The power of satellite image mining – finding and monitoring changes on the surface of the earth over time – leaped to center stage recently. News outlets around the globe quickly published “before and after” open source satellite pictures of Syrian military sites hit by cruise missiles on April 13.
Commercial companies have built a sky full of eyes that delivers rapid, frequent imagery. Firms such as Planet Labs and DigitalGlobe have amassed petabytes of economic and strategic value, photographing the entire surface of the earth in deal for months and years. Planet’s fleet of hundreds of satellites collects six terabytes of imagery of the entire earth’s surface every single day. Last week, Planet said it had over 400 images for every piece of territory on the globe.
On the evening of April 13, cruise missiles from the U.S., France, and the United Kingdom hit three alleged Syrian chemical facilities. Roughly twelve hours later, arms control specialists from around the world started publishing the first imagery from Planet on Twitter. The “before” image showed a group of buildings at a complex on the outskirts of Damascus, while the “after” taken on the morning of April 14 revealed a lot of grey rubble.
Not to be outdone, DigitalGlobe quickly supplied high-resolution before-after imagery of two sites outside of Homs, Syria, showing the effects of multiple missiles on the facilities.
Planet has a fleet of over 200 observation satellites, currently the largest constellation in history for any purpose. The majority of them are small, bread box-size “Dove” satellites with a resolution per pixel between 3 to 5 meters – more than sufficient to make out the shape of buildings and structures. DigitalGlobe owns and operates five satellites delivering color images at a resolution up to of 1.2 meters and black and white imagery of 30 centimeters, providing finer detail of building structures and site organization prior to explosive changes.
Over the past year, DigitalGlobal has promoted through social media its ability to rapidly provide imagery of places and things around the world. With Planet photographing the entire globe day after day and other start-ups rushing to build up large constellations of 60 or more satellites capable of delivering color video and 1-meter color imaging, DigitalGlobe has had to step up its game.
As the sky has become full of eyes, the need to rapidly process (mine) satellite imagery continues to grow. Start-ups such as Orbital Insight, Descartes Labs, and SpaceKnow have emerged to provide new tools, using the latest in analytics and machine learning to dig through imagery and deliver insights.
Imagery sources are proliferating. Companies are building radar imaging satellites to see through cloud cover and darkness and “hyperspectral” satellites with cameras capturing information in many different wavelengths of light outside what human beings can see.
Being able to process through petabytes of data across different satellite systems is one challenge for image mining companies. Another will be taking in data from non-imaging sources, such as news feeds and social media, and fusing them together with image data to deliver a deeper understanding of events beyond imagery. It’s a ripe field for developers, Big Data, analytics, machine learning, and AI tools.