Home

Animal tracking is useful in understanding a variety of topics.  These topics include animal migration and climate change.  There are a variety of ways to determine the location of an animal, and one such way is by using the Argos Doppler, a satellite system that was launched into service in 1978.  Other methods include using GPS (Global Positioning System) and radio transmitters.  Each method has its merits and drawbacks, but we will specifically be addressing how to accurately track the position of an animal with the Argos Doppler.  It is more cost-efficient and cheaper to use than GPS and more convenient than using a radio transmitter.  Argos is used to track free-ranging animals, however, one drawback to this system is that location tracking generates noisy measurements ranging in error from a few meters to hundreds of kilometers.

Fig. 1. Shows the ARGOS position measurements of 15 harbor seals tagged in the central Aleutian Islands from September 2016 to January 2017. This data was obtained from a contact of Professor Trobaugh.

The Argos system uses polar, low-orbiting satellites to collect near real-time data from a platform (the transmitter on the animal of interest).  We will develop a realistic forward model, add white noise, and code and test the Least Squares and Extended Kalman Filter on the forward model.  We wish to recreate results similar to Figure 2, which is an example of how the Kalman Filter works better for the animal tracking problem than does Least Squares.

Fig. 2. (a) Least Squares was used to estimate true trajectory of elephant seal tracks in red. (b) Multiple-model Kalman Filter is the green line, and better estimates the true trajectory of elephant seal tracks. These figures were obtained from “Improving Argos Doppler Location Using Multiple-Model Kalman Filtering” by Lopez et. al.

The forward model describes how a satellite communicates with the platform.  To develop a realistic forward model, we will use data on the trajectory of the Satellite with Argos and Altika, also known simply as SARAL, as well as the movement of a false killer whale, to calculate the frequency received by the satellite () from the platform transmitting frequency ().  This is because the Argos Doppler uses the Doppler effect, which uses the shift in frequency between measurements to calculate changes in the transmitter’s position.

Fig. 3. SARAL is a satellite carrying Argos receivers and false killer whales are tracked using Argos. These are inputs into the forward model.

We will use MATLAB to code both estimators of interest.  Once we have coded both estimators, we can conduct simulations on the forward model created to determine which estimator better estimates the true trajectory of an individual animal’s movement.

Fig. 4. The platform and Argos communicate with each other. The true animal movement is estimated from the noisy measurements received by Argos.

Alison Gu
ESE Undergraduate
xinyungu@wustl.edu
Rebecca Ho
ESE Undergraduate
ho.rebecca@wustl.edu
Natasha Zachariades
ESE Undergraduate
nzachariades@wustl.edu
We would like to thank Professor Hoven, Professor Trobaugh, and Dr. Zhang for all of their guidance and support!  Their ideas and feedback were so helpful and we are very grateful to all of them.
 –
Professor Randall Hoven
Lecturer
School of Engineering and Applied Science
hovenr@wustl.edu
Professor Jason Trobaugh
Lecturer
School of Engineering and Applied Science
jasont@wustl.edu
 –
Dr. Jinsong Zhang
Lecturer
School of Engineering and Applied Science
jinsong.zhang@wustl.edu