Geolocate images by shadow, architecture and terrain

Block

01 · Geo & Chrono

Topic

Visual Geolocation

Tools

SunCalc · Google Earth Pro · Streetview · Mapillary

Difficulty

Intermediate

Prerequisites

Legal and ethical boundaries of OSINT: GDPR, data minimisation and defensible collection.

Every unverified still is a claim. Visual geolocation is how you test it

The most consequential footage is rarely tagged. This tutorial teaches you to take a single image, work only from what is inside the frame, and produce a coordinate you can defend: shadow direction, vernacular architecture and terrain as your primary evidence.

Every still that arrives without metadata is an unverified claim. The caption says one place. The pixels may say another. Visual geolocation is the discipline of closing that gap: taking a single image and producing a coordinate or coordinate range that can be independently verified.

This matters because place is the first link in the chain that connects an image to an event, an event to a perpetrator, and a perpetrator to a record. The technique works because the world leaves signatures on images: the sun casts shadows whose direction depends on latitude, longitude, date and time, and buildings follow regional patterns of material and form. Each observable is a query you can run against the world.

In the field

In September 2018, BBC Africa Eye published "Anatomy of a Killing," an investigation that identified the location and individuals responsible for the execution of two women and two children in Cameroon. The key video had circulated without any location data. Investigators used three methods to establish that the footage was recorded in the Far North region of Cameroon, near the town of Zelevet.

Shadow direction analysis. The angle and direction of cast shadows in the footage were matched against SunCalc outputs for candidate latitudes and times of day, narrowing the possible location to a specific latitude band in sub-Saharan Africa.
Vernacular architecture matching. Distinctive building features characteristic of the Sahel region were identified in the footage and cross-referenced against satellite imagery of northern Cameroon.
Terrain and settlement pattern identification. The topography and compound layout visible in the footage were matched against Google Earth Pro satellite imagery to fix the precise location, corroborated by military unit insignia on the perpetrators' uniforms.

BBC Africa Eye · Anatomy of a Killing · 24 September 2018

Learning outcomes

By the end of this tutorial you will be able to:

Read the shadows in a still image and use SunCalc to derive a candidate latitude band, time of day or date range
Identify vernacular architecture features that narrow a location to a country or region
Build a geolocation chain with each finding rated to the confidence level framework
Confirm a candidate coordinate using Google Earth Pro historical imagery and Streetview
Produce a hash-verified evidentiary record for every step in the chain

Pinpointing location through visual clues: shadows, architecture and terrain

Every unverified still is a claim. Visual geolocation is how you test it

Learning outcomes

Keep Reading