Using GPS data to evaluate the accuracy of state-space methods for correction of Argos satellite telemetry error.

摘要: Recent studies have applied state-space models to satellite telemetry data in order remove noise from raw location estimates and infer the true tracks of animals. However, while resulting may appear plausible, it is difficult determine accuracy estimated positions, especially for position interpolated times between locations. In this study, we use two gray seals (Halichoerus grypus) carrying tags that transmitted Fastloc GPS positions via Argos satellites. This combination Service highly accurate allowed examination their uncertainty derived data. After applying a speed filter aberrant locations, fit continuous-time Kalman estimate parameters random walk, used smoothing at measurements, then compared filtered with actual measurements. We investigated effect varying maximum thresholds speed-filtering algorithm on root mean-square error (RMSE) minimum RMSE as criterion guide final choice threshold. The optimal differed animals (1.1 m/s 2.5 m/s) retained 50% 65% each seal. using 1.1 resulted very similar both For seals, Kalman-filtered were 5.9 12.76 km, respectively, 75% modeled had errors less than 6.25 km 11.7 Confidence interval coverage was close correct typical levels (80-95%), although tended be overly generous smaller sizes. reliability also affected by chosen filtering allows effective calculation indicates limits when correcting service locations linking spatial covariate habitat