Root Mean Square Error (RMSE) estimates (km) between modelled Argos locations and the true (GPS) position at varying sensitivities for three commonly-used location error correction models freely-available within the R statistical framework; crawl [14], bsam [17] and tripEstimation [31].

‘Old’ and ‘New’ reflect the effect of applying the original error structures derived from data in [10] and errors estimated from data in the present study, respectively. Variable ‘N’ for the tripEstimation results reflects the reduction in the pre-processed dataset. For all models, the new error structures incorporated a correction for differences in the distance covered by one degree of longitude at high latitude was applied. Optimal model results are highlighted in bold italics. Both bsam and tripEstimation models performed better with error structures derived from the current study, providing the most accurate location estimates at the lowest speed threshold with the exception of the bearded seal bsam model. Sensitivity of models to different speed thresholds was apparent, though the differences in error estimates were minimal.
Root Mean Square Error (RMSE) estimates (km) between modelled Argos locations and the true (GPS) position at varying sensitivities for three commonly-used location error correction models freely-available within the R statistical framework; crawl [14], bsam [17] and tripEstimation [31].