Developed as a collaborative effort between researchers at the Cornell Lab of Ornithology and UMass’s College of Information and Computer Sciences in 2018, MistNet is a machine learning tool built to pull bird location data from weather radar datasets to generate decades-long bird migration information. The deep learning aspect of MistNet has helped address the mass amount of data—over 200 million images and hundreds of terabytes—needed to create continent-scale temporal patterns of bird movements, with audio and imagery recognition mimicking human neural networks. Researchers were able to extract bird data from weather patterns in radar images to construct migration maps and animations ranging over the past 24 years to highlight the most intensive migratory areas across the United States. They were also able to calculate flight velocity and traffic rates. This tool uses the long-standing weather radar archives in the United States, but also has the capability to address data collected from various citizen science apps created by the Cornell Lab of Ornithology.

University of Massachusetts Amherst. (2019, October 9). “Using Artificial Intelligence to Track Birds’ Dark-of-Night Migrations“.


Artificial Intelligence, Biodiversity, Climate Change, Ecological Modelling, Ecological Monitoring, Monitoring, Visual Technologies