When Robert Ballard and a team of scientists sighted the clay cargo containers once used by Roman merchants in the depths of the Mediterranean this summer, they knew they were onto something big. These weighty carriers, called amphora, were often the first item overboard when a vessel began sinking - so spotting them meant an ancient ship was nearby. The tricky part was seeing enough of the container to identify it.
Fortunately, Ballard and his cohorts benefited from recent strides in underwater navigation, improvements in sonars and transponders that, together, allow a rover such as Jason to cultivate a sense of hearing that rivals the eyes. These newer devices use sound waves to locate and identify objects in murky depths of up to 6,000 meters, territory that renders traditional land positioning technologies such as radio waves impotent. They're also identifying tiny objects with such precision that an observer can use the data to determine the head or tail of a coin and which way it happens to be facing.
"The problem with navigating an underwater vehicle is that you don't know where the hell you are," said Louis Whitcomb, a mechanical engineering professor at Johns Hopkins University who took part in the Roman expedition off the coast of ancient Carthage. "We needed something that penetrates water."
Sonar systems such as the one used to aid Ballard and archaeologist Anne McCann culls technological lessons from decades of trial and use in Naval submarines. Sonar mimics the way dolphins and bats navigate sea and air by measuring the time it takes for high-frequency signals to travel to a target and back.
Whitcomb's team rolled together a combination of technologies, most notably a long baseline acoustic navigation system - one that tracks a vehicle's or diver's position relative to a series of fixed stations - and Doppler sonar, a device that can read the change in the frequency of sound waves caused by the movement of the target or of the sonar. This latter technology gave the system a way of updating position coordinates as the Jason rover moved, because it sent its signals to a network of non-fixed transponders - radio transmitters that send guidance signals - which were tethered from the ships that carried Whitcomb and the other scientists to the Mediterranean.
"Doppler sonar gave us standard time of flight navigation, the speed at every second," Whitcomb explained.
By contrast, some traditional navigation systems use only long baseline navigation in conjunction with a carpet of fixed transponders, and are limited by the speed of sound in water, about 1,500 meters per second. Incorporating the dynamic sound-wave reading capability of the Doppler along with the roving transponders allowed the scientists to overcome this limitation and let Jason explore the wreckage site more freely - as a land-bound archaeologist would - to get a close-up and more accurate picture of artifacts.
Up close and personal is just what Roman Kuc is shooting for with his sonar system. The Yale University researcher is testing an acoustical system that cuts through huge waves of sound information to the precise data to identify objects. This precision is derived from three animal-like functions that allow the sonar to move in the direction of a sound, follow the source, and pick out the part of the sound that it deems most important. Together, these operations allow the system to draw a picture of an object from sound that is more detailed than one resulting from the use of cameras, Kuc said.
"The problem with cameras is that they produce lots of data," said Kuc, the director of Yale's Intelligent Sensors Laboratory. "An image is about 2 megabits, and we're producing a one-dimensional echo that's 3 kilobits in size."
The advantage of smaller "image" files is that Kuc can teach the sonar system to identify a wide range of objects using a form of pattern recognition. Kuc teaches the sonar system the sound waves bounced off objects, such as different sizes of balls, washers, and O-rings. These wave patterns, which are 3 KB in size, are stored in a database that can easily fit onto a 1.44-MB floppy. The result is a system that's as capable as a dolphin at rooting out an object.
"All sonars generate an image, but a dolphin doesn't. It looks only at the waveform," explained Kuc. "The sensor has to go through a learning stage to train it so it can compare observed echoes with its database."
"The representation of the wave is enough to differentiate the object," Kuc continued. So the sonar system can tell, for example, if Franklin Roosevelt's head on the dime is facing up or down, he noted.
Put this system together with ever-faster processors on board computers, and a system can be pretty adept at identifying objects. It also gives Kuc a healthy respect for the sense of hearing. "We're so dependent on vision, we forget how sharp our other senses are," he said.
Not that cameras are soon to go into mothballs on expeditions such as Ballard's. In fact, they will still come in handy for close-ups, allowing sonars to provide the bigger picture, Kuc said.
