North American beavers transform ecosystems with their engineering prowess. By storing water, digging channels, and gathering nearby vegetation, you dramatically change the landscape of environments ranging from tundras to deserts.
Two centuries of the fur trade, beginning in the 17th century, depleted impasto builders, but now the beaver population is slowly recovering. This is good news for many ecosystems, as beaver construction creates valuable habitat for endangered species, traps carbon, and improves water availability in arid places.
Despite these ecological impacts, large-scale mapping of beaver habitat is lacking from scientific research. Most mapped dams are identified manually, which takes a lot of time and effort.
To accelerate the identification of beaver dams, Emily Fairfax and colleagues apply machine learning to scrutinize high-resolution geospatial imagery and examine potential dam complexes across landscape- and regional-scale areas. I was. Researchers developed an Earth Engine Automated Geospatial Element(s) Recognition (EEAGER) model that uses a neural network trained on the known locations of thousands of beaver dams contained in aerial and satellite imagery. their research Geophysical Research Journal: Biogeoscience.
For this study, the team trained the model to identify dams in the western United States, then tested it outside the training region to see if it could correctly detect other dams in the images. Overall, EAGER showed him 98.5% accuracy in characterizing whether sites in imaged landscapes had dams. Model recall (percentage of known dams correctly identified by the model) and accuracy (percentage of dams predicted by the model that were actually dams) were both 63% and 26%, respectively, with additional training may be improved by Beavers live in different regions, the authors note.
Still, the relatively high recall is a good sign, they say, suggesting the model can detect the majority of real dams. They also noted that the false-positive dam identities could easily be manually removed from the catalog, and that many of the false-positive dams were close to real beaver dams.
The study will help track beaver populations and health, ecosystem change and beaver restoration of rivers, and the methodology could be applied to monitor other complex terrain over large areas. , the researchers point out.
For more information:
Emily Fairfax et al, EEGER: A Neural Network Model for Finding Beaver Complexes in Satellite and Air Imagery, Geophysical Research Journal: Biogeoscience (2023). DOI: 10.1029/2022JG007196
This article is republished courtesy of Eos, sponsored by the American Geophysical Union. Read the original story here.
