A team of engineers, scientists, and conservationists, led by University of Minnesota beaver researcher Emily Fairfax, is using machine learning to identify beaver habitats in California, reports the CBC. The goal is to map beaver infrastructure, such as dams and wetlands, with the aim of enhancing conservation strategies that can help mitigate natural disasters like floods and wildfires.
The project – a joint effort that includes Google, the Nature Conservancy and the state of California – uses the Earth Engine Automated Geospatial Elements Recognition (EEAGER), an image recognition machine learning model that detects beaver complexes in aerial and satellite imagery.
The idea originated in 2018 when Google’s Eddie Corwin teamed up with consultant Dan Arckerstein to work on the company’s water stewardship strategy, and realised the conservation potential of beavers. These animals build dams, ponds and wetlands that can store large amounts of water, creating resilient landscapes against droughts, floods, and fires. The largest beaver dam in the world, in Wood Buffalo National Park in northern Canada, is thought to contain about 70,000 cubic metres of water.
"By storing a bunch of water both on the surface in the ponds, but also underground in the soil, they create these big spongy patches in the landscape that plants are able to access water from when you have a period of drought — and that are, honestly, just too wet to burn when you have a period of fire," said Fairfax. Moreover, they help reduce erosion and the impacts of flooding.
Scientists have been using satellite imagery to identify beaver dams for years. At the time Google was investing in artificial intelligence and machine learning, and Corwin thought whether it would be possible to teach a computer to do it as well. The EEAGER model was trained to recognize beaver dams, and in a study published in May 2023, demonstrated its effectiveness. The results have implications for beaver-based river restoration. In addition, this type of model has promise for finding other landscape features.
The project's significance lies in addressing climate change challenges by harnessing the ecological services that beavers provide. While scientists understand the positive impacts of beavers, the population distribution and their numbers remain uncertain. EEAGER aims to provide this crucial information, aiding conservationists in evaluating their efforts and identifying locations for potential beaver reintroduction.
Similar initiatives like "Beavers from Space" in Canada rely on human volunteers to analyse satellite images. While acknowledging the continued importance of on-the-ground work, experts emphasize the role of technology, including AI, in supporting conservation efforts to build resilience to climate threats.
