GPS Collars and Accuracy to Study Sloths
Studying the GPS locations of an animal that doesn’t move much is challenging compared to studying species that travel a lot, such as whales and cheetahs. When studying sloths, accuracy is what we need to focus on.
Today, we want to show you what the GPS data from the collars we use looks like, and how we work with it for the Urban Sloth Project.
Sloths and Dots
In the image below, you can see the GPS locations, represented in the green and red dots, from Lee, one of our Urban Sloths. This data is still “raw,” and needs processing: it is not as accurate as we need it to be yet (you’ll probably notice the dots in the middle of the ocean!).
The dots are generated every three hours, and are recorded based on the satellites available at that moment: the green dots are the closest to correct and have been generated by triangulating the signal from multiple satellites, while the red dots indicate an approximate location (because there were much less satellite signals at that precise moment).
So the red dots are not precise, but some of the green ones aren’t either…
How do we know which dots are the most accurate?
We filter the data to get to the most accurate GPS points by HDOP and satellite availability:
- Horizontal Dilution of Precision (HDOP) is a measure that lets us know how accurate our GPS readings are specifically in the horizontal plane. A lower HDOP value signifies higher accuracy, while higher values indicate less accuracy in the positional data.
- Satellite availability. The software also allows us to see the number of satellites within range when the GPS point was recorded. For the GPS point to be reliable, at least five satellite signals need to have been triangulated.
Finally, we overlap the filtered data with the GPS points we collect manually while tracking on the field. While we can’t be 100% on the accuracy of every single data point, this will determine the most accurate ones and the ones that can be discarded.
This is just the beginning
The sloths’ GPS locations are just one variable in the big picture of the Urban Sloth Project: in reality, we also have literally billions of different data points, such as noise levels, ambient temperature, humidity, height, and other relevant information, plus all the activity data. It’s a really long and slow road until we see the first results!
Amelia Symeou
Data Scientist /Urban Sloth Project
