Could AI save the Amazon rainforest?
It took just the month of March this year to fell an area of forest in Triunfo do Xingu equivalent to 700 football pitches. At more than 16,000 sq km, this Environmental Protection Area (APA) in the southeastern corner of the Brazilian Amazon, in the state of Pará, is one of the largest conservation areas in the world. And according to a new tool that predicts where deforestation will happen next, it’s also the APA at highest risk of even more destruction.
The tool, PrevisIA, is an artificial intelligence platform created by researchers at environmental nonprofit Imazon. Instead of trying to repair damage done by deforestation after the fact, they wanted to find a way to prevent it from happening at all. PrevisIA pinpointed Triunfo do Xingu as the APA at highest risk of deforestation in 2023, with 271.52 sq km of forest in the conservation area expected to be lost by the end of the year. About 5 sq km had already been destroyed in March.
Home to the endangered white-cheeked spider monkey and other vulnerable and near-threatened species, such as the hyacinth macaw and the jaguar, the conservation area is rich in biodiversity often found nowhere else in the world. But its land runs through two municipalities, Altamira and São Félix do Xingu, with some of the highest rates of deforestation in the country. And despite Triunfo do Xingu being protected under Brazilian law, illegal activities – mining, logging, land-grabbing – have ravaged the area, stripping it bare in places.
Nevertheless, with PrevisIA, there is the potential for change. Imazon is now establishing partnerships with authorities across the region, with the aim of stopping deforestation before it starts. Destruction across the Brazilian Amazon is creeping close to an all-time high. According to SAD, Imazon’s Deforestation Alert System, deforestation this March tripled compared to the same month last year, and the first quarter of 2023 saw 867 sq km of rainforest destroyed – the second largest area felled in the past 16 years.
The idea for PrevisIA emerged in 2016, when the team at Imazon analyzed data collected from SAD satellite images. Tired of getting notifications after large swaths of forest had already been cleared, they asked themselves: is it possible to generate short-term deforestation prediction models? “Existing deforestation prediction models were long-term, looking at what would happen in decades,” says Carlos Souza Jr, senior researcher at Imazon and project coordinator of PrevisIA and SAD. “We needed a new tool that could get ahead of the devastation.” Souza and his team began developing a new model capable of generating annual predictions.
They published their findings in the journal Spatial Statistics in August 2017. The model takes a twopronged approach. First, it focuses on trends present in the region, looking at geostatistics and historical data from Prodes, the annual government monitoring system for deforestation in the Amazon. Understanding what has happened can help make predictions more precise. When already deforested areas are recent, this indicates gangs are operating in the area, so there’s a higher risk that nearby forest will soon be wiped out. Second, it looks at variables that put the brakes on deforestation – land protected by Indigenous and quilombola (descendent of rebel slaves) communities, and areas with bodies of water, or other terrain that doesn’t lend itself to agricultural expansion, for instance – and variables that make deforestation more likely, including higher population density, the presence of settlements and rural properties, and higher density of road infrastructure, both legal and illegal.
“They are the arteries of destruction of the forest,” says Souza, referring to unofficial roads that snake through the Amazon to facilitate illegal industrial activities. “These roads create the conditions for new deforestation.” Monitoring the construction of these roads is crucial to predicting – and eventually preventing – deforestation. According to Imazon, 90% of accumulated deforestation is concentrated within 5.5km of a road. Logging is even closer, with 90% taking place within 3km, and 85% of fires within 5km. Researchers used to comb through thousands of satellite images to see whether they could spot new roads slicing through the biome. With PrevisIA, the work is handed over to an AI algorithm that automates mapping, allowing for quicker analysis and, in turn, more frequent updates. But without a robust computational platform and the ability to update road maps more quickly, PrevisIA couldn’t be put into action. It wasn’t until 2021 that the team at Imazon partnered with Microsoft and Fundo Vale, acquiring the cloud computing power they needed to run the AI algorithm for mapping roads.
LANGLOIS, Jill. Could AI save the Amazon rainforest? The Guardian, Apr. 29, 2023.
Available at: https://www.theguardian. com/technology/2023/apr/29/could-ai-save-amazon-rainforestartificial- intelligence-conservation-deforestation. Retrieved on: July 13, 2024. Adapted.
