|
Abstract:
|
Animal ecologists readily collect nearly continuous spatio-temporal data on animal locations via various satellite telemetry technologies. In addition, there is a wealth of readily available, complex environmental data via geographic information systems. Though numerous modeling approaches have been proposed for studying the relationship between animal locations and their environment, many are self-admittedly oversimplified or computationally infeasible in high dimensions. Given that computational implementations of all continuous models must be discretized on some level, we propose a framework that mimics the discrete spatio-temporal processes directly. This method allows for the statistical modeling of dynamic spatio-temporal movement behavior while remaining computationally tractable in high dimensions (e.g., simultaneously modeling several distinct individuals).
|
