Open Source geographical intelligence is no longer impossible in times of big data and commercial satellite imaging. Here are some examples of datasets and methods:


Datasets

Shapefiles, Text etc.:

  • OpenStreetMap (OSM), Google Maps, Apple Maps etc. -> typical map layers
  • Products of other OSINT users (e.g. liveuamap.com, deepstatemap.live, etc.)

DEM:

  • ALOS/PALSAR 30m global DEM, SRTM etc.

Imagery:

  • SAR -> Sentinel-1
  • Optical multispectral -> Sentinel-2, Landsat, Sentinel-3 etc.
  • Previews of commercial satellite images (https://www.euspaceimaging.com/image-library/)
  • Not georeferenced high resolution images (e.g. Maxar images on twitter)
  • Google Maps/ Google Earth/ ArcGIS/ Bing

 

 

Methods

Aim is the mapping of

  • Possible Targets
  • Enemy unit/assets locations
  • Infrastructure and environment
  • Battle activities

Methods:

  • Ready to use datasets (road networks, elevation model, OSINT-datasets, …)
  • Classic Analysis of imagery data (Sentinel-2 fire detection, geolocalizing enemy units)
  • Advanced Analysis of imagery data
  • Sentinel-2 fast moving target detection in multispectral image
  • C-band radar detection with Sentinel-1
  • Change detection

What to consider:

  • Natural borders – define strategic defense lines
  • Topographic differences – Topography as a key part in defense, supply and agility
  • Bottle neck points (bridges, main streets, …) – Key strategic points
  • Observed attacks – Determine enemies positions and strategy
  • Static military targets (bases, airbases, RADAR, etc.)
  • Infrastructure (Roads, Railroads, etc.)

 

Examples

Ready to use datasets

OSM -> Infrastructure (Roads, Railroads, etc.), Bottle neck points (bridges, main streets, …)

 
ALOS 30m DEM -> Topographic differences

Classic Analysis of imagery data

Observed attacks with Sentinel-2 data

Georeferencing of high resolution images from twitter or other channels

Static military targets (bases, airbases, RADAR, etc.)

Multi-spectral misregistration

There are also “higher-tier”-methods, which can be applied to open source datasets. Multi-spectral misintegration can be applied to multispectral satellite images, such as Sentinel-2 images. The different spectral bands are not exactly acquired at the same time, which causes very fast moving objects – such as aircrafts, ships or even trucks- to be aqcuired multiple times, hence for in each spectral band once. Detecting such misintegrations means detecting fast objects and when measuring spatial the distance between the misintegration we can calculate the direction and velocity of the object.

C-band radar detection with Sentinel-1

Have a look at the ELINT section to learn more about this method. Sentinel-1 measures the same frequency bands as some Radar stations are emitting – and can therefore be detected using these satellite images.

Back to Reports