The intrepid Autosub. Credit: BAS

When the Autosub 3 robot submarine began its pioneering 110-kilometre round trip under the Antarctic ice, there was no guarantee that it would ever come back.

Its sister craft, Autosub 2, had been lost on a similar mission in 2005. Autosub 3 was being sent on a much more ambitious mapping expedition, in an environment in which escaping to the surface is not an option.

But it was also packing the latest technology. The 7 metre-long, 3.5 tonne autonomous robot can dive to 1,600 metres under the surface of the ocean and travel 400 kilometres, powered only by 5,000 standard D-cell batteries.

Shortly after the joint UK-US mission was launched from the RV Nathaniel B. Palmer icebreaker in January, the submarine set out under the Pine Island Glacier, which juts out into the Amundsen Sea.

To find out what happened next, Nature spoke to the technical and scientific leaders of the Autosub team: Steve McPhail of the UK National Oceanography Centre in Southampton and Adrian Jenkins of the British Antarctic Survey in Cambridge.

Why was Autosub 3 sent on this mapping mission?

Adrian Jenkins: Pine Island Glacier is both thinning and accelerating. It's one of the fastest changing regions in Antarctica at the moment. What we're trying to do is understand what is driving those changes.

This thinning has been measured by radar altimetry. It's strongest on the floating parts of the glacier. The favoured hypothesis is that it is a change that has been driven by the oceans — there is some change in the supply of ocean heat and the floating parts of the glaciers, the ice shelves themselves, are now thinning. Because of the thinning, the glaciers feeding the ice shelves flow faster.

How risky was the mission?

Steve McPhail: Only the Autosub AUVs [automated underwater vehicles] have done anything like this before. In 2005, Autosub 2 went about 30 kilometres under the Fimbulisen Ice Shelf [and returned]. Unfortunately, three days later we repeated the mission and Autosub 2 was lost. Because there wasn't any telemetry coming from the lost Autosub, we couldn't be certain what the cause of the problem was. The most likely single cause was a failure in the power system because of a failure in a connector, or a failure in the communications network.

What happens on an Autosub mission?

The autosub was launched from the Nathaniel B. Palmer. Credit: James Perrett, National Oceanography Centre, Southampton

SM: At the start of each mission we check the vehicle out. Then we dive the vehicle to about 1,000 metres. Radio waves don't go through the water so we're using our acoustic system to get messages back from the vehicle, so we can check everything looks correct.

We watch the AUV and check it out for up to 5 hours. Once we're happy we send a message and off it goes under the ice shelf. You can track it for about an hour or two, by which time it had gone 10 kilometres. At that point it is just too far away to communicate with and it is off on its own to do the mission for 24 hours.

So did Autosub 3 make it?

SM: We were very happy and relived to have successfully done six missions. The longer the period under the ice shelf the greater the risk, and the further you go the closer you get to the grounding line — the point at which the ice is touching the seabed.

The sub does have a collision-avoidance system, so if it sees an object ahead it will attempt to get round it. The mission was programmed that once the AUV detected that the difference in depth between the bottom of the ice and the sea bed was below a certain threshold the AUV turns round and comes back.

It's an area where if anything goes wrong with your vehicle you're probably likely to lose it. It's also the case that you don't really know anything about the environment you're going into, it's totally unexplored.

What information did Autosub 3 bring back?

AJ: It has two sets of acoustic instruments that give a range to the ice base above the sub, and to the seabed below.

The ice base is a very interactive system. The flow of water is what brings heat to the ice-shelf base, and it is the melting of the ice shelf base that sets the shape of the base. And of course, the shape of the base controls the flows that deliver the heat to it.

We do now have a much clearer picture of the topography. The seabed topopgraphy was totally unknown before the sub went in there. We had some idea of the large scale structure of the ice, but we will certainly come away with a much better picture of how the warm water gets to the crucial regions.