Is AI Turning Satellites into All-Seeing Supercomputers?
In the winter of 2016, an analyst at an Earth-monitoring company received an alert from one of its satellites photographing a forest in Primorsky Krai, eastern Russia.
Upon closer inspection, the satellite had noticed that an area that should have been shrouded in forest, was now barren.
Within hours, a call had been made to a global conservation group, who mounted a legal case against the logging companies operating in the area. That process, historically, could have taken months of observing and recording changes. What’s more, in remote areas such as the Ussuri Taiga in Russia’s Far East, policing illegal logging operations have historically had little impact on the extraction of timber.
But thanks to artificial intelligence (AI) and satellites, the ability to observe and respond to changes has become much faster. AI is increasingly being used to help turn the hundreds of man-made objects currently orbiting earth into all-seeing supercomputers.
Geospatial Analysis: A Brief History
As of November 2017, there were 4,635 man-made satellites orbiting the earth. Of the 1,738 in operation, as of August 2017, most (742) are used for communications. The second largest portion is classified by a broad term, “earth observation.” Many of these are weather satellites, but around 300 are taking pictures of the earth and how we use it.
Geospatial analysis, as it’s known, began as a discipline in 1950, when scientists began mounting cameras on to satellites and turning them to look back at the world. It was, and remains to this day, focused on recognizing trends, patterns and anomalies in satellite images of the Earth’s surface.
But, as more and more satellites are launched, the amount of data they generate grows exponentially. It cost $200 million to launch a satellite into orbit 10 years ago. It costs about $200,000 to do the same today. That creates an opportunity for a lot more data gathering. DigitalGlobe, another earth monitoring company, has alone amassed 100 petabytes of satellite imagery since 2000 and grows by over 3,000,000 sq km of new imagery every day.
To cope with this demand, a number of new companies have emerged to teach machines how to recognize what they see on the earth’s surface.
Turning Satellites into Observatories
Computers are given millions of satellite photos, and a set of parameters on what to look for. It could be anything from forest cover, buildings, crops, roads, mines and even cars in parking lots.
Once a machine has learned to identify the objects in a given image, it then focuses on change detection, a term used to categorize shifts in what the machine sees compared with the previous photo of the same area. The result is a computer that has the ability to record and monitor tiny changes in ecosystems in real time, on a global scale.
AI in Action
Astro Digital developed an algorithm that scoured photos from the Sentinel-2 satellite and used it to search for signs of illegal logging. Using change detection, it could see tiny changes in the color of pixels, indicating the appearance of new roads and forest fires — early indicators that an area is about to be cleared.
The company has helped timber companies perform inventory mapping of some 4 billion hectares of the world’s forests. In 2015, it monitored the clearance of some 10,000 square kilometers of Brazil’s Amazon rainforest. This sort of imagery helps anti-logging groups coordinate their resources to focus on areas most at risk of deforestation.
The Human Side of AI Satellite Applications
DigitalGlobe is using their Earth-monitoring photos to help map global poverty. Leveraging its array of satellites, they use four satellite-based metrics that they have determined correlate with poverty:
- Agricultural Productivity
- Car Density
- Building Density
- Building Height
Their goal is to leverage these metrics to identify and quantify poverty at a country-wide scale, giving a clearer picture of where resources and investment are most needed.
AI-powered satellites can even help spot slavery from space.
A project led by researchers in the UK is training machines to pinpoint the location of South Asian brick kilns — frequently the site of forced labor — in satellite images.
The “Brick Belt”, an area that stretches across Pakistan, India and Nepal has historically been difficult to police and monitor. Nearly 70 per cent of the estimated 5 million brick kiln workers in South Asia are thought to be working there under force, often to pay off debts. With a smart pair of eyes (or lenses) watching from above, anti-slavery organizations on the ground have a clearer idea of where to direct their efforts.
Climate and Weather
In the future, things could get even more sophisticated. A team of scientists in Minneapolis were interested to see what would happen if they let an algorithm teach itself how to spot air-pressure patterns like El Niño in the atmosphere.
Weather currents as seen from spaceThat sort of discovery offers up a future in which smaller climactic events could be triggered by air patterns not previously known, which doesn’t sound particularly sexy, but as a machine gets smarter, it could begin to look at the incredibly complex global air-pressure patterns, especially those associated with global warming, and help governments and businesses better prepare for potential disruptions.
A World of Possibilities
While we may not literally be turning each satellite into a supercomputer, the increased edge processing power will enable supercomputer-like speed in the near future. Massive amounts of data are more and more accessible as algorithms get even better at trawling all the images generated. Who knows what we might discover next?
As costs continue to drop, a new fleet of startups are all looking to launch their own satellites into orbit to capitalize on the opportunity. Consider the potential for real estate development. Even today, interactive websites, can already read satellite imagery of two different cities and judge the income brackets of the neighborhoods within them. It can even speculate on the ramifications if a community were to acquire a luxury hotel or tennis court and how much that would impact median income.
Farming is another industry that is set to benefit. While still in the early stages, satellite imagery is now able to tell an individual farmer how much of a planted crop has emerged and monitor crop health through the growing season, especially during periods of high temperatures or drought, creating new opportunities to maximize bounties.
Like so many industries, lowering costs and rising accessibility to data storage and processing will lead to interesting and unexpected outcomes.
This content is produced by WIRED Brand Lab in collaboration with Western Digital Corporation.
Certain blog and other posts on this website may contain forward-looking statements, including statements relating to expectations in the market for our products and applications of our products. These forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially from those expressed in the forward-looking statements, including development challenges or delays, changes in markets, demand, global economic conditions and other risks and uncertainties listed in Western Digital Corporation’s most recent quarterly and annual reports filed with the Securities and Exchange Commission, to which your attention is directed. Readers are cautioned not to place undue reliance on these forward-looking statements and we undertake no obligation to update these forward-looking statements to reflect subsequent events or circumstances.