A population Monte Carlo model for underwater acoustic telemetry positioning in reflective environments

Abstract Underwater acoustic telemetry positioning is widely used to track the fine‐scale movements of aquatic animals. In study areas near acoustically reflective surfaces, reflected transmissions may cause large detection outliers that can severely reduce the accuracy of positioning models. A novel time‐of‐arrival model for telemetry positioning is presented that utilizes a population Monte Carlo algorithm to solve positions (termed PMC‐TOA). Telemetry detection error is modelled as a mixture distribution, allowing reflected detections to be identified and positions to be estimated despite their presence. Importantly, the PMC‐TOA model provides good measures of positioning uncertainty, facilitating the use of post‐processing state‐space models to further refine position estimates. A simulated telemetry study is used to validate the PMC‐TOA model and compare its performance to a conventional time‐difference‐of‐arrival positioning model. A real case study on Atlantic salmon ( Salmo salar ) smolt passage behaviour is further used to demonstrate how PMC‐TOA can be combined with post‐processing models to produce high‐resolution tracks. The resulting tracks are compared against those resulting from YAPS and TDOA positioning. The PMC‐TOA model was shown to work well as either (i) a pre‐processing step to remove reflected transmissions from time‐of‐arrival datasets, or (ii) a fast and accurate positioning method when paired with a post‐processing state‐space model. Positions returned by the model can be further used for animal movement statistics, allowing researchers to test the effects of experimental or environmental factors on the fine‐scaled movement behaviours of aquatic animals in acoustically challenging environments.