What goes up must come down: Using weather surveillance radar data to forecast migratory birds banded on the ground
Open Access
Abstract
Ecological forecasting is a powerful framework to predict future states of ecological systems, including species distributions and supporting proactive conservation, management, and policy decisions that exist in a rapidly changing world. Forecasting efforts span a diversity of systems and organisms, yet many ecological phenomena, especially those involving highly mobile species, remain difficult to predict. Migration, for example, shapes population dynamics, species interactions, ecological services, and disease spread across terrestrial and aerial environments. Migratory pulses of birds observed in airspaces and stopover habitats are often studied separately; however, the connection between these two phases of migration remains largely unquantified. Here, we evaluate whether weather surveillance radar (WSR) measures of nocturnal migration activity can predict the number of birds captured at a long‐term bird banding station, thereby bridging aerial and terrestrial perspectives of migration. Using data from two WSR sites—KBOX (Boston, MA) and KGYX (Portland, ME)—and bird banding records from Appledore Island, Maine (1996–2023), we examined how radar‐derived measures of aerial migration activity forecast terrestrial bird banding captures. Using leave‐one‐year‐out cross‐validation, generalized additive models incorporating radar measures, atmospheric conditions, and non‐environmental covariates produced skillful probabilistic forecasts of daily captures in both seasons, improving on a persistence baseline in 21 of 23 spring and 23 of 26 fall years (median CRPS skill 0.25 and 0.12). Including radar‐derived measures further improved forecasts beyond atmospheric predictors in the majority of years (spring: 18 of 23; fall: 21 of 26). The models explained 57.8% and 49.9% of the deviance in daily spring and fall captures, respectively. These findings show that linking aerial and terrestrial monitoring can enhance migration forecasts, with implications for conservation strategies such as targeted “lights out” initiatives and stopover habitat management. This approach offers a scalable pathway to forecasting the connection between large‐scale atmospheric data and ground‐level ecological processes.Keywords
Funding Information
- National Aeronautics and Space Administration (80NSSC21K0930)
- National Aeronautics and Space Administration (80NSSC26K0151)
- Amon G. Carter Foundation
This publication has 91 references indexed in Scilit:
- Structuring the skies: Diel dynamics of migratory animal movement in the lower atmosphereEcology, 2025
- Long‐term analysis of persistence and size of swallow and martin roosts in the US Great LakesRemote Sensing in Ecology and Conservation, 2023
- Prospects for monitoring bird migration along the East Asian‐Australasian Flyway using weather radarRemote Sensing in Ecology and Conservation, 2022
- Evaluating probabilistic ecological forecastsEcology, 2021
- Weather and eared grebe winter migration near the Great Salt Lake, UtahInternational Journal of Biometeorology, 2017
- Birds: blowin’ by the wind?Journal of Ornithology, 2006
- Estimating regression models with unknown break‐pointsStatistics in Medicine, 2003
- Bird Migration Speed2003
- Optimum Fuel Loads in Migratory Birds: Distinguishing Between Time and Energy MinimizationJournal of Theoretical Biology, 1997
- Forecasting spore concentrations: A time series approachInternational Journal of Biometeorology, 1990