Nature’s scale can be very difficult to comprehend. The largest rainforest on the planet, for example, the Amazon spans for almost 2.6 million square miles, crossing the borders of nine countries. That’s more than twice the size of India and equivalent to 68 percent of the US total area.
Trying to protect something so big is a challenge of similar size and complexity, but that was made easier by Google’s Earth Engine, a geospatial processing service directed to researchers and other public policy experts. Launched in 2010, it completely changed the way scientists can track forest loss all around the world by making the process much faster and wider — and it all started with one person’s curiosity.
In 2005, Google engineer Rebecca Moore learned about a new logging project near her home in the Santa Cruz Mountains of Northern California. Although the project’s plan was shared with her community by the developers, it was hardly informative or transparent.
“It was a 400-page document with a sketchy black-and-white map that no one could understand. No one could even tell where it was going to be, much less the details of what was being proposed,” she says.
A self-described map geek, she decided to use Google Earth, the global visualization tool that was released a few years before, to look into the targeted 1,000-acre area. The high-resolution satellite images showed what the 400 pages didn’t: the project would impact daily life of not only the 2,000 people in the community but also residents of the broader Silicon Valley community who got their drinking water from that region.
The area was made up of coastal redwood, an endangered sequoia species that includes some of the tallest living trees and the longest-living organisms on Earth.
“It turned out, the helicopters were going to be landing and taking off and hauling the logs over the nursery school and the daycare center,” she recalls. “It was going to go right up to within 100 meters of three schools.”
With a flyover video of less than two minutes, she was able to mobilize the community and ultimately stop the logging project. “Seeing the scale of destruction, seeing where deforestation is happening, gives you insights into what’s causing it, which areas are more threatened and need more protection,” says Moore, who is now the director of Google Earth, Earth Engine and Outreach.
Powered by the cloud
Although that was a relatively small area and tackled only a specific moment, being able to help stop the logging project using Google Earth sparked in Moore the idea that some similar monitoring could be done to the world’s forests. But for it to work, researchers would need to have access to an enormous database and to tools for processing all this data — that was the seed of Google Earth Engine.
Until then, maps and satellite images would be stored in different institutions, such as NASA and other government agencies, and scientists had to gather this information and download it in their computers, and only then could they start the analysis. It was timely, taking weeks to collect it all, and costly, due to the hardware power needed.
“Google Earth Engine changes this paradigm,” explains Gilberto Câmara, a computer scientist and former director of Brazil’s National Institute for Space Research (INPE). “Instead of the data going to where the software is, which is your computer, the software goes to where the data is, which is Google.”
The tech giant gathered the free-access satellite imagery from research institutions around the globe, from the US to Europe and Brazil, uploaded it to its cloud service, and made it available to academics, along with programming functions to process the data. “This obviously benefits greatly from Google’s ability to be a big data warehouse,” Câmara says.
“We’re using the Google data centers to bring in, every day, thousands and thousands of images from hundreds of satellites and other data sources and store them in a form that’s ready for analysis,” says Moore.
The first target of all of that machine power was the world’s largest tropical forest. At the Amazon Institute of People and the Environment (Imazon), a nonprofit organization, was built the first operational forest monitoring system on Earth Engine. It was integrated into its Deforestation Alert System to report the pace of forest degradation and deforestation in the Brazilian Amazon on a monthly basis.
That work was followed by a breakthrough analysis. Published in Science in 2013, it was the first high-resolution map of the world’s forests and how they had changed over the prior 12 years.
“It had never been done before, and the reason was: it required more than a million hours of computation. But because we ran it on 10,000 computers in parallel, we had the result in a few days. On a single computer, it would have taken 15 years,” Moore explains.
Deforestation and the fight against climate change
Energy is a huge deal when talking about climate change since the world is still heavily reliant on fossil fuels. However, deforestation is a big driver of the global temperature increase, too — not only because it reduces the amount of carbon storage available but also due to the carbon emitted by forest destruction. That is particularly important in Brazil’s case.
According to the most recent official data, the Brazilian Amazon lost over 4,466 square miles of forest coverage from August 2021 to July 2022 (deforestation numbers are measured by taking into account the rainy season in the region, which ends in July). To get a better understanding of the scale of the destruction, that is the equivalent of almost 15 times New York City’s area.
The destruction of the Amazon rainforest affects not only the region and the communities who live there but the whole world. It helps regulate the global climate by storing carbon and releasing billions of tonnes of water into the atmosphere. Protecting its biodiversity also helps to prevent new disease outbreaks and possible sources for medicines.
