating a private individual or a private home. SITREP refuses proposals whose reasoning names one or contains contact details, such as phone numbers, email addresses or home addresses, and staff remove any others. Propose the event or the scene, never the camera position: where the person filming stood is not what you are locating.
You can withdraw your own proposal while it is open and nobody has voted on it yet.
7.8 Confirming and disputing
Who can vote: members from the Established trust level upward, within a daily limit, who are not recused.
Recusal. To stop friends confirming each other's work, these people cannot count towards a proposal: the proposer, the author of the post being geolocated, anyone in a mutual follow with the proposer, anyone in a block relationship with the proposer either way, and anyone who shares a private channel or group conversation with the proposer. The app does not tell you when your vote is set aside for this reason, so the rule cannot be used to work out who is connected to whom.
How to vote. You confirm or dispute, with optional reasoning of up to 500 characters and, if you like, your own point. A confirm whose point falls outside the proposal's radius is recorded as a dispute, because it disagrees with the claimed location. You see how others voted only after you vote, and then only as a leaning in words, such as "Leaning confirmed". No counts are ever shown.
Weight. Votes are weighted. A vote counts a little more when the voter holds a verification mark, including the gray check that this course leads to, and when the voter's past votes have agreed with the final outcomes. The exact numbers are staff settings and can change.
Outcomes. A proposal becomes:
| State | What it means |
|---|---|
| Confirmed | Enough independent, weighted confirmations, a clear majority, at least one from a highly trusted or verified account, and a set of confirmers that passes checks against coordinated voting |
| Rejected | Enough weighted disputes, with a clear majority against |
| Contested | Strong support on both sides. It stays under its target as "Contested", never goes on the map, and staff may decide |
| Unresolved | No result after 7 days |
In the first 14 days after geolocations launch, the thresholds are stricter, staff audit one in five confirmations, and SITREP staff may take part in confirming. Staff confirmations are never shown as a person. Staff can reopen a confirmed geolocation, but it is never silently changed.
How it is shown. Under the target, a neutral chip reads "Community geolocation: Confirmed" (or "Proposed" or "Contested") with the radius in words, such as "within 100 m". Only confirmers who show a verification mark are listed; nobody sees a full list or a count. The map layer "Community geolocations" is off by default and shows only confirmed, released geolocations that are not military positions. Each appears as a hollow diamond with a dashed ring, and its centre is snapped to a grid a quarter of the radius wide, so the map never shows more precision than the community agreed.
Your geo score rewards confirmed proposals. Tighter radii earn more when confirmed, and every rejected or removed proposal costs points, so claim only the precision your evidence supports.
7.9 No live positions: holds, delays and pins
SITREP builds the no-live-positions rule into the product. It relies on structure, not only on people spotting problems.
Conflict contexts. Content is in a conflict context when its alert concerns a country on SITREP's staff-maintained conflict list, when it sits in the official channel of such a country or a channel tagged with one, or when its alert is high impact and less than six hours old. A post in the community feed takes its context from its country tags, and a feed post with no country tag counts as a conflict context.
Held proposals. A geolocation proposal is held when any of these apply. A held proposal is visible only to staff and to you, the proposer, and never appears on the map:
- its target is in a conflict context, whatever its age or category
- its category is military position
- the post it locates tripped a location safety rule, such as one for coordinates or troop movements
- the image it locates has not yet cleared image review
- its point lies inside a listed conflict country with a radius of 5 km or less
- its reasoning trips a safety rule
The proposer sees that the location is held for review. Staff release a held proposal only when there is no live risk. A military position proposal is never released until at least 72 hours after the event, and even when confirmed it appears only under its target, never on the map. The category exists so that historical analysis can be reviewed safely; it is not a way to report where forces are.
Delayed posts. Posts, replies and channel posts in a conflict context, and posts elsewhere that look like they contain a position, are published after a short delay unless a report or rule sends them to review. Verification marks do not exempt anyone.
Incident reports. Incident reports follow their own rules. They unlock at the Established trust level. A pin more precise than 200 m is held for staff review before anyone sees it. "Approximate my pin" counts as 500 m, and an exact pin with no radius counts as fully precise, so it is held. Reports that name or target a person, or point to someone's home, are refused. There is no military category. Reports in listed conflict areas are checked by staff before anyone sees them, and strike reports there are shown only well after the event and only on a coarse map. Pin the incident, never yourself, and never anyone's home.
7.10 Reviewing someone else's proposal
Vote on the evidence, never on who proposed it. Before you confirm, check:
- Do at least three fixed features match, in the same relative positions?
- Does the line of sight work? Check it, but never put the camera position in your vote or your reasoning.
- Is the radius honest for the evidence shown?
- Does the time of the target fit what the image shows?
- Is the proposal about an event or a place, never a private person or a home?
"I can confirm" means you checked the evidence yourself, not that it sounds right. If the place is right but the radius is too tight, dispute it and say why in your reasoning.
7.11 Your own safety and your log
Geolocation work can reveal more about you than about the image. Before you share a comparison:
- crop out your own map app's location marker, search history, open tabs and account names
- use research accounts and settings kept apart from your personal ones (module 3)
- remember that uploading an image to an online tool can share it with that service; do not upload material you are not free to share
Log each geolocation as you would any research (module 1): the image and where you found it, its first appearance, every clue, each reference you checked with its date, your matches, your radius, your time window, your confidence, and dead ends. Another analyst should be able to repeat your work from your log alone.