Intrepid explorer: an oceangoing rover gathers unprecedented data.

The autonomous underwater vehicle, stuffed to its gills with scientific instruments, motors steadily through the frigid, sunless sea. Sensors on board the 6.8-meter-long craft are constantly on alert, testing the water for minute variations in salinity and temperature. As the vehicle cruises beneath a ceiling of ice hundreds of meters thick, its sonar observes above it a rugged topography of ice unlike any previously seen.

Although scientists had previously deployed autonomous underwater vehicles (AUVs) beneath free-floating ice, until last year they'd never sent an AUV under one of Antarctica's ice shelves. On that 50-kilometer maiden voyage, the craft got a tantalizing look at one of the world's most unexplored environments-one where seas increasingly warmed by modern climate change come into contact with ancient ice flowing off the continent.

Understanding what's taking place under ice shelves is vital for predicting how those features will evolve in a warming world--for example, how fast they'll shrink or break apart, and how much fresh water they'll provide to the oceans as they melt. Scientists have developed computer models that simulate these processes, but those models are coarse.

The fine-scale data that AUVs can gather could enable researchers to create more-detailed computer models as well as to verify or fine-tune the results from existing ones, researchers say. Cameras taken along for the ride could give biologists their first look at what life is like in such an alien environment.

ON THE SHELF Ice-covered Antarctica, the world's third-largest continent, covers about 14 million square kilometers. About 90 percent of that ice sits on land. The rest, though still firmly attached to the continent's ice sheet, floats in the surrounding seas, says Start Jacobs, a glaciologist at Lamont-Doherty Earth Observatory in Palisades, N.Y. These ice shelves, many of which are hundreds of meters thick, occupy about 44 percent of Antarctica's coastline. In some places, their edges where Connecticut-size icebergs occasionally snap free and head for warmer climes (SN: 5/12/01, p. 298)--are hundreds of kilometers from land.

Shipbound Antarctic scientists, even those on icebreakers, can't directly explore areas covered by ice shelves. The scientists generally remain at the shelves' fringes, measuring the temperature and salinity of the water flowing in and out of the sheltered depths, says Jacobs. As useful as that information may be, it doesn't give scientists direct data on the water under Antarctica's ice shelves. AUVs can gather such data and at the same time map the seafloor and measure ice thickness, he notes.

Researchers have drilled holes through ice shelves and dropped instruments into the frigid water beneath, but even that's been done at fewer than 20 sites in Antarctica in the past half-century, says Keith W. Nicholls, an oceanographer with the British Antarctic Survey in Cambridge, England. Such efforts have provided data about single spots over extended periods of time, but scientists have also longed for water-temperature and salinity data gathered over wide areas during short periods of time, he notes.

So, in February 2005, Nicholls and his colleagues sent an AUV called Autosub on an extended jaunt under the Fimbul ice shelf, which fows into the South Atlantic. Autosub was designed for use in open seas, so for the under-ice missions, engineers added sonar equipment so that the AUV could navigate by features on the seafloor, says Nicholls. Deployed from a ship near the edge of the 50,000-[km.sup.2] ice shelf, Autosub traveled about 25 km into the ice-covered cavity before turning around and coming out along the same path. On the inbound leg of the mission, the AUV cruised at about 150 m above the ocean floor. Then, Autosub rose several hundred meters to a point near the ice shelf's lower surface, turned around, and headed back to the ship. For most of its outbound trip, the AUV flew about 100 m below the ice.

Sonar readings indicated that that shelf's undersurface is, for the most part, smooth. Any bumps, cracks, or other features in those regions are no more than a few millimeters across, says Nicholls. The smooth complexion isn't surprising, he notes, because an ice surfacein contact with water is naturally self-leveling. The deeper a chunk of ice sticks down into an ocean, the higher the water pressure on it and the more quickly it melts. So, features that hang like icicles underneath the shelves tend to disappear. Conversely, water within an inverted crevasse in the ice freezes readily and fills the gap with smooth ice because it's shallower and therefore experiences less pressure than water below it.

In some areas, however, the AUV spotted kilometer-wide sections of ice that were unexpectedly riven with fissures as deep as 30 m. "We're still unsure about what this means," Nicholls notes. "Nobody ... can come up with an idea of how such rough terrain is maintained."

The jumbled topography is probably a sign that melting rates in that part of the ice shelf are high, says Nicholls. However, because the rough-ice regions have well-defined edges, the chaotic patches of ice don't seem to be a result of ocean currents or widespread turbulence beneath the ice shelf.

Data from above the ice hint at what forces have roughed up the lower surface. By comparing the navigational data gathered by the AUV with satellite images of the same region, the scientists noted that the rough patches show up directly beneath shallow depressions in the upper surface of the ice shelf. These features, called flow traces, seem to have been created as the ice slowly spilling from the continent thinned after it passed over, around, or between geological features such as small islands, says Nicholls.

