This week in Cancun, in a jungle-themed conference room with green lighting and an audio track of rain forest sounds, Google launched a new technology platform designed to help scientists — and ultimately developing countries — monitor deforestation. Google Earth Engine combines LandSat satellite imagery from the last 25 years (much of which was not previously available online) with analytical tools provided by scientists, which will allow users to make fine-scale maps.
Greg Asner of the Carnegie Institute at Stanford is one of Google’s partners in the project. His lab provided some of the algorithms built into the Earth Engine that will allow users to analyze the satellite data online.
“There have been two major bottlenecks in helping people to map and keep track of deforestation and degradation: getting access to the satellite data and making it user-friendly,” said Asner.
Scientists used to pay dearly for satellite images. Matt Hansen, a remote sensing scientist at South Dakota State University who created a map of Mexico (shown above) using Google Earth Engine, said that just two years ago, accessing the 53,000 images he used to create the map would have cost $32 million dollars. “But now, we just did it in our spare time this past weekend, for free,” he said.
That’s because two years ago, the USGS started providing its satellite imagery at no cost. But even though they were free, Hansen said, they weren’t readily available until Google, with it’s massive storage capacity, loaded them all online.
Now they are accessible, and by pairing them with tools for mapmaking and analysis provided by labs like Asner’s, Google Earth Engine will allow countries to monitor and measure their forest cover on an ongoing basis, at no cost.
Hansen said that it would have taken a single computer years to create the Mexico map. But by harnessing the power of a thousand computers at Google, it took just one day to generate the map.
Google.org is donating twenty million computing hours to ensure that the technology is available to all scientists and governments. According to the company, ten million CPU-hours is enough to make 1,000 maps like the one Hansen made of Mexico, which reveals the forest cover of the entire country down to a scale of 30 meters.
Here at the UN climate talks in Cancun, it’s clear that there is a growing demand for effective methods to monitor forests. One of the biggest topics on the table here is REDD, which stands for “Reducing Emissions from Deforestation and Forest Degradation in Developing Countries.” Deforestation is a major source of global greenhouse gas emissions (estimates range from 12 to 20%), and many see it as low-hanging fruit in efforts to reduce global warming.
There are more than 30 models of how REDD could work, put forth by different countries, but the general approach is to pay developing countries to stop cutting down their forests. Currently, REDD pilot projects tend to be funded by foundations, corporations, and governments, but ultimately many of the plans call for the payment to come in the form of carbon credits. Because trees store CO2, the idea is that in a cap-and-trade system, corporations could purchase offsets to meet some of their emissions reductions goals.
The state of California is currently exploring potential REDD projects with Brazil and Mexico.
While many see REDD as a positive step, and one that looks relatively promising in the ongoing negotiations, not everyone agrees that it’s a good idea. Some argue that it would encourage land grabs and embezzlement and thus harm indigenous people who rely on the forest. There are also those that disagree with offsets in general, arguing that they could allow developed countries to skirt their obligations to reduce emissions at home. And there are questions about measuring the amount of carbon stored in a forest and verifying whether or not a project is actually preventing deforestation.
Here’s where Google Earth Engine could play a role, said Peter Holmgren, Director of the Environment, Climate Change and Bioenergy Division of the UN’s Food and Agriculture Organization.
“Of course it [Google Earth] will be useful for the countries that choose to use it,” he said. “REDD will depend on effective monitoring systems in all the developing countries that want to participate in REDD. The availability of more standardized data, more accessible data from remote sensing will be crucial.”
Google is working with the Surui tribe in the Amazon on a pilot project to monitor forests on the ground, in addition to monitoring from satellites above.
The original version of this post misstated the number of CPU hours Google is donating to the project. Google is donating twenty million CPU hours over two years, not ten million CPU hours as previously stated.