Limitations and the Future

The Extended Kalman Filter is a very valuable tool.  Animal tracking is nonlinear in nature, and as can be seen by our results, the Extended Kalman Filter does a better job of estimating the true trajectory over the Least Squares method.  The results of the Extended Kalman Filter is extremely important, because researchers use the results from animal tracking to draw conclusions about animal migration and other environmental topics.

Despite its proven importance, the future of Argos is in danger.  The oldest Argos satellite was launched in 1998, and is now operating over 15 years past its projected lifespan.  The launch of a new Argos receiver to replace the old is being pushed back to the year 2022, according to Markus Horning (2017), a professor at Oregon State University and Science Director at Alaska Sealife Center.  However, the launch of the new satellite may not happen if funding is lacking from the federal government.  This is a major issue as the future of animal tracking is in jeopardy, as there are about 22,000 active Argos transmitters out there.

We also need to note a limitation with our project.  A limitation was that we used real data in conjunction with simulated data to obtain the results shown.  For instance, the platform and satellite locations, as well as the platform transmitting frequencies, are real data that we obtained through Movebank and  However, the measured results we obtained are not the actual measured data, because we simulated white noise.  We know from prior research that white noise is an accurate representation of the actual measurement noise, but we do not have the values of the actual measurement noise that was added.  Unfortunately, despite spending weeks trying to acquire this real data, we had no luck.  We emailed an employee at Argos, and asked if it was possible to acquire real data, however, we would need to pay a set amount to be able to access it and it would take a couple weeks to get a hold of.  In addition, without getting a glimpse of the data beforehand, we cannot know for sure if the data is even applicable to our project.  However, our project still serves as an example of the advantage of the Extended Kalman Filter over the Least Squares.