Can You Tell If A Forest Is Healthy From Space?

You can count trees from space, but counting everything living underneath them is considerably harder. The Amazon stretches across around seven million square kilometres, with much of its wildlife hidden beneath dense forest canopy, so even an enormous amount of fieldwork can only ever show researchers what is happening in particular places at particular times.

Amazon.ia is trying to build a much bigger picture by creating a network of monitoring sites across the rainforest. Rather than relying on one way of observing wildlife, the project plans to combine camera traps, acoustic sensors, environmental DNA and satellite imagery. AI can then help researchers work through the huge amount of information those different sources produce and look for connections between them.

That could mean a camera recording mammals moving through one area while microphones capture the species that can be heard nearby. Environmental DNA can offer another clue about the wildlife that has passed through, while satellite imagery provides a wider view of what is happening to the surrounding landscape. None of those observations tells the whole story, but together they can reveal much more about an ecosystem than any one of them could alone.

There are still limits to what researchers can learn from them. Hearing a particular bird doesn’t necessarily tell you how many are living nearby, just as a camera can only record an animal that happens to pass in front of it. The interesting part is the possibility of collecting those imperfect glimpses across many different locations and following how they change over time.

That’s also where AI becomes particularly useful. Scientists already have plenty of ways to observe wildlife; the harder problem is dealing with the enormous amount of information those methods can generate. If AI can help connect photographs, recordings, DNA samples and satellite observations, researchers may be able to spot changes that would otherwise be buried across thousands of separate pieces of data.