Flow traces are present on top of all ice shelves, so the presumption that all ice shelves are smooth underneath may need to be reassessed, Nicholls and his team assert in the April 28 Geophysical Research Letters. That's hardly a trivial matter, since it would mean that computer models that simulate the flow of ocean currents beneath ice shelves may need to be adjusted to incorporate additional fluid friction caused by large areas of rough ice.

COLD HEAT When the water beneath ice shelves is barely above the freezing point--for seawater, that's about -1.9[degrees]C--only a few centimeters of ice melts from the shelves each year. Scientists estimate that for each 0.1[degrees]C rise in the underlying water's temperature, an extra 1 m or so of ice could melt over the course of a year (SN: 11/1/03, p. 278).

Data that Autosub gathered on its 2005 voyage under the Fimbul ice shelf indicate that the water there was somewhat warmer and saltier than water that ship-based oceanographers sampled just off the ice shelf's edge, says Nicholls. Therefore, he suspects that the water beneath the ice shelf must have migrated there during the previous winter, when storms may have blown unusually warm and salty waters to the region from depths farther offshore.

The water could also have come from a current meandering along an unusual path through the region or a passing ocean eddy, says David M. Holland, an oceanographer at New York University. The water in the center of those whirlpool-like ocean features, sometimes transported from distant regions, can be 1[degrees]C or more warmer than the surrounding ocean (SN: 6/14/03, p. 375).

Holland and his colleagues used sound-echo data gathered atop the Fimbul ice shelf to construct a computer model of the region that includes the shelf, the waters beneath it, and the underlying seafloor. Then, they used that model to predict 11 years' worth of ocean-circulation data under present conditions.

After the first 4 years, according to the model, the average temperature of the water under the ice stabilized at about -1[degrees]C, almost a full degree warmer than seawater's freezing point. In areas near where the floating ice shelf touches the seafloor near the Antarctic shore, a layer of ice more than 10 m thick melted each year, the researchers reported in the Jan. 15 Journal of Geophysical Research (Oceans). In the team's model, on average, the entire ice shelf thinned by almost 2 m per year, they note.

The water temperature actually measured by Nicholls and his team using Autosub was a degree or so warmer than the average temperature estimated by the Holland team's model. The model could be wrong, says Holland. However, the apparent difference between the two figures may not be significant, he notes.

First of all, the average temperature calculated in the team's simulation is an average for the entire year; the AUV may have taken its measurements during a warm spell. Also, the 50-km route taken by Autosub may have fallen entirely within a small, relatively warm patch of water beneath the 50,000-[km.sup.2] ice shelf. Finally, weather conditions in the region during the Nicholls team's expedition may have enabled exceptionally large quantities of warm, deep water from offshore to spill into the cavity beneath the ice shelf, says Holland.

Furthermore, the computer model doesn't now include longterm variations such as meandering ocean currents, occasional eddies, or significant changes in weather patterns. And it doesn't offer ways to incorporate sudden changes in ice shelf geometry, says Holland. "The model isn't as advanced as we'd like, but these aren't insurmountable problems" he notes.

LOSS OF A PIONEER Alas, Autosub's first round trip under Antarctica's Fimbul ice shelf was its last. On the second of several missions planned for the 2005 expedition, the AUV didn't come back. Scientists are still trying to figure out what went wrong, and biologists who'd hoped to use the AUV's camera to get a closer look at the seabed on subsequent missions lament that the craft's second excursion met an untimely end. "It's really disappointing, since the first mission had done so well," says Brian Bett, a marine biologist at the National Oceanography Centre in Southampton, England.

Despite the disappearance, Nicholls and his team proved that AUVs can operate under an ice shelf, says Jacobs. Long-range versions of such craft could cover more territory and collect much more data. Scientists could also send them to those areas where the ice shelf meets the Antarctic shore. This small-but-dynamic area is called an ice shelf's grounding line, and researchers are intensely interested in conditions there. For one thing, water samples from this area could indicate how much of the fresh water reaching the oceans in these regions melts from the ice shelf and how much originates as subglacial melt on shore.

Scientists recently put a new Autosub, built at a cost of more than $1.5 million, through sea trials in the North Atlantic. Its next mission under an ice shelf is scheduled for early 2007, when a team led by Jacobs will visit Antarctica's Pine Island Glacier.

Data from that expedition will enable scientists to compare the conditions there with those under the Fimbul ice shelf. Also, if all goes well, the marine biologists will finally see what kinds of organisms call this environment home. "There are plenty of exciting things to be found down there,' says Bert. "It'll be a voyage of discovery."

And what might the researchers find? In March 2005, scientists who explored the ocean uncovered when the Larsen B ice shelf collapsed and drifted away (SN: 3/30/02, p. 197) found thick mats of bacteria that were probably nourished by nutrient-rich water seeping from the seafloor (SN: 8/6/05, p. 94). Researchers navigating the new AUV under the Pine Island shelf might find similar life forms or others, says Bert. Organisms there might eke out their livings on the organic matter brought in by ocean currents or by scavenging the occasional carcass washed from the open ocean, he says.

"If we knew ahead of time what we were going to see, we wouldn't bother," Bett notes.