Skip to story A deforested area near Novo Progresso in Brazil's northern state of Para. Andre Penner/AP
Forests are often described as the lungs of our planet. They’re home to countless animal and plant species, provide humans with oxygen and food, and help stave off the effects of global warming. Yet we as a species are constantly imperiling these vital ecosystems.
An estimated 46 to 58 thousand square miles of forest are cut each year. At the current rate of deforestation, the world’s rain forests could be obliterated within just 100 years, according to National Geographic.
Now environmentalists are turning to artificial intelligence and satellites to combat forest destruction in real time—that is, in time to stop it.
Last year the World Resource Institute launched a website called the Global Forest Watch to provide up-to-date information about the state of forests worldwide. The trouble is that the site has often been stuck reporting deforestation after the fact. Now one startup is working to give the organization the tech to spot deforestation risks before they cause irreversible harm.
Small Changes,Serious Impact
Orbital Insight, founded by former Google and NASA robotics and artificial intelligence expert James Crawford, plans to collect satellite imagery of tropical forests to track changes over time. The company will set its machine vision algorithms loose on the images to detect small alterations in the landscape that humans might miss but that could indicate illegal logging.
At first, the visual impact of building a new logging road in a huge forest is so small that the human eye will have trouble noticing any difference, Crawford explains. But as loggers use that road to clear more and more trees, the effects become more and more obvious. If that road can be spotted before much logging has actually been done, activists and governments stand a much better chance of preventing more damage.
Crawford says Orbital Insight will draw on a wide range of satellite image sources and analyze the data using an artificial intelligence technique known as deep learning — the same approach that companies like Google and Facebook are using for image recognition and other tasks. He says the company may also pull in other data, such as temperature and rainfall, as part of the analysis.
The project is still in its earliest stages, but Orbital Insight has done this sort of work for private customers before. For example, it has helped hedge funds determine which big box stores have the most customers by comparing the number of cars that routinely appear in each company’s parking lots. It’s estimated the world’s oil reserves by analyzing changes in the shadows cast by the floating lids of oil tanks around the world. Crawford says the Orbital Insight team wanted to find ways to apply its technology to other problems that might benefit humanity, and they were all impressed with the work that WRI has been doing.
Ultimately, Orbital Insight and environmentalists don’t have the ability to stop logging. That responsibility will fall on governments and citizens. But the job might get a little easier if the people working to save forests can anticipate the damage before it’s done, rather than being left to do damage control after it’s too late.