The Netherlands Bird Avoidance Model, Final Report

The NL-BAM was developed as a web-based decision support tool to be used by the bird hazard avoidance experts in the ecology unit of the Royal Netherlands Air Force. The NL-BAM will be used together with the ROBIN 4 radar system to provide BirdTAMS, for real time warnings and flight planning and to give an overview of the predicted bird hazards in the Netherlands to air traffic controllers, flight coordinators, squadron leaders and airfield bird control units. The NL-BAM was developed by a multidisciplinary team and through the cooperation of the people from the Institute for Biodiversity and Ecosystem Dynamics (IBED) and the Informatics Institute of the University of Amsterdam, the Royal Netherlands Air Force and SOVON, the Dutch Centre for field Ornithology.
The NL-BAM is accessible to the general public on the Internet at http://ecogrid.sara.nl/bambas. The website and models are divided into two modules, 'migration prediction' and 'spatial distribution'.
The migration module of the NL-BAM is a dynamic model and predicts the density of migrants during the day and at night for 3 consecutive days based on real-time and/or forecasted meteorological data. During the migration season, migration forecasts are updated daily by members of the RNLAF. Predictions are presented in figures on the NL-BAM website. The autumn nocturnal forecasts are based on three data-driven models: a log-linear regression model, a conceptual model and a neural network model. The spring forecasts include a log-linear regression model and a neural network model. Diurnal forecasts are based on nocturnal forecasts by the log-linear relationship between nocturnal and diurnal migration intensities.
The spatial distribution module includes spatial density maps of the 62 bird species selected as most relevant for flight safety in the Netherlands in bi-weekly intervals, four time periods per day and at five altitude layers. This module is based on historic field observations and represents the density distribution expected in average circumstances. The key species were selected based on criteria such as total number of birdstrikes, number of damaging birdstrikes, availability of data, mass, behaviour of birds, expert knowledge on species and conservation importance. The spatial distribution of each species was modelled in relation to environmental conditions using a combination of regression models, spatial statistics and expert knowledge. Furthermore, the altitude distribution of several species of birds was modelled in relation to meteorological conditions. Expert knowledge and additional sources were used to complete information on the altitude distribution, temporal activity (daily), annual abundance and flight activity. The public access website includes composite maps of mass and separately for number of birds/km2 for each time of year, time of day and altitude combination (480 combinations). An authorized access page, accessible only by team members, also includes individual maps for each species as well as a summary table including total numbers of the top 10 most abundant species per region for each time of year, time of day, altitude combinations.
Finally, although much has been accomplished in the last few years, it is clear that much more can be still be achieved in the field of bird strike avoidance research. We hope that future projects will emerge and be as productive as this one, if not more.