To try to revert the high rates of deforestation, the Brazilian government just released a new plan that ranges from fighting environmental crime to promoting bioeconomy activities that generate income for local communities while keeping the forest standing.
The largest country in Latin America, Brazil is the seventh biggest annual emitter of greenhouse gasses, accounting for 3 percent of the world total, according to 2019 data compiled by the World Resources Institute. The top emitters are China (25.2 percent), the United States (12 percent), and India (7 percent).
While around 71 percent of the world’s electric power comes from nonrenewable sources, that number drops to 17 percent in Brazil. Most of the country’s carbon emissions come from deforestation and changes in land use, accounting for about 49 percent of their emissions in 2021.
Most of Brazil’s carbon emissions come from deforestation and changes in land use, accounting for about 49 percent of their emissions in 2021
This figure is generated every year by the System of Estimates of Emissions and Removal of Greenhouse Gases, a non-governmental initiative. The system has a sister platform called MapBiomas (“MapBiomes,” in a literal translation), which tracks deforestation, fires, mining, surface water, infrastructure, and quality of pastures in the whole country.
Created by a network of scientists and research institutions, MapBiomas was first launched in 2015 at the UN climate conference in Paris.
“We needed to produce a map every year to see if the public policies and recommendations we made affected emissions and removals of greenhouse gasses,” says Marcos Rosa, the technical coordinator of MapBiomas.
That was hard because it demanded a huge dataset (daily, high-resolution pictures of the entire country), a lot of processing power, and machine learning (to identify and classify changes on the satellite images of the Amazon and Brazil’s five other biomes).
“Google Earth Engine was what made it possible for [researchers in] each biome to work remotely without having to purchase a big computational infrastructure.”
That’s where the Google Earth Engine came into play. “It was essential. Google Earth Engine was what made it possible for [researchers in] each biome to work remotely without having to purchase a big computational infrastructure,” says Rosa.
This kind of monitoring was taken to a worldwide scale by the World Resources Institute’s Global Forest Watch (GFW), launched in 2014. The platform’s director, Mikaela Weisse, explains that the institute has been involved in providing transparent forest information since the 1990s but, until then, didn’t have the technology to make it global.
“A number of advancements have made that possible — including free access to satellite imagery and better internet access around the world — but a key piece of that is the cloud computing and power to analyze satellite images using Google Earth Engine,” she says.
Some of their partners are researchers at the University of Maryland and Wageningen University, who use Google’s system to create near-real-time deforestation alerts, detecting changes in the forest as soon as possible in order to inspire a response.
Nevertheless, Weisse sees some gaps in the data available. “The tropics are particularly difficult to monitor using satellite imagery because they are so often cloudy. There is also less historical and ground information available in those locations,” she says, adding that there are efforts from both public and private organizations to address this issue.
Public data managed by a private entity
Although the scientists who spoke to The Verge didn’t express particular concerns about it, Google is a huge company that, naturally, has its own economic interests — that will not always align with environmental issues. In Earth Engine’s case, problems hypothetically could arise regarding the reliability of the data being publicized or the interest of the company to keep the platform running, for example.
Google’s Rebecca Moore emphasizes that all the images they’re providing are in the public domain. “This is data coming from government satellites, for example, in the US, Brazil, Europe, Japan,” she says. “And all of the data is scientifically peer-reviewed.”
“I don’t see conflicts of interest that would have any impact on the data produced by independent researchers with Earth Engine,” says Weisse. “The biggest risk is if Google decided it wasn’t in their economic interest to continue offering this service — that would be a huge loss for the research community.”
“The biggest risk is if Google decided it wasn’t in their economic interest to continue offering this service — that would be a huge loss for the research community.”
Even though Google’s service was the first of its kind, today, there are alternatives, such as Microsoft Planetary Computing and Amazon’s Earth on AWS. “Several countries have also launched efforts to have their own Google-like platforms,” says INPE’s Gilberto Câmara.
He claims that Google democratized the access to information but thinks Earth Engine didn’t evolve much since 2010. “It has some processing capabilities, but it’s not a universal system that gives access to state-of-the-art machine learning methods.”
MapBiomas’ Marcos Rosa says that, even though they have a partnership with Google for digital storage space, the company has never tried to meddle with the content being produced by them.
“We have always kept an alternative for the data that we produce with Google Earth Engine. So, if Google has a change in policy — which has never happened — we can replicate [the model] on another platform, such as Amazon’s.”