BirdCast proposed to provide real-time predictions of bird migrations: when they migrate, where they migrate, and how far they will be flying. And the time has arrived!
Knowledge of migratory behavior will aid conservation on the ground by informing decisions for mitigation of numerous hazards including identifying nights on which lighting of tall buildings and structures could be eliminated to prevent the deaths of millions of birds. Accurate migration models also have broader application, allowing researchers to understand behavioral aspects of migration, how migration timing and pathways respond to changing climate, and whether linkages exist between variation in migration timing and subsequent changes in population size. Beginning in 2018, after many years of research and developments in machine learning, cloud-based computing, and big data analytics, the BirdCast site will feature migration forecasts that predict how many birds will be aloft over the continental US and live migration maps that report how many birds actually took flight.
Project Goals: Techniques, Infrastructure, and Visualizations
BirdCast collaborators at Oregon State University and University of Massachusetts Amherst proposed two innovative machine-learning techniques: Collective Graphical Models (CGMs) and Semi-Parametric Latent Process Models (SLPMs). These techniques would provide the abilities to identify the complex conditions governing the dynamics of migration behavior, including choice of migratory pathways, the factors that influence when birds migrate, and the speed and duration of each night’s movements.
The BirdCast team also planned to develop new interoperable data infrastructure for synthesizing bird observations, flight calls, radar data, and covariate data from multiple sources including satellite imagery, weather, and human population data. Some recent work by Frank La Sorte and Kyle Horton has furthered plans for understanding the relationships between radar and eBird data, and additional peer-reviewed research is currently underway on this subject with new NSF awards (for example, see Dark Ecology and BirdVox).
Finally, BirdCast proposed novel web-based data visualizations for communicating the migration predictions generated by BirdCast to the general public. The BirdCast site has been providing interpretation and information since 2012 towards these ends, but the team is particularly excited to provide new visualizations and interpretive tools beginning in the spring 2018 season that speak to the original vision and spirit of the project’s intent.