Earlier this week the glider team pulled in the University of Delaware’s glider, nicknamed the “Blue Hen,” from the ocean. They needed to change its batteries and download one of the instrument’s data. But they didn’t want to miss what was happening out at sea, so they pulled what they call a “Nascar turnaround,” working like a pit crew to get the work done overnight.

It gave us an opportunity to look at the inside of a glider and explore how these complicated robots manage to go where they’re told and collect the data we need, practically by themselves.

To fly, a glider needs a few basic abilities: it needs to be able to control its buoyancy so that it can either sink or rise; it needs to know where it is and where it’s going; it needs to be able to call home for new instructions; and it needs to be able to collect data. And it has to do all this with only a small allowance of space, weight, and battery power. Click through the slideshow to see how gliders manage it:

The stream of data that comes back from the gliders every hour or so is what allows the CONVERGE team to be adaptive in their sampling. With a research ship or an instrument fixed on the seafloor, it’s difficult to change your plans as you get more information. But changing a glider’s plans is as simple as uploading a new instruction file.

Over the last couple of weeks the team has settled on a sampling plan they’re calling the “iron cross.” Each of the three gliders takes a different part of the cross (see the map), and together they reveal what the waters of the Palmer Deep are doing. The Blue Hen (blue on the map) flies to the center of the cross and just goes up and down in the same spot. The glider from the University of Alaska, Fairbanks (green), flies a straight line along the length of the Palmer Deep canyon. The Rutgers RU05 glider (red), flies across the canyon.

RU05 goes at right angles to the Alaska glider, and those two gliders cross at the spot where the Blue Hen is working. Altogether, this gives the scientists a way to separate out ways the water conditions could changing.

Here’s an example of the data they’re collecting, and why it helps to have data from more than one glider. This graph shows one round trip of the Alaska glider from the northeast end of its route, down to the southwest end, and back. The dashed line shows where the glider turned around—the two parts are different lengths because it took longer to fly out to the dashed line than to fly back. That’s because it was fighting the current in the left section of the graph; then it turned around and rode with the current.

The graph shows the temperature of the water as the glider flew up and down on its way. Do you see any differences? First of all, it’s interesting that the water is warmer at the surface and at the bottom (orange colors), and coldest in the middle (blue colors). This is because the sun has warmed the waters at the surface; the middle waters are still cold from winter; and the deepest waters represent a deep current that has brought in warmer water underneath it all. Layering of different water types is a very common feature in the ocean.

Now look at the left and right edges of the graph. Remember that the glider was doing a round trip, so these two parts of the graph represent the same patch of water at the northeast end of its route. On the left side compared to the right side, there’s less warm water at the surface and the colder middle water doesn’t extend as deep.

This is an interesting development, but it leaves the scientists with a problem: did the water change in temperature between the glider’s first pass and its second pass? Or did a current carry in different water, with a different temperature (as well as other characteristics) and push the original water out of the way? With only one glider taking measurements, it would be hard to figure this out. But with three gliders working the same stretch of water from different angles, they’re much more likely to be able to put together the answer.

Huge thanks to Laura Palmara for putting the map and data plot together for this post.