How did scientist Andy Jarvis come up with a cutting-edge way to measure how much humans are degrading nature?
Simple: He watched grass turn green.
“When it rains, green grass gets greener, and when it doesn’t rain, it goes brown,” says Jarvis, a geographer at the Center for Tropical Agriculture (CIAT) in Cali, Colombia.
That simple observation from his office window led to Terra-i, an artificial intelligence software program funded by The Nature Conservancy and now used by a growing number of Latin American governments to determine the levels of habitat loss in their countries — information that will be crucial as international carbon markets grow.
Here’s how it works: Terra-i uses real-time rainfall data to predict how green a given habitat should be — and then matches that prediction up against images of the habitat from an Earth-monitoring satellite. Differences in greenness between what is predicted and what is observed — right down to the pixel — suggest habitat conversion by human activity.
The coolest part: Terra-i “learns” as it analyzes, using a neural network to “learn” which actual levels of greenness go with which amounts of precipitation during the year. The result is an unprecedentedly broad look at how nature is doing across a whole continent — not just tropical forests, but every habitat across Latin America.
“It’s the concept of an early warning system,” says Jerry Touval, director of science for The Nature Conservancy in Latin America. “With one tool we can see where we’re losing habitat, analyze the drivers of that loss, and determine where we should be working next.”
What Terra-i Has Learned So Far
Terra-i has already uncovered a number of startling trends:
• Between 2004 and 2009, Latin American countries lost approximately 22 million hectares of forested areas — nearly 85,000 square miles.
• In the Amazon, Terra-i has detected an average annual deforestation rate of 2.6 million hectares per year. Brazil, Bolivia and Colombia showed the greatest losses.
• Terra-i reveals that the habitats most at risk in South America are tropical and subtropical forests (located mainly in the Amazon) as well as tropical and subtropical grassland savannas, which are mainly located in Uruguay, Paraguay and southern Brazil.
Putting What Terra-i Has Learned to Use
Terra-i has started to impact conservation policy in Latin America, too. Shortly after launching, the tool detected strange deforestation patterns occurring in 2008 and 2009 along the Amazonian edge of Colombia’s Caquetá region.
“The Colombian government wasn’t truly aware of the problem,” says Jose Yunis, director of The Nature Conservancy’s work in Colombia. “Essentially, Terra-i was the first to start seeing this.”
Terra-i’s discovery of deforestation in the Caquetá proved that Colombia’s deforestation rate had doubled in five years — twice the official rate that the Colombian government was using. Armed with evidence from Terra-i, Yunis was able to convince the Colombian government to change its official deforestation estimates ahead of the 2009 climate negotiations in Copenhagen.
And, as international carbon markets develop, Terra-i can become a premier go-to source for countries to get a quick handle on their baseline deforestation rates.
“Back when we started this in 2005, the objective was monitoring conservation success across Latin America for The Nature Conservancy,” says Jarvis. “Now because of the climate crisis, we’re in a position of massive demand for Terra-i.”