Incorporating sea-surface temperature to the light-based geolocation model TrackIt

Archival and pop-up satellite archival tags have been widely used to study the movement dynamics in many marine pelagic species. Recent advances in light-based geolocation models have enabled better estimation of the geographical positions of tagged animals. In particular, TrackIt, a state-space model with the Kalman filter, uses only light data from a tag for estimating positions and does so independently of manufacturer calculations.

This approach is a complete break from previous geolocation methodologies, which rely on manufacturer-processed positions as an input. In this paper, a unified model is presented to extend TrackIt to incorporate satellite sea-surface temperature (SST) matching, an approach that has been demonstrated to improve accuracy.

The performance of various satellite SST imagery products is also evaluated by the comparison of SST-inclusive models against the basic TrackIt model using only light information. Three new model parameters (bias, error and smoothing radius) are introduced for handling SST observations. Analyses based on double-tagging comparisons show that the overall accuracy of TrackIt increases with the incorporation of SST, even when the resolution of the matching satellite product is rather coarse.

At the same time, the model accuracy can decrease when the SST observations do not exhibit any strong trends, rendering SST matching less informative. The incorporation of SST within this generic and statistically sound modeling framework illustrates how TrackIt can readily be extended to utilize new data streams, such as geomagnetic data, which will become available with the next generation of archival tags.