The next big oil spill could be out of sight. Climate warming has packs of Arctic sea ice in retreat, opening up vast areas for oil and gas drilling. That is posing a new problem for spill detectors: There is still a lot of ice in the region, and people cannot see through it. Remember that giant oil slick on the surface of the Gulf of Mexico after the Deepwater Horizon oil rig blowout? Off the north coast of Alaska that kind of slick would likely be shielded by miles of drifting ice. “The risk of a serious oil spill in the Arctic is escalating,” the National Research Council warned in a report just last month. And, the council added, the U.S. is not ready to respond.

One answer could be to use sound rather than sight. High-frequency sonar chirps can reveal oil underneath ice, even when it is sandwiched between ice layers. “We were able to distinguish two different signatures: oil together with ice versus just ice alone,” says Christopher Bassett, a postdoctoral researcher at Woods Hole Oceanographic Institution. He and his colleagues presented their work Wednesday at the Acoustical Society of America’s meeting in Providence. Other researchers showed that sonar was sensitive enough to detect even tiny leaks, down to the level of individual oil and gas bubbles.

Flying spotting planes or sending scout ships across the chilly Chukchi and Beaufort seas may not find oil spills after they have drifted miles from their origins, Bassett says. The oil could be hidden under moving packs of seasonal ice that form and melt every year in the region—hidden from visual observation, that is.

But not from sound waves. In a cold seawater tank in Hamburg, Germany, Bassett and his group grew a 12-centimeter-thick layer of ice. Then they squirted 50 liters of North Sea crude oil under it and continued to freeze the water until they had created an ice and oil sandwich.

At the bottom of the tank, about a meter away, they placed several sonar emitters, aimed them at the ice and analyzed the echoes. Unlike the single-frequency sound waves used in commercial depth sounders and fish finders, these instruments gave off a quickly ascending burst of frequencies, spanning a range  from 200 kilohertz up to one megahertz. “It’s beyond the range of human hearing, but if you could hear, it would sound like a rising ‘wooooop!’” Bassett says. (He adds that most of these sounds are beyond the range of marine mammal hearing and are very brief. Unlike the long, low drone of ship motors it is unlikely to bother whales and the like, he says.) The broad band of frequencies allowed the scientists to trace different kinds of echoes from the same spot. That gave them more information about the material that made up each target. Bare ice, it turned out, looked different on a sonar plot than did oil under ice or oil trapped within ice.

Sonar detection can also get very fine-grained. At the acoustics meeting, Geir Pedersen of Christian Michelsen Research in Norway, a not-for-profit institute affiliated with the University of Bergen, showed they could use multifrequency sonar to monitor streams of bubbles from leaks. He and his colleagues placed their equipment five meters away from pipes leaking vegetable oil in a secluded Norway fjord. “At that range we could detect down to single bubbles and oil droplets,” Pedersen says. The technique could be particularly useful for finding the source of a subsurface leak.

Both Bassett and Pedersen stress that they were working in controlled conditions, very different from the storm-tossed, equipment-wrecking Arctic Ocean. Durable instruments, and better ways to  filter out confusing real-world noises, are going to be needed before this technology can be deployed. But to the scientists, it does sound promising.