Hierarchical Bayesian computing of 3-dimensional whale trajectories from tag and acoustic data
Speaker: Christophe Laplanche (EcoLab > – Laboratoire d’écologie fonctionnelle )
Researchers mainly use tagging and passive acoustics to study the underwater behavior of toothed whales in the field. Both approaches have divergences which mainly originate from dogmatic (how much tagging/recording alters whale behavior?) and technical (equipments are different and so are the way they are operated and the information they provide) aspects. Are tagging and passive acoustics that different? Both approach share the common goal of computing 3-dimensional whale trajectories from noisy data collected on a network of sensors. We actually show that an identical approach – under the form of a hierachical Bayesian model (HBM) – can be used to compute 3-dimensional whale trajectories from tag (including depth meter, accelerometer, magnetometer, gyroscope, GPS) and/or passive acoustic data. This approach has the advantages of (1) efficiently extracting information from the data, (2) efficiently merging different sources of data, and (3) propagating measurement errors to whale trajectory estimates. We illustrate the capabilities of the approach with simulated and field data.