Close

Phase Two implementation (2014 Sea Turtle Season)

A project log for Turtle Sense

Cell phones for Sea Turtles and beyond -- creating an extremely low powered remote data recorder and sensor for monitoring wildlife, etc...

Samuel WantmanSamuel Wantman 08/05/2014 at 20:280 Comments

Test units started being installed in sea turtle nests in June of 2014. The plan is to monitor about 20 nests at the Cape Hatteras National Seashore and a few more at the Bald Head Island Conservancy, both in North Carolina.

The units constantly monitor and analyze motion to create a profile of its magnitude over time. The motion detector measures the change in acceleration (or "jolt") multiple times per second. The magnitude of the jolt is placed on a logarithmic scale divided into 25 different ranges. The ratio between the lower and higher limit of each range is the square root of 2. A counter for each of the different ranges is incremented whenever a jolt reading is in that range. You can think of each counter as a bin. If a magnitude is in range, it is thrown into the bin for that range. After 6 minutes the results are stored in a record along with a temperature and orientation reading, and a new set of bin counters is started. By looking at how many readings were in each bin, we can get an idea of what jolts occurred while the bins for that record were being filled. This allows us to compress several thousand readings into approximately 32 bytes of information. We lose precision (including the exact sequence) with all the readings, but we suspect that those details are not important. The 240 records created each day give us a very good idea of what is happening in the nest.

Research indicates that before emerging from the nest in a "boil," turtle hatchlings congregate underground near the top of the nest. It is thought that this motion stimulates the hatching of the turtles that haven't yet emerged. Our sensors, situated at the top of the nest, should record some large disturbances when the first turtles emerge. It is our hope that our readings will indicate these early hatchings so that we can predict that a boil is imminent.

The nests with first units installed should be hatching around mid-to-late August. We will be generating algorithms to predict hatching from our data, and we plan to refine those algorithms so that we can come up with a reliable process for predicting hatching a few days in advance.

Discussions