Silicon Valley data scientists are teaming up with rainforest conservationists to predict where and when the planet's jungles will be cut down ahead of time using new kinds of algorithms and satellite imagery.
This marriage of big data and environmental activism is designed to slow the rapid pace of illegal logging that is both endangering wildlife and making climate change worse.
PHOTOS: Global Warming Right Before Your Eyes
"If those who have the motivation and the will to stop deforestation don't have the information and know who is responsible, there's not much they can do," said James Crawford, CEO and founder of Mountain View-based Orbital Insight. "Our goal is to take the data and turn it into meaningful information to change practices on the ground."
Crawford's two-year old space-tech startup is partnering with the World Resources Institute (WRI), a global environmental group that has already developed a cool, interactive map showing where deforestation is underway as part of its Global Forest Watch program.
Using a technique called deep learning, Orbital Insight says it will analyze satellite images of tens of millions of acres of forest provided by WRI to discover patterns that identify indicators of deforestation risk. The images come from public sources, such as the University of Maryland's Department of Geographical Sciences, NASA's Landsat satellite, as well as some higher-resolution images from partner tech firms.
ANALYSIS: Amazon Rainforest Seen Burning From Space
Some of these images can resolve features only 50 centimeters wide -- the size of a laptop. Satellites snap images of reflective colors on the forest canopy, so that the color reflecting off roads looks different to a space satellite than the color reflecting off leaves.
Deep learning is a form of artificial intelligence that involves processing tremendous amounts of data to solve problems in a way that roughly mimics the human brain, the firm says.
Global Forest Watch researchers will be able to use the deep learning algorithms to recognize patterns, such as road-building in previously undisturbed areas, and forecast deforestation before the destruction occurs. The technology will also increase transparency around the impact of global commodity supply chains on forest loss. So-called "forest-risk commodities" such as palm oil, beef, soy, pulp, and paper has led to over 70 percent of deforestation in tropical forests, often illegally.
10 Signs Climate Change Is Already Happening
"Rates of deforestation are continuing to accelerate across the tropics," said WRI's Crystal Davis. "And emissions from deforestation are contributing 15 to 20 percent to overall greenhouse gas emissions. There's still a lot of work to be done."
The predictive algorithms will allow WRI on the ground to notify government officials and local conservationists before the chain saws and bulldozers move in to specific areas.
"Right now we have the ability to see on a monthly basis where trees are disappearing," Crawford said. "We are seeing that with medium resolution imagery, with Landsat. Moving forward, our access to higher resolution data is coming. As the pool of data grows, the challenge is how to extract useful information and deliver it."
WRI's Davis says that there are some bright spots in places like Brazil, which has taken advantage of satellite imagery to more quickly respond to reports of illegal timber harvesting. Part of the problem, however, is that local villagers are often to blame for the harvesting, and need to have some kind of substitute for the quick money available by cutting and clearing.