Continuous-Time Modeling of Marine Mammal Telemetry Data

• Activity Number: 326
• Type: Topic Contributed
• Date/Time: Tuesday, July 31, 2007 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Devin Johnson*+ and John Durbin and Mary-Anne Lea and Joshua London
• Companies: National Marine Mammal Laboratory and National Marine Mammal Laboratory and National Marine Mammal Laboratory and National Marine Mammal Laboratory
• Keywords: ARGOS ; Continuous-time ; Animal movement ; Stochastic process ; Marine Mammals ; Telemetry data

Abstract

Marine mammal telemetry data is obtained by determining the location of an animal in space at several points in time. The observed locations are function of two important factors, true location and measurement error. In order to handle the fact that most telemetry data is collected opportunistically on an irregular basis we consider a movement model stochastic process in continuous time. Using this approach likelihood can be formed using the raw data instead of aggregated, thinned, or interpolated data. Bayesian methods are explored for making inference on animal locations as well as other movement quantities of interest, such as travel speed. The Bayesian inference paradigm allows for inclusion of parameter uncertainty in location estimation, as well as, propagation of uncertainty through nonlinear relationships in the movement parameters of interest.