Researchers at Cornell Lab of Ornithology, University of Massachusetts, and University of Illinois have developed breakthrough methods to track the migration of individual bird species by combining participatory science data with weather radar technology. This advancement addresses a long-standing limitation in migration monitoring: while radar can detect birds in flight, it cannot identify which species are migrating.
The research, published in two papers in Global Ecology and Biogeography and Movement Ecology , introduces novel approaches that provide species-specific migration estimates across North America using data from over 2 billion bird observations submitted by citizen scientists to eBird. This new research is part of the BirdFlow project, a project from the Cornell Lab of Ornithology and the University of Massachusetts that uses AI models to predict movements of bird populations during migration using data from the Cornell Lab of Ornithology’s eBird Status & Trend project.
The first new method the team developed is called BirdFlow Migration Traffic Rate (BMTR) that uses data from over 2 billion bird observations submitted by participatory scientists to eBird to provide species-specific estimates of migration patterns across North America. The model successfully identified major flyway patterns and provided weekly species-specific migration estimates, even in areas with gaps in weather radar coverage, according to the team. "BirdFlow opens up exciting new directions for monitoring and forecasting bird migration in real-time," said Adriaan Dokter, project leader for BirdCast and BirdFlow at Cornell Lab. "The new BMTR metrics allow us to estimate the most likely species responsible for the movements we detect with radar, which detects the numbers of birds migrating aloft but not which species."
The second approach demonstrates how BirdFlow models incorporate data from individually tracked birds using GPS, Motus radio telemetry, and banding data to reconstruct population-level movements across the continent. The team produced migration models for 153 North American migratory bird species spanning 14 orders and 39 families.
To validate their methods, researchers compared BirdFlow-derived estimates with 28 years of data from 152 weather surveillance radars across North America, finding strong correlations that confirm the new method's accuracy. The models were also validated against real GPS-tracked birds, demonstrating biologically realistic migration route predictions.
"We find that incorporating such individual and species-specific differences—as captured directly by tracking and banding data—greatly improve our population-level movement models," said Dokter. "We like to think of BirdFlow as a way of synergizing all the available information on the movements of individual species that is out there."
The innovation has immediate practical applications for bird conservation, particularly in reducing collision risks during peak migration periods. "Different bird groups are prone to different levels of risk of colliding with windows," explained Yuting Deng, a postdoctoral associate at Cornell Lab who participated in both studies.
The method also shows promise for tracking disease spread. Agencies and researchers are collaborating with the research team to use BirdFlow models for monitoring avian influenza transmission routes, particularly for waterfowl species. “By resolving these population-level movements, BirdFlow can support research and applications in migration ecology, conservation planning, disease surveillance, aviation risk assessment, and public outreach,” said Yangkang Chen, a PhD student at University of Illinois Urbana-Champaign, who lead the fine-tuning of the BirdFlow models.
These new methodologies also make it possible to study migration at a scale that was previously difficult: across entire species ranges, across the full annual cycle, and across hundreds of species. Understanding how migration varies is important because bird populations within the same species may use different routes, face different threats, and experience different environmental conditions.
The research team envisions integrating BMTR into existing monitoring systems like BirdCast, which currently provides migration forecasts but lacks species-specific information. The BirdFlow project has also published a new model collection expanding from 4 to 60 vetted models for use with BirdFlowR software.
The methodology could potentially expand worldwide given adequate citizen science data coverage, offering hope for global migration monitoring capabilities.
###
Plunkett, E., Deng, Y. and D. L. Slager, M. Fuentes, Y. Chen, B. M. Van Doren, A. M. Dokter, and D. Sheldon (2026). Novel Estimates of Bird Migration Traffic at the Continental Scale Using Participatory Science Data. Global Ecology and Biogeography https://doi.org/10.1111/geb.70236
Chen, Y. and D. L. Slager, E. Plunkett, M. Fuentes, Y. Deng, S. A. Mackenzie, L. E. Berrigan, D. Fink, D. Sheldon, B. M. Van Doren, and A. M. Dokter. (2026). Population-level migration modeling of North America’s birds through data integration with BirdFlow. Movement Ecology. https://doi.org/10.1186/s40462-026-00651-z
The Cornell Lab of Ornithology is a nonprofit, member-supported organization dedicated to the understanding and protection of birds, wildlife, and our shared planet through research, education, participatory science, and conservation. birds.cornell.edu
Kathi Borgmann, Cornell Lab of Ornithology, (607) 254-2137, klb274@cornell.edu
Global Ecology and Biogeography
Data/statistical analysis
Animals
Novel Estimates of Bird Migration Traffic at the Continental Scale Using Participatory Science Data
18-Apr-2026